pytorch / pytorch/vision

v0.8.X Build failure due to binding type error

Open
#3,216 1 comment 0 reactions 0 assignees View on GitHub

Nobody has claimed this yet.

topic: build
Dominant language
Python
Stars
17.9k
Forks
7.3k
Avg merge
1d 15h
Merged PRs (30d)
13

Description

🐛 Bug

Build fails due to a binding type error on v0.8.X.

To Reproduce

Steps to reproduce the behavior:

  1. Clone the repo.

    $ git clone git@github.com:pytorch/vision.git
    
  2. Checkout any 0.8.X version.

    $ git checkout v0.8.2
    
  3. Build and install from source (my setup requires a home directory installation).

    $ python setup.py install --home "${HOME}"
    

Build failure error message:

[REDACTED_INSTALL_DIR]/torch/include/ATen/core/boxing/impl/make_boxed_from_unboxed_functor.h:242:170: error: cannot bind non-const lvalue reference of type ‘at::Tensor&’ to an rvalue of type ‘at::Tensor’
  242 | move_cv_t<std::remove_reference_t<guts::typelist::element_t<ivalue_arg_indices, IValueArgTypes>>>, AllowDeprecatedTypes>::call(
      | ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~^

  243 | alue_arg_indices, num_ivalue_args))
      | ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

  244 |
      |

error: command 'gcc' failed with exit status 1

Expected behavior

Build success

Environment

PyTorch version: 1.6.0a0+b31f58d
Is debug build: False
CUDA used to build PyTorch: False
ROCM used to build PyTorch: N/A

OS: Manjaro ARM Linux (aarch64)
GCC version: (GCC) 10.2.0
Clang version: N/A
CMake version: version 3.19.1

Python version: 3.8 (64-bit runtime)
Is CUDA available: False
CUDA runtime version: No CUDA
GPU models and configuration: No CUDA
Nvidia driver version: No CUDA
cuDNN version: No CUDA
HIP runtime version: N/A
MIOpen runtime version: N/A

Versions of relevant libraries:
[pip3] numpy==1.19.4
[pip3] torch==1.6.0a0+b31f58d
[conda] N/A

Additional context

  • At first, I tried to install with the master branch, but I faced #3210. Thanks to @qyb, I was able to work around this by checking out a specific version (git checkout v0.8.2).
  • I am able to successfully build and install with v0.7.0 (git checkout v0.7.0; python setup.py install --home "${HOME}").

Contributor guide

Open the contributing guide

First steps

  1. Read the whole issue, then the project's contributing guide.
  2. Comment on the issue to say you are picking it up — it saves two people doing the same work.
  3. Fork the repository and make your change on a branch.
  4. Open a pull request that references the issue number.

Research direction

Reproduce the v0.8.2 source build with python setup.py install --home "${HOME}" on the reported ARM environment, then inspect the binding error in make_boxed_from_unboxed_functor.h. Done means the v0.8.X build completes successfully; compare with the working v0.7.0 build and the related #3210 context before changing anything.

Written by the indexing model from the issue text.

Assessment

Tech stack
cpp, python
Domain
build-system, computer-vision
Issue type
Bug
Difficulty
4/5
Estimated time
3-5 days
Activity status
Stale
Clarity
Mostly clear
Newbie friendliness
35/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.