On Tue, Apr 17, 2018 at 13:18:02 +0100, Daniel Berrange wrote: > On Tue, Apr 17, 2018 at 01:16:42PM +0200, Michal Privoznik wrote: > > https://bugzilla.redhat.com/show_bug.cgi?id=1480668 > > > > The cases when we cannot enable this optimization are: > > 1) nvdimms > > 2) if memAccess='shared' > > > > Otherwise it is safe to enable it. > > This is basically saying that if memAccess==private, then no person/app cares > about the contents of the memory-backend-file storage after QEMU has exited, > because (implicitly) whatever data was in that storage is not going to be used > again. While I accept that's going to be a common case, I'm wary of saying that > is a 100% safe/valid assumption for every user. > > I tend think it should require an explicit opt-in to turn on this behaviour Well, we used that behaviour prior to storing the backing files in a "predictible" path as the file was unlinked right after creating it (so that some memory management software can do some magic behind our backs with that) so I don't think anybody would depend on the described behaviour.
Attachment:
signature.asc
Description: PGP signature
-- libvir-list mailing list libvir-list@xxxxxxxxxx https://www.redhat.com/mailman/listinfo/libvir-list
