Reland of "gpu: workaround force_cube_map_positive_x_allocation fixes Android crash."

gpu/command_buffer/service/texture_manager.h

Index: gpu/command_buffer/service/texture_manager.h
diff --git a/gpu/command_buffer/service/texture_manager.h b/gpu/command_buffer/service/texture_manager.h
index a255ad40983f17f30a5a70eae8c42f81db4bcaaa..479678cddb0ed39b0145df1bfba26b965e8ab57e 100644
--- a/gpu/command_buffer/service/texture_manager.h
+++ b/gpu/command_buffer/service/texture_manager.h
@@ -13,6 +13,7 @@
 #include "base/basictypes.h"
 #include "base/containers/hash_tables.h"
 #include "base/memory/ref_counted.h"
+#include "gpu/command_buffer/service/feature_info.h"
 #include "gpu/command_buffer/service/gl_utils.h"
 #include "gpu/command_buffer/service/memory_tracking.h"
 #include "gpu/gpu_export.h"
@@ -535,10 +536,14 @@ class GPU_EXPORT TextureRef : public base::RefCounted<TextureRef> {
 struct DecoderTextureState {
   // total_texture_upload_time automatically initialized to 0 in default
   // constructor.
-  explicit DecoderTextureState(bool texsubimage_faster_than_teximage)
+  explicit DecoderTextureState(const FeatureInfo::Workarounds& workarounds)
       : tex_image_failed(false),
         texture_upload_count(0),
-        texsubimage_faster_than_teximage(texsubimage_faster_than_teximage) {}
+        texsubimage_faster_than_teximage(
+            workarounds.texsubimage_faster_than_teximage),
+        force_cube_map_positive_x_allocation(
+            workarounds.force_cube_map_positive_x_allocation),
+        force_cube_complete(workarounds.force_cube_complete) {}
 
   // This indicates all the following texSubImage*D calls that are part of the
   // failed texImage*D call should be ignored. The client calls have a lock
@@ -551,6 +556,8 @@ struct DecoderTextureState {
   base::TimeDelta total_texture_upload_time;
 
   bool texsubimage_faster_than_teximage;
+  bool force_cube_map_positive_x_allocation;
+  bool force_cube_complete;
 };
 
 // This class keeps track of the textures and their sizes so we can do NPOT and