如何修复 OSX 上的 tensorflow protobuf 编译错误？

Question

我在查看 repo 后尝试编译 TensorFlow。 我已经到了遇到 google protobuf 错误的地步：

INFO: From Compiling tensorflow/core/kernels/histogram_op_gpu.cu.cc:
./tensorflow/core/lib/core/status.h(32): warning: attribute "warn_unused_result" does not apply here

external/protobuf_archive/src/google/protobuf/arena.h(719): error: more than one instance of overloaded function "google::protobuf::Arena::CreateMessageInternal" matches the argument list:
            function template "T *google::protobuf::Arena::CreateMessageInternal<T>(google::protobuf::Arena *)"
            function template "T *google::protobuf::Arena::CreateMessageInternal<T,Args...>(Args &&...)"
            argument types are: (google::protobuf::Arena *)
          detected during:
            instantiation of "Msg *google::protobuf::Arena::CreateMaybeMessage<Msg>(google::protobuf::Arena *, google::protobuf::internal::true_type) [with Msg=tensorflow::TensorShapeProto_Dim]" 
(729): here
            instantiation of "T *google::protobuf::Arena::CreateMaybeMessage<T>(google::protobuf::Arena *) [with T=tensorflow::TensorShapeProto_Dim]" 
external/protobuf_archive/src/google/protobuf/repeated_field.h(648): here
            instantiation of "GenericType *google::protobuf::internal::GenericTypeHandler<GenericType>::New(google::protobuf::Arena *) [with GenericType=tensorflow::TensorShapeProto_Dim]" 
external/protobuf_archive/src/google/protobuf/repeated_field.h(675): here
            instantiation of "GenericType *google::protobuf::internal::GenericTypeHandler<GenericType>::NewFromPrototype(const GenericType *, google::protobuf::Arena *) [with GenericType=tensorflow::TensorShapeProto_Dim]" 
external/protobuf_archive/src/google/protobuf/repeated_field.h(1554): here
            instantiation of "TypeHandler::Type *google::protobuf::internal::RepeatedPtrFieldBase::Add<TypeHandler>(TypeHandler::Type *) [with TypeHandler=google::protobuf::RepeatedPtrField<tensorflow::TensorShapeProto_Dim>::TypeHandler]" 
external/protobuf_archive/src/google/protobuf/repeated_field.h(2001): here
            instantiation of "Element *google::protobuf::RepeatedPtrField<Element>::Add() [with Element=tensorflow::TensorShapeProto_Dim]" 
bazel-out/local_darwin-opt/genfiles/tensorflow/core/framework/tensor_shape.pb.h(471): here

....

有人遇到过这个问题吗？关于如何解决这个问题的任何想法？

（我在 OSX 10.11.5 的虚拟环境中使用 Python 2.7）

Answer 1

幸运的是，其他人不仅遇到了同样的问题，而且还找到了解决方法并分享了它。感谢 Daniel Trebbien 的 comments on protobuf 和 eigen 我可以在 OSX 上编译带有 GPU 支持的 tensorflow：

>>> import tensorflow as tf
>>> tf.__version__
'1.6.0-rc0'
>>> sess = tf.Session(config=tf.ConfigProto(log_device_placement=True))
2018-02-19 22:22:12.194516: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:859] OS X does not support NUMA - returning NUMA node zero
2018-02-19 22:22:12.195011: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1331] Found device 0 with properties: 
name: GeForce GT 750M major: 3 minor: 0 memoryClockRate(GHz): 0.9255
pciBusID: 0000:01:00.0
totalMemory: 2.00GiB freeMemory: 12.58MiB
2018-02-19 22:22:12.195038: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1410] Adding visible gpu devices: 0
2018-02-19 22:22:14.563665: I tensorflow/core/common_runtime/gpu/gpu_device.cc:911] Device interconnect StreamExecutor with strength 1 edge matrix:
2018-02-19 22:22:14.563700: I tensorflow/core/common_runtime/gpu/gpu_device.cc:917]      0 
2018-02-19 22:22:14.563707: I tensorflow/core/common_runtime/gpu/gpu_device.cc:930] 0:   N 
2018-02-19 22:22:14.563798: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1021] Creating TensorFlow device (/job:localhost/replica:0/task:0/device:GPU:0 with 65 MB memory) -> physical GPU (device: 0, name: GeForce GT 750M, pci bus id: 0000:01:00.0, compute capability: 3.0)
Device mapping:
/job:localhost/replica:0/task:0/device:GPU:0 -> device: 0, name: GeForce GT 750M, pci bus id: 0000:01:00.0, compute capability: 3.0
2018-02-19 22:22:14.697626: I tensorflow/core/common_runtime/direct_session.cc:297] Device mapping:
/job:localhost/replica:0/task:0/device:GPU:0 -> device: 0, name: GeForce GT 750M, pci bus id: 0000:01:00.0, compute capability: 3.0

作为参考，以下是 cmets 中提出的补丁：

--- a/tensorflow/workspace.bzl
+++ b/tensorflow/workspace.bzl
@@ -353,11 +353,11 @@ def tf_workspace(path_prefix="", tf_repo_name=""):
   tf_http_archive(
       name = "protobuf_archive",
       urls = [
-          "https://mirror.bazel.build/github.com/google/protobuf/archive/396336eb961b75f03b25824fe86cf6490fb75e3a.tar.gz",
-          "https://github.com/google/protobuf/archive/396336eb961b75f03b25824fe86cf6490fb75e3a.tar.gz",
+          "https://mirror.bazel.build/github.com/dtrebbien/protobuf/archive/50f552646ba1de79e07562b41f3999fe036b4fd0.tar.gz",
+          "https://github.com/dtrebbien/protobuf/archive/50f552646ba1de79e07562b41f3999fe036b4fd0.tar.gz",
       ],
-      sha256 = "846d907acf472ae233ec0882ef3a2d24edbbe834b80c305e867ac65a1f2c59e3",
-      strip_prefix = "protobuf-396336eb961b75f03b25824fe86cf6490fb75e3a",
+      sha256 = "eb16b33431b91fe8cee479575cee8de202f3626aaf00d9bf1783c6e62b4ffbc7",
+      strip_prefix = "protobuf-50f552646ba1de79e07562b41f3999fe036b4fd0",
   )

--- a/tensorflow/workspace.bzl
+++ b/tensorflow/workspace.bzl
@@ -120,11 +120,11 @@ def tf_workspace(path_prefix="", tf_repo_name=""):
   tf_http_archive(
       name = "eigen_archive",
       urls = [
-          "https://mirror.bazel.build/bitbucket.org/eigen/eigen/get/2355b229ea4c.tar.gz",
-          "https://bitbucket.org/eigen/eigen/get/2355b229ea4c.tar.gz",
+          "https://mirror.bazel.build/bitbucket.org/dtrebbien/eigen/get/374842a18727.tar.gz",
+          "https://bitbucket.org/dtrebbien/eigen/get/374842a18727.tar.gz",
       ],
-      sha256 = "0cadb31a35b514bf2dfd6b5d38205da94ef326ec6908fc3fd7c269948467214f",
-      strip_prefix = "eigen-eigen-2355b229ea4c",
+      sha256 = "fa26e9b9ff3a2692b092d154685ec88d6cb84d4e1e895006541aff8603f15c16",
+      strip_prefix = "dtrebbien-eigen-374842a18727",
       build_file = str(Label("//third_party:eigen.BUILD")),
   )