mirror of
https://github.com/zebrajr/pytorch.git
synced 2025-12-07 00:21:07 +01:00
328cfd50e7
Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/72607
since python isn't available from libtorch, most of lazy tensor
code can't depend on python.
separate python_util into libtorch_python library
make debug_util and IR dump work with or without python by providing
a default function for 'maybe getting python stacktrace' that returns
an empty stacktrace
use a registration mechanism on libtorch_python library load to update
the 'maybe' function to use the real python stacktrace getter
Test Plan:
OSS build tests:
- test_ptltc by itself works
- LTC_SAVE_TENSORS_FILE=log test_ptltc works, and log contains
empty stacktrces
- python examply.py by itself works
- LTC_SAVE_TENSORS_FILE=log test_ptltc works, and log contains
real stacktraces
fbcode build: rely on CI to run test/lazy
Reviewed By: desertfire
Differential Revision: D34115046
fbshipit-source-id: 8d6222963c146da36b3c1b5ff8a638bbc3f1442e
(cherry picked from commit 3717688ade)
22 lines
659 B
C++
22 lines
659 B
C++
#include <torch/csrc/lazy/python/init.h>
|
|
|
|
#include <torch/csrc/lazy/core/debug_util.h>
|
|
#include <torch/csrc/lazy/python/python_util.h>
|
|
|
|
namespace torch {
|
|
namespace lazy {
|
|
|
|
void initLazyBindings(PyObject* /* module */){
|
|
#ifndef USE_DEPLOY
|
|
// When libtorch_python is loaded, we register the python frame getter
|
|
// otherwise, debug util simply omits python frames
|
|
// TODO(whc) can we make this work inside torch deploy interpreter?
|
|
// it doesn't work as-is, possibly becuase GetPythonFrames resolves to external
|
|
// cpython rather than embedded cpython
|
|
GetPythonFramesFunction() = GetPythonFrames;
|
|
#endif
|
|
}
|
|
|
|
} // namespace lazy
|
|
} // namespace torch
|