Re:Should you be scared of Unix signals?
by jasonzemos on Monday June 13, 2016 @11:19PM
(#11899523)
Sure, and then the handler only sees one integer as updated and the other integer will be updated after the handler. The lock gets silently broken unfortunately, but there's probably a useful reason for why this is. It could just deadlock instead.
[ Parent
| on HN ]
Deadlocking would be the ideal outcome. Deadlocks are easy to debug. Silent concurrent memory corruptions not so much.