Invalid cross-device link

[jonathanpwang]

Recently, this workflow is giving errors when used on AWS self-hosted GPU runners:

Example: https://github.com/openvm-org/stark-backend/actions/runs/17200888597/job/48791119926?pr=102

Error snippet:

error: could not rename component file from '/home/runner/.rustup/toolchains/stable-x86_64-unknown-linux-gnu/share/doc/clippy' to '/home/runner/.rustup/tmp/qzwl4905618fqaq1_dir/bk': Invalid cross-device link (os error 18)

Internet/LLM searches say GPU instances are virtualized differently, but the same workflows used to run fine until 2 days ago, so I wonder if something has changed in the workflow to cause this?

I now add

      - run: |
          rustup component remove clippy || true
          rm -rf ~/.rustup/toolchains/stable-x86_64-unknown-linux-gnu || true

prior to

      - uses: dtolnay/rust-toolchain@stable

in all workflows to fix this.

[jonathanpwang]

To clarify, the workflow continues to work on all CPU runners, and the only change necessary to get to fail is to switch to a GPU runner.

[hamedse7en6]

name: test suite on: [push, pull_request]

jobs: test: name: cargo test runs-on: ubuntu-latest steps: - uses: actions/checkout@v5 - uses: dtolnay/rust-toolchain@stable - run: cargo test --all-features