Use texture barriers to read directly from the RT

src/gpu/GrTargetCommands.h

Index: src/gpu/GrTargetCommands.h
diff --git a/src/gpu/GrTargetCommands.h b/src/gpu/GrTargetCommands.h
index cff9898dadcbab75c606ec7284646dd26b11bbf7..4ab41dbaf1e311b01bfc87e86d1b28116c9b5890 100644
--- a/src/gpu/GrTargetCommands.h
+++ b/src/gpu/GrTargetCommands.h
@@ -56,6 +56,7 @@ public:
             kDrawPath_CmdType          = 6,
             kDrawPaths_CmdType         = 7,
             kDrawBatch_CmdType         = 8,
+            kXferBarrier_CmdType       = 9,
         };
 
         Cmd(CmdType type) : fMarkerID(-1), fType(type) {}
@@ -151,6 +152,8 @@ private:
                                                           GrBatch*,
                                                           const GrDrawTarget::PipelineInfo&);
 
+    void recordXferBarrierIfNecessary(GrInOrderDrawBuffer*, const GrDrawTarget::PipelineInfo&);
+
     struct Draw : public Cmd {
         Draw(const GrDrawTarget::DrawInfo& info) : Cmd(kDraw_CmdType), fInfo(info) {}
 
@@ -276,6 +279,12 @@ private:
         const GrPipeline* getPipeline() const {
             return reinterpret_cast<const GrPipeline*>(fPipeline.get());
         }
+        GrRenderTarget* getRenderTarget() const {
+            return this->getPipeline()->getRenderTarget();
+        }
+        const GrXferProcessor* getXferProcessor() const {
+            return this->getPipeline()->getXferProcessor();
+        }
 
         void execute(GrGpu*, const SetState*) override;
 
@@ -303,6 +312,14 @@ private:
         GrBatchTarget*         fBatchTarget;
     };
 
+    struct XferBarrier : public Cmd {
+        XferBarrier() : Cmd(kXferBarrier_CmdType) {}
+
+        void execute(GrGpu*, const SetState*) override;
+
+        GrXferBarrierType   fBarrierType;
+    };
+
      static const int kCmdBufferInitialSizeInBytes = 8 * 1024;
 
      typedef void* TCmdAlign; // This wouldn't be enough align if a command used long double.