1 files changed, 200 insertions, 72 deletions

diff --git a/drivers/gpu/drm/virtio/virtgpu_vq.c b/drivers/gpu/drm/virtio/virtgpu_vq.c
index 981ee16e3ee9..5914e79d3429 100644
--- a/drivers/gpu/drm/virtio/virtgpu_vq.c
+++ b/drivers/gpu/drm/virtio/virtgpu_vq.c
@@ -26,19 +26,31 @@
  * OTHER DEALINGS IN THE SOFTWARE.
  */
 
-#include <drm/drmP.h>
-#include "virtgpu_drv.h"
-#include "virtgpu_trace.h"
+#include <linux/dma-mapping.h>
 #include <linux/virtio.h>
 #include <linux/virtio_config.h>
 #include <linux/virtio_ring.h>
 
+#include "virtgpu_drv.h"
+#include "virtgpu_trace.h"
+
 #define MAX_INLINE_CMD_SIZE   96
 #define MAX_INLINE_RESP_SIZE  24
 #define VBUFFER_SIZE          (sizeof(struct virtio_gpu_vbuffer) \
 			       + MAX_INLINE_CMD_SIZE		 \
 			       + MAX_INLINE_RESP_SIZE)
 
+static void convert_to_hw_box(struct virtio_gpu_box *dst,
+			      const struct drm_virtgpu_3d_box *src)
+{
+	dst->x = cpu_to_le32(src->x);
+	dst->y = cpu_to_le32(src->y);
+	dst->z = cpu_to_le32(src->z);
+	dst->w = cpu_to_le32(src->w);
+	dst->h = cpu_to_le32(src->h);
+	dst->d = cpu_to_le32(src->d);
+}
+
 void virtio_gpu_ctrl_ack(struct virtqueue *vq)
 {
 	struct drm_device *dev = vq->vdev->priv;
@@ -154,7 +166,7 @@ static void free_vbuf(struct virtio_gpu_device *vgdev,
 {
 	if (vbuf->resp_size > MAX_INLINE_RESP_SIZE)
 		kfree(vbuf->resp_buf);
-	kfree(vbuf->data_buf);
+	kvfree(vbuf->data_buf);
 	kmem_cache_free(vgdev->vbufs, vbuf);
 }
 
@@ -191,7 +203,7 @@ void virtio_gpu_dequeue_ctrl_func(struct work_struct *work)
 	} while (!virtqueue_enable_cb(vgdev->ctrlq.vq));
 	spin_unlock(&vgdev->ctrlq.qlock);
 
-	list_for_each_entry_safe(entry, tmp, &reclaim_list, list) {
+	list_for_each_entry(entry, &reclaim_list, list) {
 		resp = (struct virtio_gpu_ctrl_hdr *)entry->resp_buf;
 
 		trace_virtio_gpu_cmd_response(vgdev->ctrlq.vq, resp);
@@ -218,14 +230,18 @@ void virtio_gpu_dequeue_ctrl_func(struct work_struct *work)
 		}
 		if (entry->resp_cb)
 			entry->resp_cb(vgdev, entry);
-
-		list_del(&entry->list);
-		free_vbuf(vgdev, entry);
 	}
 	wake_up(&vgdev->ctrlq.ack_queue);
 
 	if (fence_id)
 		virtio_gpu_fence_event_process(vgdev, fence_id);
+
+	list_for_each_entry_safe(entry, tmp, &reclaim_list, list) {
+		if (entry->objs)
+			virtio_gpu_array_put_free_delayed(vgdev, entry->objs);
+		list_del(&entry->list);
+		free_vbuf(vgdev, entry);
+	}
 }
 
 void virtio_gpu_dequeue_cursor_func(struct work_struct *work)
@@ -251,26 +267,67 @@ void virtio_gpu_dequeue_cursor_func(struct work_struct *work)
 	wake_up(&vgdev->cursorq.ack_queue);
 }
 
