Well, I've been reviewing this change and I'm not sure how it does improve our situation. It's hard to commit a change without any sort of explanation on why it improve things.
The poll backend has its own synchronization on its modify callback. The other two, kqueue and epool, simply do a syscall to queue the new event. Maybe what needs to be done is to have those two use the same principle and use a pair of pipes so only a single thread change the kernel side of things. On Fri, Aug 17, 2012 at 1:04 PM, Greg Young <gregoryyou...@gmail.com> wrote: > Thanks guys. There is a matching push on list object pool I believe the > two are needed together. We have a few others as well... > > > On Friday, August 17, 2012, Rodrigo Kumpera wrote: > >> Ok, I'll push to 2.10/2.12 >> >> On Fri, Aug 17, 2012 at 1:19 AM, Gonzalo Paniagua Javier < >> gonzalo.m...@gmail.com> wrote: >> >>> On Thu, Aug 9, 2012 at 11:57 AM, Greg Young <gregoryyou...@gmail.com> >>> wrote: >>> > This would be a good one to get into trunk >>> > >>> https://github.com/ysw/mono-socket-problem/blob/master/Patches/cb_fix.patch >>> >>> Rodrigo, I did the patch above and it seems to make things better for >>> Greg. I can't push it right now, can you? >>> >>> -Gonzalo >>> >> >> > > -- > Le doute n'est pas une condition agréable, mais la certitude est absurde. >
_______________________________________________ Mono-devel-list mailing list Mono-devel-list@lists.ximian.com http://lists.ximian.com/mailman/listinfo/mono-devel-list