Re: [PATCH 01/19] drm: Wire up proper acquire ctx for plane functions

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 





On 2017-03-28 03:02 AM, Daniel Vetter wrote:
On Tue, Mar 28, 2017 at 08:23:53AM +0200, Daniel Vetter wrote:
On Mon, Mar 27, 2017 at 04:12:05PM -0400, Harry Wentland wrote:
On Wednesday, March 22, 2017 10:50:40 PM EDT Daniel Vetter wrote:
This is just prep work to get an acquire ctx into every place where we
call ->update_plane or ->disable_plane.

v2: Keep the hidden acquire_ctx pointers valid while transitioning.

Signed-off-by: Daniel Vetter <daniel.vetter@xxxxxxxxx>
---
 drivers/gpu/drm/drm_plane.c | 52
++++++++++++++++++++++++++++++++++++--------- 1 file changed, 42
insertions(+), 10 deletions(-)

diff --git a/drivers/gpu/drm/drm_plane.c b/drivers/gpu/drm/drm_plane.c
index a22e76837065..0d04888172a7 100644
--- a/drivers/gpu/drm/drm_plane.c
+++ b/drivers/gpu/drm/drm_plane.c
@@ -457,7 +457,8 @@ static int __setplane_internal(struct drm_plane *plane,
 			       uint32_t crtc_w, uint32_t crtc_h,
 			       /* src_{x,y,w,h} values are 16.16 fixed point */
 			       uint32_t src_x, uint32_t src_y,
-			       uint32_t src_w, uint32_t src_h)
+			       uint32_t src_w, uint32_t src_h,
+			       struct drm_modeset_acquire_ctx *ctx)
 {
 	int ret = 0;

@@ -537,13 +538,26 @@ static int setplane_internal(struct drm_plane *plane,
 			     uint32_t src_x, uint32_t src_y,
 			     uint32_t src_w, uint32_t src_h)
 {
+	struct drm_modeset_acquire_ctx ctx;
 	int ret;

-	drm_modeset_lock_all(plane->dev);
+	drm_modeset_acquire_init(&ctx, 0);
+retry:
+	ret = drm_modeset_lock_all_ctx(plane->dev, &ctx);
+	if (ret)
+		goto fail;
+	plane->dev->mode_config.acquire_ctx = &ctx;
 	ret = __setplane_internal(plane, crtc, fb,
 				  crtc_x, crtc_y, crtc_w, crtc_h,
-				  src_x, src_y, src_w, src_h);
-	drm_modeset_unlock_all(plane->dev);
+				  src_x, src_y, src_w, src_h, &ctx);
+
+fail:
+	if (ret == -EDEADLK) {
+		drm_modeset_backoff(&ctx);
+		goto retry;
+	}
+	drm_modeset_drop_locks(&ctx);
+	drm_modeset_acquire_fini(&ctx);

 	return ret;
 }
@@ -613,11 +627,22 @@ static int drm_mode_cursor_universal(struct drm_crtc
*crtc, int32_t crtc_x, crtc_y;
 	uint32_t crtc_w = 0, crtc_h = 0;
 	uint32_t src_w = 0, src_h = 0;
+	struct drm_modeset_acquire_ctx ctx;
 	int ret = 0;

 	BUG_ON(!crtc->cursor);
 	WARN_ON(crtc->cursor->crtc != crtc && crtc->cursor->crtc != NULL);

+	drm_modeset_acquire_init(&ctx, 0);
+retry:
+	ret = drm_modeset_lock(&crtc->mutex, &ctx);
+	if (ret)
+		goto fail;
+	ret = drm_modeset_lock(&crtc->cursor->mutex, &ctx);
+	if (ret)
+		goto fail;
+	crtc->acquire_ctx = &ctx;
+
 	/*
 	 * Obtain fb we'll be using (either new or existing) and take an extra
 	 * reference to it if fb != null.  setplane will take care of dropping
@@ -662,7 +687,7 @@ static int drm_mode_cursor_universal(struct drm_crtc
*crtc, */
 	ret = __setplane_internal(crtc->cursor, crtc, fb,
 				crtc_x, crtc_y, crtc_w, crtc_h,
-				0, 0, src_w, src_h);
+				0, 0, src_w, src_h, &ctx);

 	/* Update successful; save new cursor position, if necessary */
 	if (ret == 0 && req->flags & DRM_MODE_CURSOR_MOVE) {
@@ -670,6 +695,15 @@ static int drm_mode_cursor_universal(struct drm_crtc
*crtc, crtc->cursor_y = req->y;
 	}

+fail:
+	if (ret == -EDEADLK) {
+		drm_modeset_backoff(&ctx);
+		goto retry;
+	}
+
+	drm_modeset_drop_locks(&ctx);
+	drm_modeset_acquire_fini(&ctx);
+
 	return ret;
 }

@@ -696,12 +730,10 @@ static int drm_mode_cursor_common(struct drm_device
*dev, * If this crtc has a universal cursor plane, call that plane's update
* handler rather than using legacy cursor handlers.
 	 */
-	drm_modeset_lock_crtc(crtc, crtc->cursor);
-	if (crtc->cursor) {
-		ret = drm_mode_cursor_universal(crtc, req, file_priv);
-		goto out;
-	}
+	if (crtc->cursor)
+		return drm_mode_cursor_universal(crtc, req, file_priv);

+	drm_modeset_lock_crtc(crtc, crtc->cursor);

Why are you moving the lock? Shouldn't it still be covering the call to
drm_mode_cursor_universal?

It would deadlock, because the new acquire ctx dance for universal planes
also takes the same locks as this helper?

But I just realized that crtc->cursor is NULL here, I can simplify this
some more.

Specifically this gets reshuffled in patch 14, but that is blocked on the
vmwgfx story for now.

Ah, right, the lock moves down into drm_mode_cursor_universal. I don't know how I missed that.

Reviewed-by: Harry Wentland <harry.wentland@xxxxxxx>

Harry


-Daniel

_______________________________________________
Intel-gfx mailing list
Intel-gfx@xxxxxxxxxxxxxxxxxxxxx
https://lists.freedesktop.org/mailman/listinfo/intel-gfx




[Index of Archives]     [Linux USB Devel]     [Linux Audio Users]     [Yosemite News]     [Linux Kernel]     [Linux SCSI]
  Powered by Linux