From: Michael P. <mic...@gm...> - 2010-01-24 21:42:17
|
Pete Batard wrote: >> The safe way to fix this is probably to reintroduce the >> handle_timeouts_locked, as it existed prior to 102. It'll slightly >> duplicate code, but it should restore the behaviour for all the >> platforms that directly called handle_timeouts_locked exactly. >> >> Unless someone has a better idea, I'll include this patch in one of my >> next commits then. This is good for now. Is it possible that Windows has something similar to timerfd that someone may want to use in the future, such as CreateWaitableTimer()? If so, a partial revert is only a temporary solution, since someone may want to put that functionality into the Windows version. I'll look at it more when I've finished studying the source. I would presume that if it were possible to just lock the mutex before rearming the timerfd, the code would have already been simplified that way, so I need to read more. Thanks, Michael |