pytorch 源码分析之 调用栈

(lldb) bt
* thread #1, queue = 'com.apple.main-thread', stop reason = step over
  * frame #0: 0x000000032727a57c libtorch_cpu.dylib`torch::autograd::VariableType::(anonymous namespace)::add_Tensor(ks=(repr_ = 137438986241), self=0x000000010019e7b8, other=0x000000016fdfe838, alpha=0x000000016fdfe940) at VariableType_2.cpp:6345:3
    frame #1: 0x000000032727b738 libtorch_cpu.dylib`c10::impl::wrap_kernel_functor_unboxed_<c10::impl::detail::WrapFunctionIntoFunctor_<c10::CompileTimeFunctionPointer<at::Tensor (c10::DispatchKeySet, at::Tensor const&, at::Tensor const&, c10::Scalar const&), &torch::autograd::VariableType::(anonymous namespace)::add_Tensor(c10::DispatchKeySet, at::Tensor const&, at::Tensor const&, c10::Scalar const&)>, at::Tensor, c10::guts::typelist::typelist<c10::DispatchKeySet, at::Tensor const&, at::Tensor const&, c10::Scalar const&>>, at::Tensor (c10::DispatchKeySet, at::Tensor const&, at::Tensor const&, c10::Scalar const&)>::call(c10::OperatorKernel*, c10::DispatchKeySet, at::Tensor const&, at::Tensor const&, c10::Scalar const&) [inlined] c10::impl::detail::WrapFunctionIntoFunctor_<c10::CompileTimeFunctionPointer<at::Tensor (c10::DispatchKeySet, at::Tensor const&, at::Tensor const&, c10::Scalar const&), &torch::autograd::VariableType::(anonymous namespace)::add_Tensor(c10::DispatchKeySet, at::Tensor const&, at::Tensor const&, c10::Scalar const&)>, at::Tensor, c10::guts::typelist::typelist<c10::DispatchKeySet, at::Tensor const&, at::Tensor const&, c10::Scalar const&>>::operator()(this=0x00006000008dbe00, args=(repr_ = 137438986241), args=0x000000010019e7b8, args=0x000000016fdfe838, args=0x000000016fdfe940) at WrapFunctionIntoFunctor.h:13:16
    frame #2: 0x000000032727b710 libtorch_cpu.dylib`c10::impl::wrap_kernel_functor_unboxed_<c10::impl::detail::WrapFunctionIntoFunctor_<c10::CompileTimeFunctionPointer<at::Tensor (c10::DispatchKeySet, at::Tensor const&, at::Tensor const&, c10::Scalar const&), &torch::autograd::VariableType::(anonymous namespace)::add_Tensor(c10::DispatchKeySet, at::Tensor const&, at::Tensor const&, c10::Scalar const&)>, at::Tensor, c10::guts::typelist::typelist<c10::DispatchKeySet, at::Tensor const&, at::Tensor const&, c10::Scalar const&>>, at::Tensor (c10::DispatchKeySet, at::Tensor const&, at::Tensor const&, c10::Scalar const&)>::call(functor=0x00006000008dbe00, dispatchKeySet=(repr_ = 137438986241), args=0x000000010019e7b8, args=0x000000016fdfe838, args=0x000000016fdfe940) at make_boxed_from_unboxed_functor.h:485:14
    frame #3: 0x0000000323408ea8 libtorch_cpu.dylib`at::Tensor c10::callUnboxedKernelFunction<at::Tensor, at::Tensor const&, at::Tensor const&, c10::Scalar const&>(unboxed_kernel_func=0x000000032727b6a4, functor=0x00006000008dbe00, dispatchKeySet=(repr_ = 137438986241), args=0x000000010019e7b8, args=0x000000016fdfe838, args=0x000000016fdfe940) at KernelFunction_impl.h:53:12
    frame #4: 0x0000000323814984 libtorch_cpu.dylib`at::_ops::add_Tensor::call(at::Tensor const&, at::Tensor const&, c10::Scalar const&) at KernelFunction_impl.h:104:16
    frame #5: 0x0000000323814934 libtorch_cpu.dylib`at::_ops::add_Tensor::call(at::Tensor const&, at::Tensor const&, c10::Scalar const&) [inlined] at::Tensor c10::Dispatcher::call<at::Tensor, at::Tensor const&, at::Tensor const&, c10::Scalar const&>(this=0x000000032c0c6ab8, op=0x000000032c0db150, args=0x000000010019e7b8, args=0x000000016fdfe838, args=0x000000016fdfe940) const at Dispatcher.h:698:28
    frame #6: 0x0000000323814650 libtorch_cpu.dylib`at::_ops::add_Tensor::call(at::Tensor const&, at::Tensor const&, c10::Scalar const&) [inlined] c10::TypedOperatorHandle<at::Tensor (at::Tensor const&, at::Tensor const&, c10::Scalar const&)>::call(this=0x000000032c0db150, args=0x000000010019e7b8, args=0x000000016fdfe838, args=0x000000016fdfe940) const at Dispatcher.h:531:41
    frame #7: 0x0000000323814590 libtorch_cpu.dylib`at::_ops::add_Tensor::call(self=0x000000010019e7b8, other=0x000000016fdfe838, alpha=0x000000016fdfe940) at Operators_2.cpp:1077:15
    frame #8: 0x0000000104e1c804 libtorch_python.dylib`at::Tensor::add(this=0x000000010019e7b8, other=0x000000016fdfe838, alpha=0x000000016fdfe940) const at TensorBody.h:1664:12
    frame #9: 0x0000000104e1c79c libtorch_python.dylib`torch::autograd::THPVariable_add(_object*, _object*, _object*)::$_1::operator()(this=0x000000016fdfe84f, self=0x000000010019e7b8, other=0x000000016fdfe838, alpha=0x000000016fdfe940) const at python_variable_methods.cpp:2743:21
    frame #10: 0x0000000104d4a248 l

在mac上手动编译pytorch源码，并且安装这个编译的模块，然后使用lldb查看完整的add 调用栈。 mac 上目前还无法使用gdb。  

lldb python3.10
b at::Tensor::add
run                       # 执行完run就进入了python环境
import torch
a= torch.Tensor([1,2])
b= torch.Tensor([1,2])
a + b

 

build/aten/./src/ATen/Operators_2.cpp   pytorch这个文件是自动生成的 

 

 

 

 

参考：

https://zhuanlan.zhihu.com/p/567978232