hrm... this doesn't seem fun...

looking at the stack that ::mutex shows as owning a lock, it's stack shows mutex_adaptive_tryenter() (at the offset where it's doing the jne check to see if it succeeded swapping the thread address.. e.g. +0x10) followed immediately by _interrupt()

hrm.. actually nevermind.. the caller is doing a recursive enter check (just didn't see it in the stack because of no frame pointer)

which is what i was concerned about..