When investigating the sha2-0.10 transition being stuck, I noticed the test
seems unrelated to sha2, but is instead related to filesystem permissions. I'm
not sure why this was only flagged on riscv64, and how the test has passed in
the past.

I also noticed there have been new uploads to tame-index in the meantime, so
this problem may have been solved already, but the patches I looked into seemed
unrelated to this specific test.

https://web.archive.org/web/20260525194927/https://ci.debian.net/data/autopkgtest/testing/riscv64/r/rust-tame-index/71210417/log.gz
--- 8< ---

816s thread 'builds_local_registry' (14929) panicked at tests/local.rs:22:25:
816s failed to gather metadata: CargoMetadata { stderr: "error: failed to write
/usr/share/cargo/registry/tame-index-0.26.2/Cargo.lock\n\nCaused by:\n  failed
to open: /usr/share/cargo/registry/tame-index-0.26.2/Cargo.lock\n\nCaused by:\n
Permission denied (os error 13)\n" }
816s stack backtrace:
816s    0: __rustc::rust_begin_unwind
816s              at /usr/src/rustc-1.94.1/library/std/src/panicking.rs:689:5
816s    1: core::panicking::panic_fmt
816s              at /usr/src/rustc-1.94.1/library/core/src/panicking.rs:80:14
816s    2: core::result::unwrap_failed
816s              at /usr/src/rustc-1.94.1/library/core/src/result.rs:1867:5
816s    3: core::result::Result<T,E>::expect
816s              at /usr/src/rustc-1.94.1/library/core/src/result.rs:1185:23
816s    4: local::builds_local_registry
816s              at ./tests/local.rs:22:25
816s    5: local::builds_local_registry::{{closure}}
816s              at ./tests/local.rs:10:27
816s    6: core::ops::function::FnOnce::call_once
816s              at /usr/src/rustc-1.94.1/library/core/src/ops/function.rs:250:5
816s    7: core::ops::function::FnOnce::call_once
816s              at /usr/src/rustc-1.94.1/library/core/src/ops/function.rs:250:5
816s note: Some details are omitted, run with `RUST_BACKTRACE=full` for a
verbose backtrace.
816s
816s
816s failures:
816s     builds_local_registry
816s
816s test result: FAILED. 2 passed; 1 failed; 0 ignored; 0 measured; 0 filtered
out; finished in 1.27s
816s
816s error: test failed, to rerun pass `--test local`

--- >8 ---

Severity 1137611 important
thanks
I was also confused and had to re-read the test logs several times before I
figured out what was going on.

The "builds_local_registry" failure is not the cause of the test failures on
riscv64. That test only runs in the all-features test which is already
marked
as "flaky".

The actual failure seems to be a bench that tries to download stuff from the
Internet and fails to do so.
I disabled said bench in 0.26.2-2