TensorFlow: upstream changes to git.

Change 109695551
	Update FAQ
Change 109694725
	Add a gradient for resize_bilinear op.
Change 109694505
	Don't mention variables module in docs

	variables.Variable should be tf.Variable.
Change 109658848
	Adding an option to create a new thread-pool for each session.
Change 109640570

	Take the snapshot of stream-executor.
	+ Expose an interface for scratch space allocation in the interface.

Change 109638559
	Let image_summary accept uint8 input

	This allows users to do their own normalization / scaling if the default
	(very weird) behavior of image_summary is undesired.

	This required a slight tweak to fake_input.cc to make polymorphically typed
	fake inputs infer if their type attr is not set but has a default.

	Unfortunately, adding a second valid type to image_summary *disables* automatic
	implicit conversion from np.float64 to tf.float32, so this change is slightly
	backwards incompatible.
Change 109636969
	Add serialization operations for SparseTensor.
Change 109636644
	Update generated Op docs.
Change 109634899
	TensorFlow: add a markdown file for producing release notes for our
	releases.  Seed with 0.5.0 with a boring but accurate description.
Change 109634502
	Let histogram_summary take any realnumbertype

	It used to take only floats, not it understands ints.
Change 109634434
	TensorFlow: update locations where we mention python 3 support, update
	them to current truth.
Change 109632108
	Move HSV <> RGB conversions, grayscale conversions, and adjust_* ops back to tensorflow
	- make GPU-capable version of RGBToHSV and HSVToRGB, allows only float input/output
	- change docs to reflect new size constraints
	- change HSV format to be [0,1] for all components
	- add automatic dtype conversion for all adjust_* and grayscale conversion ops
	- fix up docs
Change 109631077
	Improve optimizer exceptions

	1. grads_and_vars is now a tuple, so must be wrapped when passed to format.
	2. Use '%r' instead of '%s' for dtype formatting

Base CL: 109697989

1 files changed, 11 insertions, 6 deletions

diff --git a/tensorflow/stream_executor/stream_executor_pimpl.h b/tensorflow/stream_executor/stream_executor_pimpl.h
index 5a59a5a26d..f624e0fcdb 100644
--- a/tensorflow/stream_executor/stream_executor_pimpl.h
+++ b/tensorflow/stream_executor/stream_executor_pimpl.h
@@ -71,9 +71,7 @@ class StreamExecutor {
  public:
   explicit StreamExecutor(PlatformKind kind,
                           const PluginConfig &plugin_config = PluginConfig());
-
-  // Primarily used for testing.
-  StreamExecutor(PlatformKind kind,
+  StreamExecutor(const Platform *platform,
                  internal::StreamExecutorInterface *implementation);
 
   ~StreamExecutor();
@@ -81,9 +79,13 @@ class StreamExecutor {
   port::Status Init();
   port::Status Init(int device_ordinal, DeviceOptions device_options);
 
+  // DEPRECATED: Do not use; use platform() instead.
   // Returns the platform that this StreamExecutor is acting upon.
   PlatformKind platform_kind() const { return platform_kind_; }
 
+  // Returns a reference to the platform that created this executor.
+  const Platform *platform() const { return platform_; }
+
   // Retrieves (loads) a kernel for the platform this StreamExecutor is acting
   // upon, if one exists.
   //
@@ -538,15 +540,18 @@ class StreamExecutor {
   // can acquire the lock on their first (mutating) call as well.
   mutable mutex mu_;
 
-  // A mapping of pointer (to GPU memory) to string representation of the stack
-  // (of the allocating thread) at the time at which the pointer was allocated.
-  std::map<void *, AllocRecord> mem_allocs_ GUARDED_BY(mu_);
+  // Reference to the platform that created this executor.
+  const Platform *platform_;
 
   // Pointer to the platform-specific-interface implementation. This is
   // delegated to by the interface routines in pointer-to-implementation
   // fashion.
   std::unique_ptr<internal::StreamExecutorInterface> implementation_;
 
+  // A mapping of pointer (to GPU memory) to string representation of the stack
+  // (of the allocating thread) at the time at which the pointer was allocated.
+  std::map<void *, AllocRecord> mem_allocs_ GUARDED_BY(mu_);
+
   // Memoized BLAS support object -- we only want to create this once when asked
   // for a BLAS interface.
   std::unique_ptr<blas::BlasSupport> blas_ GUARDED_BY(mu_);