-static int virtio_gpu_queue_ctrl_buffer_locked(struct virtio_gpu_device *vgdev,
-					       struct virtio_gpu_vbuffer *vbuf)
+/* Create sg_table from a vmalloc'd buffer. */
+static struct sg_table *vmalloc_to_sgt(char *data, uint32_t size, int *sg_ents)
+{
+	int ret, s, i;
+	struct sg_table *sgt;
+	struct scatterlist *sg;
+	struct page *pg;
+
+	if (WARN_ON(!PAGE_ALIGNED(data)))
+		return NULL;
+
+	sgt = kmalloc(sizeof(*sgt), GFP_KERNEL);
+	if (!sgt)
+		return NULL;
+
+	*sg_ents = DIV_ROUND_UP(size, PAGE_SIZE);
+	ret = sg_alloc_table(sgt, *sg_ents, GFP_KERNEL);
+	if (ret) {
+		kfree(sgt);
+		return NULL;
+	}
+
+	for_each_sg(sgt->sgl, sg, *sg_ents, i) {
+		pg = vmalloc_to_page(data);
+		if (!pg) {
+			sg_free_table(sgt);
+			kfree(sgt);
+			return NULL;
+		}
+
+		s = min_t(int, PAGE_SIZE, size);
+		sg_set_page(sg, pg, s, 0);
+
+		size -= s;
+		data += s;
+	}
+
+	return sgt;
+}
+
+static bool virtio_gpu_queue_ctrl_buffer_locked(struct virtio_gpu_device *vgdev,
+						struct virtio_gpu_vbuffer *vbuf,
+						struct scatterlist *vout)
 		__releases(&vgdev->ctrlq.qlock)
 		__acquires(&vgdev->ctrlq.qlock)
 {
 	struct virtqueue *vq = vgdev->ctrlq.vq;
-	struct scatterlist *sgs[3], vcmd, vout, vresp;
+	struct scatterlist *sgs[3], vcmd, vresp;
 	int outcnt = 0, incnt = 0;
+	bool notify = false;
 	int ret;
 
 	if (!vgdev->vqs_ready)
-		return -ENODEV;
+		return notify;
 
 	sg_init_one(&vcmd, vbuf->buf, vbuf->size);
 	sgs[outcnt + incnt] = &vcmd;
 	outcnt++;
 
-	if (vbuf->data_size) {
-		sg_init_one(&vout, vbuf->data_buf, vbuf->data_size);
-		sgs[outcnt + incnt] = &vout;
+	if (vout) {
+		sgs[outcnt + incnt] = vout;
 		outcnt++;
 	}
 
@@ -291,32 +348,35 @@ retry:
 		trace_virtio_gpu_cmd_queue(vq,
 			(struct virtio_gpu_ctrl_hdr *)vbuf->buf);
 
-		virtqueue_kick(vq);
+		notify = virtqueue_kick_prepare(vq);
 	}
-
-	if (!ret)
-		ret = vq->num_free;
-	return ret;
-}
-
-static int virtio_gpu_queue_ctrl_buffer(struct virtio_gpu_device *vgdev,
-					struct virtio_gpu_vbuffer *vbuf)
-{
-	int rc;
-
-	spin_lock(&vgdev->ctrlq.qlock);
-	rc = virtio_gpu_queue_ctrl_buffer_locked(vgdev, vbuf);
-	spin_unlock(&vgdev->ctrlq.qlock);
-	return rc;
+	return notify;
 }
 
-static int virtio_gpu_queue_fenced_ctrl_buffer(struct virtio_gpu_device *vgdev,
-					       struct virtio_gpu_vbuffer *vbuf,
-					       struct virtio_gpu_ctrl_hdr *hdr,
-					       struct virtio_gpu_fence *fence)
+static void virtio_gpu_queue_fenced_ctrl_buffer(struct virtio_gpu_device *vgdev,
+						struct virtio_gpu_vbuffer *vbuf,
+						struct virtio_gpu_ctrl_hdr *hdr,
+						struct virtio_gpu_fence *fence)
 {
 	struct virtqueue *vq = vgdev->ctrlq.vq;
-	int rc;
+	struct scatterlist *vout = NULL, sg;
+	struct sg_table *sgt = NULL;
+	bool notify;
+	int outcnt = 0;
+
+	if (vbuf->data_size) {
+		if (is_vmalloc_addr(vbuf->data_buf)) {
+			sgt = vmalloc_to_sgt(vbuf->data_buf, vbuf->data_size,
+					     &outcnt);
+			if (!sgt)
+				return;
+			vout = sgt->sgl;
+		} else {
+			sg_init_one(&sg, vbuf->data_buf, vbuf->data_size);
+			vout = &sg;
+			outcnt = 1;
+		}
+	}
 
