[PATCH v2 6/6] drm/v3d: Invalidate the caches from the outside in.

From: Eric Anholt <eric@anholt.net>
To: dri-devel@lists.freedesktop.org
Cc: linux-kernel@vger.kernel.org,
	Dave Emett <david.emett@broadcom.com>,
	Thomas Spurden <thomas.spurden@broadcom.com>,
	Eric Anholt <eric@anholt.net>
Subject: [PATCH v2 6/6] drm/v3d: Invalidate the caches from the outside in.
Date: Mon,  3 Dec 2018 14:24:38 -0800	[thread overview]
Message-ID: <20181203222438.25417-6-eric@anholt.net> (raw)
In-Reply-To: <20181203222438.25417-1-eric@anholt.net>

This would be a fairly obscure race, but let's make sure we don't ever
lose it.

Signed-off-by: Eric Anholt <eric@anholt.net>
---
 drivers/gpu/drm/v3d/v3d_gem.c | 8 ++++++--
 1 file changed, 6 insertions(+), 2 deletions(-)

diff --git a/drivers/gpu/drm/v3d/v3d_gem.c b/drivers/gpu/drm/v3d/v3d_gem.c
index 8a4be9515179..443b1c53117a 100644
--- a/drivers/gpu/drm/v3d/v3d_gem.c
+++ b/drivers/gpu/drm/v3d/v3d_gem.c
@@ -171,11 +171,15 @@ v3d_invalidate_slices(struct v3d_dev *v3d, int core)
 void
 v3d_invalidate_caches(struct v3d_dev *v3d)
 {
+	/* Invalidate the caches from the outside in.  That way if
+	 * another CL's concurrent use of nearby memory were to pull
+	 * an invalidated cacheline back in, we wouldn't leave stale
+	 * data in the inner cache.
+	 */
 	v3d_flush_l3(v3d);
-
 	v3d_invalidate_l2c(v3d, 0);
-	v3d_invalidate_slices(v3d, 0);
 	v3d_flush_l2t(v3d, 0);
+	v3d_invalidate_slices(v3d, 0);
 }
 
 static void
-- 
2.20.0.rc1

