On 10/09/2015 15:04, alvise rigo wrote: > Hi Paolo, > > A brief update on this. I have a first implementation of the idea you > proposed, though it's not working really well. The failing rate of SCs > for some reason is very high. > Instead of trying to fix it, I came up with this alternative design: > we still use 8 bits per page and we group the smp_cpus in clusters of > maximum size ceiling(smp_cpus / 8). > Given a byte describing a certain page, its first bit will represents > the vCPUs of index 0, 9, 17, ... and so on for the other bits. > > Note that systems with eight vCPUs or less will not be affected by > this simplified model. > > Now, unless I'm missing some advantages of your proposal except the > less aggressive memory consumption, I would go with this design.
Sounds great. Paolo