 again:
 	spin_lock(&vgdev->ctrlq.qlock);
@@ -329,29 +389,66 @@ again:
 	 * to wait for free space, which can result in fence ids being
 	 * submitted out-of-order.
 	 */
-	if (vq->num_free < 3) {
+	if (vq->num_free < 2 + outcnt) {
 		spin_unlock(&vgdev->ctrlq.qlock);
 		wait_event(vgdev->ctrlq.ack_queue, vq->num_free >= 3);
 		goto again;
 	}
 
-	if (fence)
+	if (hdr && fence) {
 		virtio_gpu_fence_emit(vgdev, hdr, fence);
-	rc = virtio_gpu_queue_ctrl_buffer_locked(vgdev, vbuf);
+		if (vbuf->objs) {
+			virtio_gpu_array_add_fence(vbuf->objs, &fence->f);
+			virtio_gpu_array_unlock_resv(vbuf->objs);
+		}
+	}
+	notify = virtio_gpu_queue_ctrl_buffer_locked(vgdev, vbuf, vout);
 	spin_unlock(&vgdev->ctrlq.qlock);
-	return rc;
+	if (notify) {
+		if (vgdev->disable_notify)
+			vgdev->pending_notify = true;
+		else
+			virtqueue_notify(vgdev->ctrlq.vq);
+	}
+
+	if (sgt) {
+		sg_free_table(sgt);
+		kfree(sgt);
+	}
+}
+
+void virtio_gpu_disable_notify(struct virtio_gpu_device *vgdev)
+{
+	vgdev->disable_notify = true;
+}
+
+void virtio_gpu_enable_notify(struct virtio_gpu_device *vgdev)
+{
+	vgdev->disable_notify = false;
+
+	if (!vgdev->pending_notify)
+		return;
+	vgdev->pending_notify = false;
+	virtqueue_notify(vgdev->ctrlq.vq);
+}
+
+static void virtio_gpu_queue_ctrl_buffer(struct virtio_gpu_device *vgdev,
+					 struct virtio_gpu_vbuffer *vbuf)
+{
+	virtio_gpu_queue_fenced_ctrl_buffer(vgdev, vbuf, NULL, NULL);
 }
 
-static int virtio_gpu_queue_cursor(struct virtio_gpu_device *vgdev,
-				   struct virtio_gpu_vbuffer *vbuf)
+static void virtio_gpu_queue_cursor(struct virtio_gpu_device *vgdev,
+				    struct virtio_gpu_vbuffer *vbuf)
 {
 	struct virtqueue *vq = vgdev->cursorq.vq;
 	struct scatterlist *sgs[1], ccmd;
+	bool notify;
 	int ret;
 	int outcnt;
 
 	if (!vgdev->vqs_ready)
-		return -ENODEV;
+		return;
 
 	sg_init_one(&ccmd, vbuf->buf, vbuf->size);
 	sgs[0] = &ccmd;
@@ -369,14 +466,13 @@ retry:
 		trace_virtio_gpu_cmd_queue(vq,
 			(struct virtio_gpu_ctrl_hdr *)vbuf->buf);
 
-		virtqueue_kick(vq);
+		notify = virtqueue_kick_prepare(vq);
 	}
 
 	spin_unlock(&vgdev->cursorq.qlock);
 
-	if (!ret)
-		ret = vq->num_free;
-	return ret;
+	if (notify)
+		virtqueue_notify(vq);
 }
 
 /* just create gem objects for userspace and long lived objects,
@@ -387,6 +483,7 @@ retry:
 void virtio_gpu_cmd_create_resource(struct virtio_gpu_device *vgdev,
 				    struct virtio_gpu_object *bo,
 				    struct virtio_gpu_object_params *params,
+				    struct virtio_gpu_object_array *objs,
 				    struct virtio_gpu_fence *fence)
 {
 	struct virtio_gpu_resource_create_2d *cmd_p;
@@ -394,6 +491,7 @@ void virtio_gpu_cmd_create_resource(struct virtio_gpu_device *vgdev,
 
