Miri reports Race condition on Collector Test

[luca3s]

Miri currently reports a race condition in the Collector test. I narrowed it down to for _ in 0..100 { collector.collect() } which races with the threads created before that. I have a branch where i changed two Atomic Accesses and miri didn't complain anymore, but i honestly know too little about the linked list used to determine if that really fixes it or just doesn't trigger in that test anymore.

---

I orginally wanted to build T: ?Sized Support for the Library and just got sidetracked. That PR sadly isn't going happen, as AtomicPtr can't deal with T: ?Sized, due to platform limitations and i can't see a way around that. This (as far as i can see) also makes #3 completely impossible, which i hoped would be able to happen soon (with the Rust for Linux push for custom Smart Pointers).

---

Miri Output:

Finished test profile [unoptimized + debuginfo] target(s) in 0.00s
Running unittests src/lib.rs (target/miri/aarch64-apple-darwin/debug/deps/basedrop-43f3ef052c3bee08)

running 1 test
test collector::tests::collector ... error: Undefined Behavior: Data race detected between (1) non-atomic write on thread unnamed-97 and (2) atomic store on thread collector::tests::collector at alloc1585770. (2) just happened here
--> src/collector.rs:281:21
|
281 | (*tail).link.next.store(head, Ordering::Relaxed);
| ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ Data race detected between (1) non-atomic write on thread unnamed-97 and (2) atomic store on thread collector::tests::collector at alloc1585770. (2) just happened here
|
help: and (1) occurred earlier here
--> src/collector.rs:99:9
|
99 | (*node).header.link.next = ManuallyDrop::new(AtomicPtr::new(core::ptr::null_mut()));
| ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
= help: this indicates a bug in the program: it performed an invalid operation, and caused Undefined Behavior
= help: see https://doc.rust-lang.org/nightly/reference/behavior-considered-undefined.html for further information
= note: BACKTRACE (of the first span) on thread collector::tests::collector:
= note: inside collector::Collector::collect_one at src/collector.rs:281:21: 281:69
note: inside collector::Collector::collect
--> src/collector.rs:238:15
|
238 | while self.collect_one() {}
| ^^^^^^^^^^^^^^^^^^
note: inside collector::tests::collector
--> src/collector.rs:391:13
|
391 | collector.collect();
| ^^^^^^^^^^^^^^^^^^^
note: inside closure
--> src/collector.rs:362:19
|
361 | #[test]
| ------- in this procedural macro expansion
362 | fn collector() {
| ^
= note: this error originates in the attribute macro test (in Nightly builds, run with -Z macro-backtrace for more info)

note: some details are omitted, run with MIRIFLAGS=-Zmiri-backtrace=full for a verbose backtrace

error: aborting due to 1 previous error

error: test failed, to rerun pass --lib

Caused by:
process didn't exit successfully: /Users/lucasbaumann/.rustup/toolchains/nightly-aarch64-apple-darwin/bin/cargo-miri runner /Users/lucasbaumann/Documents/GitHub/basedrop/target/miri/aarch64-apple-darwin/debug/deps/basedrop-43f3ef052c3bee08 collector (exit status: 1)
note: test exited abnormally; to see the full output pass --nocapture to the harness.

[micahrj]

Interesting, thanks for reporting this! I'll look into it.