[PATCH v2 2/3] drm/vc4: Export fence through syncobj

[Stefan Schake]

Allow specifying a syncobj on render job submission where we store the
fence for the job. This gives userland flexible access to the fence.

v2: Use 0 as invalid syncobj to drop flag (Eric)
    Don't reintroduce the padding (Eric)

Signed-off-by: Stefan Schake <stschake at gmail.com>
---
 drivers/gpu/drm/vc4/vc4_gem.c | 30 ++++++++++++++++++++++++++++--
 include/uapi/drm/vc4_drm.h    |  6 ++++++
 2 files changed, 34 insertions(+), 2 deletions(-)

diff --git a/drivers/gpu/drm/vc4/vc4_gem.c b/drivers/gpu/drm/vc4/vc4_gem.c
index e305ccde..a4c4be3 100644
--- a/drivers/gpu/drm/vc4/vc4_gem.c
+++ b/drivers/gpu/drm/vc4/vc4_gem.c
@@ -656,7 +656,8 @@ vc4_lock_bo_reservations(struct drm_device *dev,
  */
 static int
 vc4_queue_submit(struct drm_device *dev, struct vc4_exec_info *exec,
-		 struct ww_acquire_ctx *acquire_ctx)
+		 struct ww_acquire_ctx *acquire_ctx,
+		 struct drm_syncobj *out_sync)
 {
 	struct vc4_dev *vc4 = to_vc4_dev(dev);
 	struct vc4_exec_info *renderjob;
@@ -679,6 +680,9 @@ vc4_queue_submit(struct drm_device *dev, struct vc4_exec_info *exec,
 	fence->seqno = exec->seqno;
 	exec->fence = &fence->base;
 
+	if (out_sync)
+		drm_syncobj_replace_fence(out_sync, exec->fence);
+
 	vc4_update_bo_seqnos(exec, seqno);
 
 	vc4_unlock_bo_reservations(dev, exec, acquire_ctx);
@@ -1114,6 +1118,7 @@ vc4_submit_cl_ioctl(struct drm_device *dev, void *data,
 	struct vc4_dev *vc4 = to_vc4_dev(dev);
 	struct vc4_file *vc4file = file_priv->driver_priv;
 	struct drm_vc4_submit_cl *args = data;
+	struct drm_syncobj *out_sync = NULL;
 	struct vc4_exec_info *exec;
 	struct ww_acquire_ctx acquire_ctx;
 	struct dma_fence *in_fence;
@@ -1201,12 +1206,33 @@ vc4_submit_cl_ioctl(struct drm_device *dev, void *data,
 	if (ret)
 		goto fail;
 
+	if (args->out_sync) {
+		out_sync = drm_syncobj_find(file_priv, args->out_sync);
+		if (!out_sync) {
+			ret = -EINVAL;
+			goto fail;
+		}
+
+		/* We replace the fence in out_sync in vc4_queue_submit since
+		 * the render job could execute immediately after that call.
+		 * If it finishes before our ioctl processing resumes the
+		 * render job fence could already have been freed.
+		 */
+	}
+
 	/* Clear this out of the struct we'll be putting in the queue,
 	 * since it's part of our stack.
 	 */
 	exec->args = NULL;
 
-	ret = vc4_queue_submit(dev, exec, &acquire_ctx);
+	ret = vc4_queue_submit(dev, exec, &acquire_ctx, out_sync);
+
+	/* The syncobj isn't part of the exec data and we need to free our
+	 * reference even if job submission failed.
+	 */
+	if (out_sync)
+		drm_syncobj_put(out_sync);
+
 	if (ret)
 		goto fail;
 
diff --git a/include/uapi/drm/vc4_drm.h b/include/uapi/drm/vc4_drm.h
index d97065b..2be4fe3 100644
--- a/include/uapi/drm/vc4_drm.h
+++ b/include/uapi/drm/vc4_drm.h
@@ -187,6 +187,12 @@ struct drm_vc4_submit_cl {
 	 * will not start until the syncobj is signaled. 0 means ignore.
 	 */
 	__u32 in_sync;
+
+	/* Syncobj handle to export fence to. If set, the fence in the syncobj
+	 * will be replaced with a fence that signals upon completion of this
+	 * render job. 0 means ignore.
+	 */
+	__u32 out_sync;
 };
 
 /**
-- 
2.7.4