hwc: prevent bandwidth calculation in GPU fall-back case

    In prepare, we check if MDP can handle the number of layers
    given to it. If it cannot, we fall-back. Additionally, we must
    not proceed with any MDP bytes calculations otherwise we will
    see out of bounds errors.

Change-Id: Ifae45aefbc8732a7f21f5652758ece3a7639be9a
diff --git a/libhwcomposer/hwc_mdpcomp.cpp b/libhwcomposer/hwc_mdpcomp.cpp
index 44db2bc..cbfc297 100644
--- a/libhwcomposer/hwc_mdpcomp.cpp
+++ b/libhwcomposer/hwc_mdpcomp.cpp
@@ -796,10 +796,10 @@
     //do not cache the information for next draw cycle.
     if(numLayers > MAX_NUM_APP_LAYERS) {
         mCachedFrame.updateCounts(mCurrentFrame);
-        ALOGD_IF(isDebug(), "%s: Number of App layers exceeded the limit ",
-                __FUNCTION__);
+        ALOGE("%s: Number of App layers exceeded the limit ",
+        __FUNCTION__);
         ret = -1;
-        goto exit;
+        return ret;
     }
 
     //Hard conditions, if not met, cannot do MDP comp