diff options
| author | Hongxu Jia <hongxu.jia@windriver.com> | 2020-07-06 19:11:29 -0700 |
|---|---|---|
| committer | Hongxu Jia <hongxu.jia@windriver.com> | 2020-07-07 10:15:51 +0800 |
| commit | 17fe99fa101a51d62e55e0d4372b9301f1f5fbe3 (patch) | |
| tree | 20cb420ae60cff04828c96dd818aa5cf737a4db7 | |
| parent | 973ea0e89cf0059a8efd0cfe3dad60d8a553a99f (diff) | |
| download | meta-tensorflow-master.tar.gz meta-tensorflow-master.tar.bz2 meta-tensorflow-master.zip | |
Since python3-numpy: update 1.18.5 -> 1.19.0, there is
a GCC 10.1 compile error. Backport a patch from upstream
to fix it [1]
[1] https://github.com/tensorflow/tensorflow/pull/40654
Signed-off-by: Hongxu Jia <hongxu.jia@windriver.com>
| -rw-r--r-- | recipes-framework/tensorflow/files/0001-Provide-overload-to-cope-with-const-ness-change-of-N.patch | 52 | ||||
| -rw-r--r-- | recipes-framework/tensorflow/tensorflow.inc | 1 |
2 files changed, 53 insertions, 0 deletions
diff --git a/recipes-framework/tensorflow/files/0001-Provide-overload-to-cope-with-const-ness-change-of-N.patch b/recipes-framework/tensorflow/files/0001-Provide-overload-to-cope-with-const-ness-change-of-N.patch new file mode 100644 index 0000000..8d9411c --- /dev/null +++ b/recipes-framework/tensorflow/files/0001-Provide-overload-to-cope-with-const-ness-change-of-N.patch @@ -0,0 +1,52 @@ +From 75ea0b31477d6ba9e990e296bbbd8ca4e7eebadf Mon Sep 17 00:00:00 2001 +From: Christian Sigg <csigg@google.com> +Date: Fri, 26 Jun 2020 05:08:10 -0700 +Subject: [PATCH] Provide overload to cope with const-ness change of NumPy's + PyUFuncGenericFunction. + +See https://github.com/tensorflow/tensorflow/issues/40688, https://github.com/tensorflow/tensorflow/pull/40654. + +PiperOrigin-RevId: 318452381 +Change-Id: Icc5152f2b020ef19882a49e3c86ac80bbe048d64 + +Upstream-Status: Backport +Signed-off-by: Hongxu Jia <hongxu.jia@windriver.com> + +--- + tensorflow/python/lib/core/bfloat16.cc | 8 +++++++- + 1 file changed, 7 insertions(+), 1 deletion(-) + +diff --git a/tensorflow/python/lib/core/bfloat16.cc b/tensorflow/python/lib/core/bfloat16.cc +index feb01f11a1..bb6b720feb 100644 +--- a/tensorflow/python/lib/core/bfloat16.cc ++++ b/tensorflow/python/lib/core/bfloat16.cc +@@ -517,7 +517,7 @@ bool RegisterBfloat16Cast(int numpy_type, bool cast_is_safe) { + } + + template <typename InType, typename OutType, typename Functor> +-void BinaryUFunc(char** args, npy_intp* dimensions, npy_intp* steps, ++void BinaryUFunc(char** args, const npy_intp* dimensions, const npy_intp* steps, + void* data) { + const char* i0 = args[0]; + const char* i1 = args[1]; +@@ -532,11 +532,17 @@ void BinaryUFunc(char** args, npy_intp* dimensions, npy_intp* steps, + } + } + ++// Numpy changed const-ness of PyUFuncGenericFunction, provide overload. + template <typename Functor> + void CompareUFunc(char** args, npy_intp* dimensions, npy_intp* steps, + void* data) { + BinaryUFunc<bfloat16, npy_bool, Functor>(args, dimensions, steps, data); + } ++template <typename Functor> ++void CompareUFunc(char** args, const npy_intp* dimensions, ++ const npy_intp* steps, void* data) { ++ BinaryUFunc<bfloat16, npy_bool, Functor>(args, dimensions, steps, data); ++} + + struct Bfloat16EqFunctor { + npy_bool operator()(bfloat16 a, bfloat16 b) { return a == b; } +-- +2.21.0 + diff --git a/recipes-framework/tensorflow/tensorflow.inc b/recipes-framework/tensorflow/tensorflow.inc index 4b06c5d..a729bdd 100644 --- a/recipes-framework/tensorflow/tensorflow.inc +++ b/recipes-framework/tensorflow/tensorflow.inc @@ -12,6 +12,7 @@ SRC_URI = "git://github.com/tensorflow/tensorflow.git;branch=r1.13 \ file://0001-Fix-TensorFlow-on-Python-3.8-logger-issue.patch \ file://0001-Rename-gettid-functions.patch \ file://0001-third_party-eigen_archive-workaround-ice-failure-whi.patch \ + file://0001-Provide-overload-to-cope-with-const-ness-change-of-N.patch \ " S = "${WORKDIR}/git" |