A few fixes to prevent compositor crashes

If a client application exits abruptly while in direct
rendering mode, the surface buffer could reference a null buffer.
We now check for this before attempting to use such a surface
buffer.  It is also possible for WaylandSurfaceNode::preprocess to
be called after WaylandSurfaceItem has called surfaceDestroyed, so
now we check for a valid surface in preprocess as well.

Change-Id: Ibca71349b7a24687944f8b6cd505941b580e0c2a
Reviewed-by: Paul Olav Tvete <paul.tvete@nokia.com>
Reviewed-by: Jørgen Lind <jorgen.lind@nokia.com>

1 files changed, 6 insertions, 4 deletions

diff --git a/src/compositor/compositor_api/waylandsurfacenode.cpp b/src/compositor/compositor_api/waylandsurfacenode.cpp
index e6571751c..a4063ccb4 100644
--- a/src/compositor/compositor_api/waylandsurfacenode.cpp
+++ b/src/compositor/compositor_api/waylandsurfacenode.cpp
@@ -39,10 +39,12 @@ void WaylandSurfaceNode::preprocess()
 {
     QMutexLocker locker(WaylandSurfaceItem::mutex);
 
-    //Update if the item is dirty and we haven't done an updateTexture for this frame
-    if (m_item && m_item->m_damaged && !m_textureUpdated) {
-        m_item->updateTexture();
-        updateTexture();
+    if (m_item && m_item->surface()) {
+        //Update if the item is dirty and we haven't done an updateTexture for this frame
+        if (m_item->m_damaged && !m_textureUpdated) {
+            m_item->updateTexture();
+            updateTexture();
+        }
     }
     //Reset value for next frame: we have not done updatePaintNode yet
     m_textureUpdated = false;