io_uring fs operations can panic when runtime IO is disabled

[fallintoplace]

Description

With tokio_unstable, the io-uring feature, and Linux fs support enabled, several tokio::fs APIs probe the runtime io_uring driver before falling back to the blocking filesystem path. Those probes call the panicking driver().io() accessor.

If a Tokio runtime exists but was built without IO enabled, these filesystem operations can panic before reaching the existing spawn_blocking fallback.

Affected paths

Examples include:

• tokio::fs::read
• tokio::fs::write
• tokio::fs::OpenOptions::open
• tokio::fs::File reads from File::from_std

Example

let rt = tokio::runtime::Builder::new_current_thread()
    .build()
    .unwrap();

rt.block_on(async {
    let _ = tokio::fs::read("foo.txt").await;
});

In an io_uring-enabled build, this can panic with the IO-disabled runtime message before falling back to the blocking filesystem implementation.

Expected behavior

If the runtime has no IO driver, the optional io_uring path should be skipped and filesystem operations should use the existing blocking fallback. This matches the existing fallback behavior when io_uring is unavailable or unsupported.

Notes

This is limited to Linux tokio_unstable + io-uring builds, so it does not affect normal stable Tokio builds.