On Tue, Dec 9, 2014 at 1:43 PM, Ryota Ozaki <[email protected]> wrote: > On Tue, Dec 9, 2014 at 1:17 PM, Thor Lancelot Simon <[email protected]> wrote: >> On Tue, Dec 09, 2014 at 12:52:35PM +0900, Ryota Ozaki wrote: >>> On Fri, Dec 5, 2014 at 10:02 AM, Ryota Ozaki <[email protected]> wrote: >>> > On Fri, Dec 5, 2014 at 7:07 AM, Mindaugas Rasiukevicius >>> > <[email protected]> wrote: >>> >> Ryota Ozaki <[email protected]> wrote: >>> >>> Can we use pserialize for readers in hardware interrupt? >>> >>> If we use splhigh instead of splsoftserial (pserialize_read_enter) >>> >>> for critical sections of the readers, we can guarantee that >>> >>> the readers don't touch a removed object anymore after >>> >>> pserialize_perform. It this correct? >>> >> >>> >> It is correct, however we are trying to do less work in the hardware >>> >> interrupt handlers by moving it to the software interrupts. >>> > >>> > Thanks. I think we need a gradual migration plan to the direction. >>> >>> Anyway I wrote a patch of a softint-based interrupt handler for if_vioif: >>> http://www.netbsd.org/~ozaki-r/vioif-softint-intr.diff >>> >>> It works well without problems, but I see performance degradation on >>> L2 forwarding. So the feature is disabled by default. >> >> Can you try increasing HZ? > > Thank you for the suggestion. Which HZ is good for the purpose? > 1000 was not good for my environment (KVM for now) and I'm trying > other HZ (500, 200, etc.).
I couldn't get good results on this approach... ozaki-r
