frozenleaves/pytorch
1
0
Fork 0
mirror of https://github.com/pytorch/pytorch.git synced 2025-10-20 21:14:14 +08:00
Code Issues Packages Projects Releases Wiki Activity

Debug with gdb

Justin Chu
2022-07-06 14:20:45 -07:00
parent 361a0d3c02
commit a16a640691
1 changed files with 74 additions and 29 deletions
Download Patch File Download Diff File Expand all files Collapse all files

103
PyTorch-ONNX-exporter.md

@ -6,22 +6,25 @@ Documentation for developing the PyTorch-ONNX exporter (`torch.onnx`). For an in
<!-- TOC generated with https://github.com/ekalinin/github-markdown-toc --> <!-- TOC generated with https://github.com/ekalinin/github-markdown-toc -->
* [Development process](#development-process) - [Table of Contents](#table-of-contents)
* [Environment setup](#environment-setup) - [Development process](#development-process)
* [Fork PyTorch](#fork-pytorch) - [Environment setup](#environment-setup)
* [Build PyTorch](#build-pytorch) - [Fork PyTorch](#fork-pytorch)
* [Optional build tips](#optional-build-tips) - [Build PyTorch](#build-pytorch)
* [Install additional dependencies](#install-additional-dependencies) - [Optional build tips](#optional-build-tips)
* [ONNX and ONNX Runtime](#onnx-and-onnx-runtime) - [Install additional dependencies](#install-additional-dependencies)
* [TorchVision](#torchvision) - [ONNX and ONNX Runtime](#onnx-and-onnx-runtime)
* [Sanity check](#sanity-check) - [TorchVision](#torchvision)
* [VS Code](#vs-code) - [Sanity check](#sanity-check)
* [Pull requests](#pull-requests) - [VS Code](#vs-code)
* [Tests](#tests) - [Recommended settings and extensions](#recommended-settings-and-extensions)
* [Links](#links) - [Debugging with `gdb`](#debugging-with-gdb)
* [Relevant parts of PyTorch repo](#relevant-parts-of-pytorch-repo) - [Pull requests](#pull-requests)
* [Features](#features) - [Tests](#tests)
* [Quantized model export](#quantized-model-export) - [Links](#links)
- [Relevant parts of PyTorch repo](#relevant-parts-of-pytorch-repo)
- [Features](#features)
- [Quantized model export](#quantized-model-export)
# Development process # Development process
@ -135,6 +138,8 @@ If the second command succeeds, then probably python is finding a PyTorch that w
### VS Code ### VS Code
#### Recommended settings and extensions
You can place this recommended `settings.json` under `.vscode/` You can place this recommended `settings.json` under `.vscode/`
```jsonc ```jsonc
@ -207,17 +212,57 @@ If you use error lens, I recommend the following settings
} }
``` ```
#### Debugging with `gdb`
You can set up VS Code to run `gdb` and set breakpoints when debugging c++ code. In `launch.json` add the configuration
```jsonc
// ...
"configurations": [
{
"name": "(gdb) Launch",
"type": "cppdbg",
"request": "launch",
"program": "<path to python bin>",
"args": [
"-m",
"pytest",
"<test file and test name in pytest format>"
],
"stopAtEntry": false,
"cwd": "path/to/repo/of/pytorch",
"environment": [],
"externalConsole": false,
"MIMode": "gdb",
"setupCommands": [
{
"description": "Enable pretty-printing for gdb",
"text": "-enable-pretty-printing",
"ignoreFailures": true
},
{
"description": "Set Disassembly Flavor to Intel",
"text": "-gdb-set disassembly-flavor intel",
"ignoreFailures": true
}
]
},
]
```
You can then set breakpoints in the c++ source and run the debugger in VS Code.
## Pull requests ## Pull requests
PRs should be opened directly against master. PRs can be directly merged into master as long as it satisfies the [ONNX merge rule](https://github.com/pytorch/pytorch/blob/master/.github/merge_rules.json#L3): PRs should be opened directly against master. PRs can be directly merged into master as long as it satisfies the [ONNX merge rule](https://github.com/pytorch/pytorch/blob/master/.github/merge_rules.json#L3):
