ITADN

Bug: MIRIFLAGS never passed to miri

#3160OpenOakchris1955 创建于 2026-03-15
bug
O
Oakchris1955commented
### Description of the issue **Description:** `MIRIFLAGS` environment variable is apparently never passed to `miri` **Reproduction** Run a heavy test with `cargo +nightly miri nextest run`, then run it normally with `cargo +nightly miri test` and pass `MIRIFLAGS="-Zmir-opt-level=2"`. The nextest version will take noticeably longer than the ordinary one, indicating that nextest doesn't pass `MIRIFLAGS` to `miri` ### Expected outcome nextest should pass `MIRIFLAGS` to `miri`. For the example above, both tests should take roughly thr ame time to complete. ### Actual result When nextest run my tests, it took way longer (~5x times more) to complete than with plain `miri` ### Nextest version ```text cargo-nextest 0.9.129 (1358a2296 2026-02-22) release: 0.9.129 commit-hash: 1358a22964df2595d30d830ad669b37093e4a732 commit-date: 2026-02-22 host: x86_64-unknown-linux-gnu ``` ### Additional context In my setup, I run cargo with the `--target=powerpc-unknown-linux-gnu` parameter. However, I don't expect it to impact this use case.
7 条评论