 	cmd_p = virtio_gpu_alloc_cmd(vgdev, &vbuf, sizeof(*cmd_p));
 	memset(cmd_p, 0, sizeof(*cmd_p));
+	vbuf->objs = objs;
 
 	cmd_p->hdr.type = cpu_to_le32(VIRTIO_GPU_CMD_RESOURCE_CREATE_2D);
 	cmd_p->resource_id = cpu_to_le32(bo->hw_res_handle);
@@ -480,12 +578,13 @@ void virtio_gpu_cmd_resource_flush(struct virtio_gpu_device *vgdev,
 }
 
 void virtio_gpu_cmd_transfer_to_host_2d(struct virtio_gpu_device *vgdev,
-					struct virtio_gpu_object *bo,
 					uint64_t offset,
-					__le32 width, __le32 height,
-					__le32 x, __le32 y,
+					uint32_t width, uint32_t height,
+					uint32_t x, uint32_t y,
+					struct virtio_gpu_object_array *objs,
 					struct virtio_gpu_fence *fence)
 {
+	struct virtio_gpu_object *bo = gem_to_virtio_gpu_obj(objs->objs[0]);
 	struct virtio_gpu_transfer_to_host_2d *cmd_p;
 	struct virtio_gpu_vbuffer *vbuf;
 	bool use_dma_api = !virtio_has_iommu_quirk(vgdev->vdev);
@@ -497,14 +596,15 @@ void virtio_gpu_cmd_transfer_to_host_2d(struct virtio_gpu_device *vgdev,
 
 	cmd_p = virtio_gpu_alloc_cmd(vgdev, &vbuf, sizeof(*cmd_p));
 	memset(cmd_p, 0, sizeof(*cmd_p));
+	vbuf->objs = objs;
 
 	cmd_p->hdr.type = cpu_to_le32(VIRTIO_GPU_CMD_TRANSFER_TO_HOST_2D);
 	cmd_p->resource_id = cpu_to_le32(bo->hw_res_handle);
 	cmd_p->offset = cpu_to_le64(offset);
-	cmd_p->r.width = width;
-	cmd_p->r.height = height;
-	cmd_p->r.x = x;
-	cmd_p->r.y = y;
+	cmd_p->r.width = cpu_to_le32(width);
+	cmd_p->r.height = cpu_to_le32(height);
+	cmd_p->r.x = cpu_to_le32(x);
+	cmd_p->r.y = cpu_to_le32(y);
 
 	virtio_gpu_queue_fenced_ctrl_buffer(vgdev, vbuf, &cmd_p->hdr, fence);
 }
@@ -825,34 +925,38 @@ void virtio_gpu_cmd_context_destroy(struct virtio_gpu_device *vgdev,
 
 void virtio_gpu_cmd_context_attach_resource(struct virtio_gpu_device *vgdev,
 					    uint32_t ctx_id,
-					    uint32_t resource_id)
+					    struct virtio_gpu_object_array *objs)
 {
+	struct virtio_gpu_object *bo = gem_to_virtio_gpu_obj(objs->objs[0]);
 	struct virtio_gpu_ctx_resource *cmd_p;
 	struct virtio_gpu_vbuffer *vbuf;
 
 	cmd_p = virtio_gpu_alloc_cmd(vgdev, &vbuf, sizeof(*cmd_p));
 	memset(cmd_p, 0, sizeof(*cmd_p));
+	vbuf->objs = objs;
 
 	cmd_p->hdr.type = cpu_to_le32(VIRTIO_GPU_CMD_CTX_ATTACH_RESOURCE);
 	cmd_p->hdr.ctx_id = cpu_to_le32(ctx_id);
-	cmd_p->resource_id = cpu_to_le32(resource_id);
+	cmd_p->resource_id = cpu_to_le32(bo->hw_res_handle);
 	virtio_gpu_queue_ctrl_buffer(vgdev, vbuf);
 
