https://gcc.gnu.org/bugzilla/show_bug.cgi?id=98737
--- Comment #2 from Uroš Bizjak <ubizjak at gmail dot com> ---
This can be optimized with peephole2, we already have similar case in sync.md:
;; This peephole2 and following insn optimize
;; __sync_fetch_and_add (x, -N) == N into just lock {add,sub,inc,dec}
;; followed by testing of flags instead of lock xadd and comparisons.
