Refactor vulkan buffer mapping and unmapping

src/gpu/vk/GrVkGpu.cpp

Index: src/gpu/vk/GrVkGpu.cpp
diff --git a/src/gpu/vk/GrVkGpu.cpp b/src/gpu/vk/GrVkGpu.cpp
index a3393777bb924324e5f7af7f304716c71a014a28..82e7394330459fa6ccd78c4db6cac7ad63503682 100644
--- a/src/gpu/vk/GrVkGpu.cpp
+++ b/src/gpu/vk/GrVkGpu.cpp
@@ -601,14 +601,6 @@ bool GrVkGpu::uploadTexDataOptimal(GrVkTexture* tex,
 
     transferBuffer->unmap();
 
-    // make sure the unmap has finished
-    transferBuffer->addMemoryBarrier(this,

[egdaniel, 2016/09/19 15:17:24]

Removed this barrier, but I actually don't think this was ever needed since cpu
writes are sinked implicitly on commnad buffer submits to queue.

-                                     VK_ACCESS_HOST_WRITE_BIT,
-                                     VK_ACCESS_TRANSFER_READ_BIT,
-                                     VK_PIPELINE_STAGE_TOP_OF_PIPE_BIT,
-                                     VK_PIPELINE_STAGE_TRANSFER_BIT,
-                                     false);
-
     // Change layout of our target so it can be copied to
     tex->setImageLayout(this,
                         VK_IMAGE_LAYOUT_TRANSFER_DST_OPTIMAL,