`vecadd` example: couldn't load codegen backend, undefined symbol: _ZN4llvm23EnableABIBreakingChecksE

[hduoc2003]

✅ Solution: set LLVM_LINK_SHARED=1 when building like #7

I attempted to run the vecadd example but encountered the following errors:

➜  vecadd git:(main) ✗ LLVM_LINK_STATIC=1 cargo build
   Compiling vecadd v0.1.0 (/root/hduoc2003/code/Rust-CUDA/examples/cuda/vecadd)
error: failed to run custom build command for `vecadd v0.1.0 (/root/hduoc2003/code/Rust-CUDA/examples/cuda/vecadd)`

Caused by:
  process didn't exit successfully: `/root/hduoc2003/code/Rust-CUDA/target/debug/build/vecadd-dbe33157c5bca8c3/build-script-build` (exit status: 101)
  --- stdout
  cargo:rerun-if-changed=build.rs
  cargo:rerun-if-changed=kernels
  cargo:rerun-if-changed=kernels

  --- stderr
  error: failed to run `rustc` to learn about target-specific information

  Caused by:
    process didn't exit successfully: `/root/.rustup/toolchains/nightly-2025-03-02-x86_64-unknown-linux-gnu/bin/rustc - --crate-name ___ --print=file-names -Zcodegen-backend=/root/hduoc2003/code/Rust-CUDA/target/debug/deps/librustc_codegen_nvvm.so '-Zcrate-attr=feature(register_tool)' '-Zcrate-attr=register_tool(nvvm_internal)' -Zcrate-attr=no_std -Zsaturating_float_casts=false '-Cllvm-args=-arch=compute_61 --override-libm' --target nvptx64-nvidia-cuda --crate-type bin --crate-type rlib --crate-type dylib --crate-type cdylib --crate-type staticlib --crate-type proc-macro --print=sysroot --print=split-debuginfo --print=crate-name --print=cfg -Wwarnings` (exit status: 1)
    --- stderr
    error: couldn't load codegen backend /root/hduoc2003/code/Rust-CUDA/target/debug/deps/librustc_codegen_nvvm.so: undefined symbol: _ZN4llvm23EnableABIBreakingChecksE


  thread 'main' panicked at crates/cuda_builder/src/lib.rs:508:10:
  Did not find output file in rustc output
  note: run with `RUST_BACKTRACE=1` environment variable to display a backtrace

---

Context

I'm using Ubuntu 22.04 and installed all required packages as in ubuntu22-cuda12 dockerfile.

---

Environment

$ llvm-config --version
7.0.1

$ nvcc --version
nvcc: NVIDIA (R) Cuda compiler driver
Copyright (c) 2005-2024 NVIDIA Corporation
Built on Tue_Feb_27_16:19:38_PST_2024
Cuda compilation tools, release 12.4, V12.4.99
Build cuda_12.4.r12.4/compiler.33961263_0

---

rust-toolchain.toml

Same as main branch.

[toolchain]
channel = "nightly-2025-03-02"
components = [
    "clippy",
    "llvm-tools-preview",
    "rust-src",
    "rustc-dev",
    "rustfmt",
    "rust-analyzer"
]

[adamcavendish]

From googling, it seems that you're having a debug version of LLVM? We cannot reproduce with the github action pipelines.