pytorch/torch/csrc/jit/runtime/interpreter
Riham Selim b963d96bad [Torchscript] Add a flag to use mangled names instead of demangled (#148906)
Summary: Optionally keep mangled names when expanding torchscript stacks

Test Plan:
```
buck2 build mode/opt //scripts/rihams/LearnPyTorch:torch_script_generate --show-full-output

/data/users/rihams/fbsource/buck-out/v2/gen/fbcode/0bd9d136228ad8a7/scripts/rihams/LearnPyTorch/__torch_script_generate__/torch_script_generate.par

buck2 build mode/opt //scripts/rihams/LearnPyTorch:torch_script_execute --show-full-output
```

- With `--torch_jit_expanded_stacks_mangled` Flag:

/data/users/rihams/fbsource/buck-out/v2/gen/fbcode/ef35e45045e8164c/scripts/rihams/LearnPyTorch/__torch_script_execute__/torch_script_execute fbcode/model.pt  --torch_jit_expanded_stacks_mangled --torch_jit_enable_expanded_stacks

https://fburl.com/scuba/strobelight_function_tracer/8die4rvm

{F1975933247}

Without Flag:

/data/users/rihams/fbsource/buck-out/v2/gen/fbcode/ef35e45045e8164c/scripts/rihams/LearnPyTorch/__torch_script_execute__/torch_script_execute ./model.pt   --torch_jit_enable_expanded_stacks

https://fburl.com/scuba/strobelight_function_tracer/x3nladpf

 {F1975933268}

Reviewed By: bbus

Differential Revision: D70905872

Pull Request resolved: https://github.com/pytorch/pytorch/pull/148906
Approved by: https://github.com/zdevito
2025-03-19 07:53:02 +00:00
..
can_emit_inline.h
code_impl.h [Torchscript] Add a flag to use mangled names instead of demangled (#148906) 2025-03-19 07:53:02 +00:00
frame.cpp
frame.h
preprocess_graph.cpp
preprocess_graph.h