As Daniel pointed out in: https://postgr.es/m/FB948276-7B32-4B77-83E6-D00167F8EEB4@yesql.se the pg_atomic_flag fallback implementation is broken. That has gone unnoticed because the fallback implementation wasn't testable until now: - /* --- - * Can't run the test under the semaphore emulation, it doesn't handle - * checking two edge cases well: - * - pg_atomic_unlocked_test_flag() always returns true - * - locking a already locked flag blocks - * it seems better to not test the semaphore fallback here, than weaken - * the checks for the other cases. The semaphore code will be the same - * everywhere, whereas the efficient implementations wont. - * --- - */ and flags weren't used until recently.
The attached fixes the bug and removes the edge-cases by storing a value separate from the semaphore. I should have done that from the start. This is an ABI break, but given the fallback didn't work at all, I don't think that's a problem for backporting.