- Approved by one of torch.onnx developers listed in `approved_by` section. * Approved by one of torch.onnx developers listed in `approved_by` section.
- All modified files fall under the `patterns` section. * All modified files fall under the `patterns` section.
Pay special attention to the following GitHub checks: Pay special attention to the following GitHub checks:
- Has "onnx" in the name, which runs ONNX related tests. * Has "onnx" in the name, which runs ONNX related tests.
- Has "Lint" in the name, which does code format checks. * Has "Lint" in the name, which does code format checks.
Regarding other failing checks: if you are certain the failure is unrelated to your change, try rebasing on master. Often these failures are caused by a branch being out of sync with master. Regarding other failing checks: if you are certain the failure is unrelated to your change, try rebasing on master. Often these failures are caused by a branch being out of sync with master.
You can ignore the failing check if it is a regression in master. This can be verified by checking if master is also failing from [CI HUD](https://hud.pytorch.org/ci/pytorch/pytorch/master). You can ignore the failing check if it is a regression in master. This can be verified by checking if master is also failing from [CI HUD](https://hud.pytorch.org/ci/pytorch/pytorch/master).
@ -242,9 +287,9 @@ Most relevant tests are in [test/onnx/](https://github.com/pytorch/pytorch/tree/
The most used test file is [test_pytorch_onnx_onnxruntime.py](https://github.com/pytorch/pytorch/blob/onnx_ms_1/test/onnx/test_pytorch_onnx_onnxruntime.py). The tests in this file generally: The most used test file is [test_pytorch_onnx_onnxruntime.py](https://github.com/pytorch/pytorch/blob/onnx_ms_1/test/onnx/test_pytorch_onnx_onnxruntime.py). The tests in this file generally:
- Define a subclass of `torch.nn.Module`. * Define a subclass of `torch.nn.Module`.
- Define some inputs. * Define some inputs.
- Call `self.run_test()` with the instantiated module and inputs. * Call `self.run_test()` with the instantiated module and inputs.
`run_test()` converts the module to ONNX and compares the output between PyTorch and ONNX Runtime. `run_test()` converts the module to ONNX and compares the output between PyTorch and ONNX Runtime.
@ -267,15 +312,15 @@ python -m pytest -n auto --cov --cov-report "xml:test/coverage.xml" test/onnx/te
# Links # Links
- [User-facing docs](https://pytorch.org/docs/master/onnx.html). * [User-facing docs](https://pytorch.org/docs/master/onnx.html).
## Relevant parts of PyTorch repo ## Relevant parts of PyTorch repo
- User-facing doc: [docs/source/onnx.rst](https://github.com/pytorch/pytorch/blob/onnx_ms_1/docs/source/onnx.rst) * User-facing doc: [docs/source/onnx.rst](https://github.com/pytorch/pytorch/blob/onnx_ms_1/docs/source/onnx.rst)
- Python tests: [test/onnx/](https://github.com/pytorch/pytorch/tree/onnx_ms_1/test/onnx) * Python tests: [test/onnx/](https://github.com/pytorch/pytorch/tree/onnx_ms_1/test/onnx)
- More Python tests: [test/jit/test_onnx_export.py](https://github.com/pytorch/pytorch/tree/onnx_ms_1/test/jit/test_onnx_export.py) * More Python tests: [test/jit/test_onnx_export.py](https://github.com/pytorch/pytorch/tree/onnx_ms_1/test/jit/test_onnx_export.py)
- Python code: [torch/onnx/](https://github.com/pytorch/pytorch/tree/onnx_ms_1/torch/onnx) * Python code: [torch/onnx/](https://github.com/pytorch/pytorch/tree/onnx_ms_1/torch/onnx)
- C++ code: [torch/csrc/jit/passes/onnx/](https://github.com/pytorch/pytorch/tree/onnx_ms_1/torch/csrc/jit/passes/onnx) * C++ code: [torch/csrc/jit/passes/onnx/](https://github.com/pytorch/pytorch/tree/onnx_ms_1/torch/csrc/jit/passes/onnx)
# Features # Features