https://gcc.gnu.org/bugzilla/show_bug.cgi?id=69204
--- Comment #6 from Dmitry Vyukov <dvyukov at google dot com> --- The problem is that gcc does not instrument atomic operations for some reason. Here is disasm of future::set_result: Dump of assembler code for function std::__future_base::_State_baseV2::_M_set_result(std::function<std::unique_ptr<std::__future_base::_Result_base, std::__future_base::_Result_base::_Deleter> ()>, bool): ... 0x000000000048bebd <+301>: mov -0x40(%rbp),%rax 0x000000000048bec1 <+305>: mov -0x44(%rbp),%edx 0x000000000048bec4 <+308>: xchg %edx,(%rax) 0x000000000048bec6 <+310>: mov %edx,%eax 0x000000000048bec8 <+312>: shr $0x1f,%eax 0x000000000048becb <+315>: test %al,%al 0x000000000048becd <+317>: je 0x48bef5 <std::__future_base::_State_baseV2::_M_set_result(std::function<std::unique_ptr<std::__future_base::_Result_base, std::__future_base::_Result_base::_Deleter> ()>, bool)+357> 0x000000000048becf <+319>: mov -0x38(%rbp),%rax 0x000000000048bed3 <+323>: mov %rax,%rdi 0x000000000048bed6 <+326>: callq 0x407550 <_ZNSt28__atomic_futex_unsigned_base19_M_futex_notify_allEPj@plt> ... 0x000000000048bf1c <+396>: retq That xchg is supposed to be std::atomic<unsigned>::store(memory_order_release) that marks the future as ready. I have no idea what has regressed.
