* Anthony Liguori (anth...@codemonkey.ws) wrote: > Andrea Arcangeli wrote: >> On Fri, Dec 12, 2008 at 12:54:21PM +0100, Jens Axboe wrote: >>> I added CLONE_IO some time ago to avoid that, so it's perfectly possible >>> to share cfq io contexts with threads or processes even in userspace!
Yes, I've used it in this context and it solves the throughput issue we see with cfq idling the independent threads. Not doing vectored io all the way through only exacerbates the problem, of course. >> It's available in recent kernels I see! so the fix is easy. Only >> problem is how to pass CLONE_IO to pthread_create... We'll have to >> make a linux-only change and call clone by hand under some #ifdef >> CLONE_IO. > > I have no problem with this and I believe Chris was going to attempt an > implementation. I have an implementation that is just raw clone and futex (so not portable from linux/x86). I'll post it later today. We'll need glibc changes down the road to properly handle. thanks, -chris -- To unsubscribe from this list: send the line "unsubscribe kvm" in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html