4.0.3 pooled request race can still leak `exit:{normal,_}` to caller

[mashton]

On 4.0.3 we are seeing regular caller crashes caused by normal EXITs bubbling from gen_statem:call paths, instead of receiving an error tuple.

This appears to be related to the pooled connection race handling described here:

hackney/src/hackney_conn.erl

Lines 1448 to 1453 in c93ce0a

	%% Pooled connections used to stop immediately here, but that made
	%% late-arriving {call, From, {request, _}} messages from workers that
	%% raced the pool checkout race a terminating gen_statem — which
	%% surfaces as `exit:{normal, _}` in the caller (issue #836). Stay
	%% alive briefly so those late calls get a proper `{error, {closed, _}}`
	%% reply via handle_common's invalid_state fallback, then stop.

This became visible immediately after upgrading from 4.0.2 -> 4.0.3. As mitigation, we were able to simply disable pooling for the requests in question. This seems to have been a successful mitigation.

Could 4.0.3 timing or dependency changes make this race more visible?

hackney 4.0.3
erlang 27.3

[benoitc]

do you still reproduce it in 4.1 or latest masterr branch?