On Sun, Nov 29, 2009 at 3:32 PM, Jan Kiszka <jan.kis...@web.de> wrote: > Glauber Costa wrote: >> Hi guys, >> >> This is an early version of smp support in kvm that kinda works. >> It has some known problems that I am still tracking. For example, >> it does not reset very well. Also, initialization is a bit slow, >> probably because of the number of remote ioctl calls involved. >> But I believe Jan's patch to do a single ioctl can make it better. > > That approach has been buried, we decided to continue extending via new > ioctls. > > But why do you need that many remote calls during init? Aren't the > initial ioctls issued by the vcpu threads themselves?
Yes, but system_reset is called from the io-thread, and it fires a lot of ioctls. However, I have an already working version that does a much better job than that. Will post today with details -- Glauber Costa. "Free as in Freedom" http://glommer.net "The less confident you are, the more serious you have to act."
