gistfile1.txt

From a4cf36d4f918195c7d39e58a893aac6f528514a4 Mon Sep 17 00:00:00 2001
From: Themaister <[email protected]>
Date: Thu, 20 Sep 2012 16:09:33 +0200
Subject: [PATCH] Always assume that page flip works in radeon.

This is a hack that avoids prolonged stalling in userspace when a page
flip doesn't go through on first try. This patch assumes that it is more
important that userspace is allowed to continue rendering new frames
rather than stalling for a whole frame waiting for old data to render.
---
 drivers/gpu/drm/radeon/radeon_display.c | 7 +++++++
 1 file changed, 7 insertions(+)

diff --git a/drivers/gpu/drm/radeon/radeon_display.c b/drivers/gpu/drm/radeon/radeon_display.c
index 7ddef8f..aed70cb 100644
--- a/drivers/gpu/drm/radeon/radeon_display.c
+++ b/drivers/gpu/drm/radeon/radeon_display.c
@@ -297,6 +297,13 @@ void radeon_crtc_handle_flip(struct radeon_device *rdev, int crtc_id)
 		radeon_crtc->deferred_flip_completion = 0;
 	}
 
+	/* HACK: Assume that page flip has completed to avoid stalling,
+	 * waiting for a page-flip that isn't *that* important anyways. */
+	if (update_pending) {
+		DRM_DEBUG_KMS("Page flip was deferred, but ignored it.\n");
+		update_pending = 0;
+	}
+
 	/* Has the pageflip already completed in crtc, or is it certain
 	 * to complete in this vblank?
 	 */
-- 
1.7.12