Wrong memory orderings inside the RwLock implementation allow for two writers to acquire the lock at the same time. The drop implementation used Ordering::Relaxed, which allows the compiler or CPU to reorder a mutable access on the locked data after the lock has been yielded.

Only users of the RwLock implementation are affected. Users of Once (including users of lazy_static with the spin_no_std feature enabled) are NOT affected.

On strongly ordered CPU architectures like x86, the only real way that this would lead to a memory corruption is if the compiler reorders an access after the lock is yielded, which is possible but in practice unlikely. It is a more serious issue on weakly ordered architectures such as ARM which, except in the presence of certain instructions, allow the hardware to decide which accesses are seen at what times. Therefore on an ARM system it is likely that using the wrong memory ordering would result in a memory corruption, even if the compiler itself doesn't reorder the memory accesses in a buggy way.

The flaw was corrected by https://github.com/mvdnes/spin-rs/pull/66.

A bug in the SmallVec::insert_many method caused it to allocate a buffer that was smaller than needed. It then wrote past the end of the buffer, causing a buffer overflow and memory corruption on the heap.

This bug was only triggered if the iterator passed to insert_many yielded more items than the lower bound returned from its size_hint method.

The flaw was corrected in smallvec 0.6.14 and 1.6.1, by ensuring that additional space is always reserved for each item inserted. The fix also simplified the implementation of insert_many to use less unsafe code, so it is easier to verify its correctness.

Thank you to Yechan Bae (@Qwaz) and the Rust group at Georgia Tech’s SSLab for finding and reporting this bug.