 }
 
 void virtio_gpu_cmd_context_detach_resource(struct virtio_gpu_device *vgdev,
 					    uint32_t ctx_id,
-					    uint32_t resource_id)
+					    struct virtio_gpu_object_array *objs)
 {
+	struct virtio_gpu_object *bo = gem_to_virtio_gpu_obj(objs->objs[0]);
 	struct virtio_gpu_ctx_resource *cmd_p;
 	struct virtio_gpu_vbuffer *vbuf;
 
 	cmd_p = virtio_gpu_alloc_cmd(vgdev, &vbuf, sizeof(*cmd_p));
 	memset(cmd_p, 0, sizeof(*cmd_p));
+	vbuf->objs = objs;
 
 	cmd_p->hdr.type = cpu_to_le32(VIRTIO_GPU_CMD_CTX_DETACH_RESOURCE);
 	cmd_p->hdr.ctx_id = cpu_to_le32(ctx_id);
-	cmd_p->resource_id = cpu_to_le32(resource_id);
+	cmd_p->resource_id = cpu_to_le32(bo->hw_res_handle);
 	virtio_gpu_queue_ctrl_buffer(vgdev, vbuf);
 }
 
@@ -860,6 +964,7 @@ void
 virtio_gpu_cmd_resource_create_3d(struct virtio_gpu_device *vgdev,
 				  struct virtio_gpu_object *bo,
 				  struct virtio_gpu_object_params *params,
+				  struct virtio_gpu_object_array *objs,
 				  struct virtio_gpu_fence *fence)
 {
 	struct virtio_gpu_resource_create_3d *cmd_p;
@@ -867,6 +972,7 @@ virtio_gpu_cmd_resource_create_3d(struct virtio_gpu_device *vgdev,
 
 	cmd_p = virtio_gpu_alloc_cmd(vgdev, &vbuf, sizeof(*cmd_p));
 	memset(cmd_p, 0, sizeof(*cmd_p));
+	vbuf->objs = objs;
 
 	cmd_p->hdr.type = cpu_to_le32(VIRTIO_GPU_CMD_RESOURCE_CREATE_3D);
 	cmd_p->resource_id = cpu_to_le32(bo->hw_res_handle);
@@ -887,12 +993,13 @@ virtio_gpu_cmd_resource_create_3d(struct virtio_gpu_device *vgdev,
 }
 
 void virtio_gpu_cmd_transfer_to_host_3d(struct virtio_gpu_device *vgdev,
-					struct virtio_gpu_object *bo,
 					uint32_t ctx_id,
 					uint64_t offset, uint32_t level,
-					struct virtio_gpu_box *box,
+					struct drm_virtgpu_3d_box *box,
+					struct virtio_gpu_object_array *objs,
 					struct virtio_gpu_fence *fence)
 {
+	struct virtio_gpu_object *bo = gem_to_virtio_gpu_obj(objs->objs[0]);
 	struct virtio_gpu_transfer_host_3d *cmd_p;
 	struct virtio_gpu_vbuffer *vbuf;
 	bool use_dma_api = !virtio_has_iommu_quirk(vgdev->vdev);
@@ -905,10 +1012,12 @@ void virtio_gpu_cmd_transfer_to_host_3d(struct virtio_gpu_device *vgdev,
 	cmd_p = virtio_gpu_alloc_cmd(vgdev, &vbuf, sizeof(*cmd_p));
 	memset(cmd_p, 0, sizeof(*cmd_p));
 
+	vbuf->objs = objs;
+
 	cmd_p->hdr.type = cpu_to_le32(VIRTIO_GPU_CMD_TRANSFER_TO_HOST_3D);
 	cmd_p->hdr.ctx_id = cpu_to_le32(ctx_id);
 	cmd_p->resource_id = cpu_to_le32(bo->hw_res_handle);
-	cmd_p->box = *box;
+	convert_to_hw_box(&cmd_p->box, box);
 	cmd_p->offset = cpu_to_le64(offset);
 	cmd_p->level = cpu_to_le32(level);
 
@@ -916,21 +1025,25 @@ void virtio_gpu_cmd_transfer_to_host_3d(struct virtio_gpu_device *vgdev,
 }
 
 void virtio_gpu_cmd_transfer_from_host_3d(struct virtio_gpu_device *vgdev,
-					  uint32_t resource_id, uint32_t ctx_id,
+					  uint32_t ctx_id,
 					  uint64_t offset, uint32_t level,
-					  struct virtio_gpu_box *box,
+					  struct drm_virtgpu_3d_box *box,
+					  struct virtio_gpu_object_array *objs,
 					  struct virtio_gpu_fence *fence)
 {
+	struct virtio_gpu_object *bo = gem_to_virtio_gpu_obj(objs->objs[0]);
 	struct virtio_gpu_transfer_host_3d *cmd_p;
 	struct virtio_gpu_vbuffer *vbuf;
 
 	cmd_p = virtio_gpu_alloc_cmd(vgdev, &vbuf, sizeof(*cmd_p));
 	memset(cmd_p, 0, sizeof(*cmd_p));
 
