On Tue, Feb 04, 2014 at 05:26:31PM +0000, Daniel P. Berrange wrote: > On Tue, Feb 04, 2014 at 12:22:00PM -0500, Marcelo Tosatti wrote: > > > > You're OK with StrictHugePageSize=N element ? > > I'm not entirely sure what that would do ? Is that for when people want > to have a specific size of hugepages ? If so, then I guess we might > actually need some kind of granularity here. Yes, they want to enforce a specific hugepage size (so as to fail guest initialization if guest RAM cannot be backed by the particular size). > eg if the user has a 1200 MB guest, and they want to use 1 GB huge > pages, then there's a choice between 2 * 1GB huge pages, or 1 x 1GB > and n * 4M huge pages, or an error requiring that their guest RAM is > a multiple of requested huge page size. ATM QEMU does not support different huge page sizes, so the only option would be 2 * 1GB pages. > So perhaps we do need some "policy" attribute on the <hugepages/> > element to indicate desired behaviour here. What about the following new element under <hugepages/> ? enforce_hugepage_size=integer -- libvir-list mailing list libvir-list@xxxxxxxxxx https://www.redhat.com/mailman/listinfo/libvir-list