Hello everyone, On Mon, Jul 02, 2018 at 02:10:54PM +0100, Stefan Hajnoczi wrote: > Marcelo, Andrea, Paolo: There was a more complex local migration > approach in 2013 with fd passing and vmsplice. They specifically > avoided the approach proposed in this patch, but I don't remember why. > > The closest to an explanation I've found is this message from Marcelo: > > Another possibility is to use memory that is not anonymous for guest > RAM, such as hugetlbfs or tmpfs. > > IIRC ksm and thp have limitations wrt tmpfs. > > https://www.spinics.net/lists/linux-mm/msg67437.html > > Have the limitations been been solved since then?
tmpfs supports THP upstream nawadays, but it's not the default. About KSM on non-anonymous memory it's not happening any time soon, I'm not aware of anybody working on it and it's major change not easily staying contained within KSM so it would take time. There have been multiple requests for this feature (but usually for bare metal containers). The higher priority items for KSM are in terms of xxhash and/or accellerated crc32 and perhaps multithreading the scanner to use more cores. Those are fully self contained issues and they won't make the rest of the kernel more complex. Such kind of setup simply brings some of the limitations that vhost-user/DPDK also has. Thanks, Andrea