+	vbuf->objs = objs;
+
 	cmd_p->hdr.type = cpu_to_le32(VIRTIO_GPU_CMD_TRANSFER_FROM_HOST_3D);
 	cmd_p->hdr.ctx_id = cpu_to_le32(ctx_id);
-	cmd_p->resource_id = cpu_to_le32(resource_id);
-	cmd_p->box = *box;
+	cmd_p->resource_id = cpu_to_le32(bo->hw_res_handle);
+	convert_to_hw_box(&cmd_p->box, box);
 	cmd_p->offset = cpu_to_le64(offset);
 	cmd_p->level = cpu_to_le32(level);
 
@@ -939,7 +1052,9 @@ void virtio_gpu_cmd_transfer_from_host_3d(struct virtio_gpu_device *vgdev,
 
 void virtio_gpu_cmd_submit(struct virtio_gpu_device *vgdev,
 			   void *data, uint32_t data_size,
-			   uint32_t ctx_id, struct virtio_gpu_fence *fence)
+			   uint32_t ctx_id,
+			   struct virtio_gpu_object_array *objs,
+			   struct virtio_gpu_fence *fence)
 {
 	struct virtio_gpu_cmd_submit *cmd_p;
 	struct virtio_gpu_vbuffer *vbuf;
@@ -949,6 +1064,7 @@ void virtio_gpu_cmd_submit(struct virtio_gpu_device *vgdev,
 
 	vbuf->data_buf = data;
 	vbuf->data_size = data_size;
+	vbuf->objs = objs;
 
 	cmd_p->hdr.type = cpu_to_le32(VIRTIO_GPU_CMD_SUBMIT_3D);
 	cmd_p->hdr.ctx_id = cpu_to_le32(ctx_id);
@@ -964,17 +1080,21 @@ int virtio_gpu_object_attach(struct virtio_gpu_device *vgdev,
 	bool use_dma_api = !virtio_has_iommu_quirk(vgdev->vdev);
 	struct virtio_gpu_mem_entry *ents;
 	struct scatterlist *sg;
-	int si, nents;
+	int si, nents, ret;
 
 	if (WARN_ON_ONCE(!obj->created))
 		return -EINVAL;
+	if (WARN_ON_ONCE(obj->pages))
+		return -EINVAL;
 
-	if (!obj->pages) {
-		int ret;
+	ret = drm_gem_shmem_pin(&obj->base.base);
+	if (ret < 0)
+		return -EINVAL;
 
-		ret = virtio_gpu_object_get_sg_table(vgdev, obj);
-		if (ret)
-			return ret;
+	obj->pages = drm_gem_shmem_get_sg_table(&obj->base.base);
+	if (obj->pages == NULL) {
+		drm_gem_shmem_unpin(&obj->base.base);
+		return -EINVAL;
 	}
 
 	if (use_dma_api) {
@@ -1013,6 +1133,9 @@ void virtio_gpu_object_detach(struct virtio_gpu_device *vgdev,
 {
 	bool use_dma_api = !virtio_has_iommu_quirk(vgdev->vdev);
 
+	if (WARN_ON_ONCE(!obj->pages))
+		return;
+
 	if (use_dma_api && obj->mapped) {
 		struct virtio_gpu_fence *fence = virtio_gpu_fence_alloc(vgdev);
 		/* detach backing and wait for the host process it ... */
@@ -1028,6 +1151,11 @@ void virtio_gpu_object_detach(struct virtio_gpu_device *vgdev,
 	} else {
 		virtio_gpu_cmd_resource_inval_backing(vgdev, obj->hw_res_handle, NULL);
 	}
+
+	sg_free_table(obj->pages);
+	obj->pages = NULL;
+
+	drm_gem_shmem_unpin(&obj->base.base);
 }
 
 void virtio_gpu_cursor_ping(struct virtio_gpu_device *vgdev,