Finalizer is not thread safe and may end up breaking some modules about pam HOT 2 CLOSED

Now that I think more on this, I wonder if the problem you originally described is really solved. If the goroutine that calls pam.Start suspends before Transaction.End is called it's possible (even likely) that it may be scheduled on a different OS thread when it continues. It looks like runtime.LockOSThread might help, but I don't think it can be used from within this library. The caller would need to arrrange to lock/unlock the thread at the start/end of the PAM transaction. What do you think?

from pam.

I think it has, at least in all the tests I've been doing.

Before we could both end up returning an invalid state or being finalized in a thread different than the calling one.

Indeed things may not go as expected depending how the library is used, but at least we never saw again the crashes at finalization time.

from pam.