On Mon, Feb 17, 2014 at 8:56 AM, Michael S. Tsirkin <m...@redhat.com> wrote:
> On Mon, Feb 17, 2014 at 12:42:45AM +0100, Paolo Bonzini wrote: > > Il 15/02/2014 19:10, Michael Tokarev ha scritto: > > > 13 февраля 2014 г. 16:03:12 GMT+04:00, Antonios Motakis < > a.mota...@virtualopensystems.com> пишет: > > >> Extend -mem-path with additional properties: > > >> > > >> - prealloc=on|off - default off, same as -mem-prealloc > > >> - share=on|off - default off, memory is mmapped with MAP_SHARED > > > > > > Maybe we should combine -m and -mem-path options together to form > something more sane? > > > > It's on the way: it would be something like > > > > -object > mem-file,size=1024M,path=/path/to/foo,share=on,prealloc=on,id=mem \ > > -numa node,memdev=mem > > > > using the same host/guest split model that is already in use > > in many other places. Not 2.0 material though. > > > > Paolo > > Hmm in that case, let's not add prealloc as a property here. > Stick to existing flag for that, this way we don't need > to support 3 ways to do this. > We'll remove the prealloc property then for the next version. Otherwise, I wonder if the long term memdev discussion can impose additional constraints on our current implementation. It seems to me that this shouldn't be the case, right? -- Antonios Motakis Virtual Open Systems
