CompositorImplAndroid sets a texture size that may not be aligned FencedAllocator allocation size.

This CL ensures chunk size is aligned with the FencedAllocator allocation size.

BUG=680150
CQ_INCLUDE_TRYBOTS=master.tryserver.chromium.linux:linux_optional_gpu_tests_rel;master.tryserver.chromium.mac:mac_optional_gpu_tests_rel;master.tryserver.chromium.win:win_optional_gpu_tests_rel

Review-Url: https://codereview.chromium.org/2623223002
Cr-Commit-Position: refs/heads/master@{#443435}

gpu/command_buffer/client/gles2_implementation.cc

@@ -18,6 +18,7 @@
 #include <sstream>
 #include <string>
 #include "base/atomic_sequence_num.h"
+#include "base/bits.h"
 #include "base/compiler_specific.h"
 #include "base/numerics/safe_math.h"
 #include "base/strings/string_split.h"
@@ -219,6 +220,8 @@ bool GLES2Implementation::Initialize(
   if (mapped_memory_limit != SharedMemoryLimits::kNoLimit) {
     // Use smaller chunks if the client is very memory conscientious.
     chunk_size = std::min(mapped_memory_limit / 4, chunk_size);
+    chunk_size = base::bits::Align(chunk_size,
+                                   FencedAllocator::kAllocAlignment);
   }
   mapped_memory_->set_chunk_size_multiple(chunk_size);