Dan, is tmem supposed to be usable with localhost migration? Was it ever tested that way? Thanks, Jan _______________________________________________ Xen-devel mailing list Xen-devel@lists.xensource.com http://lists.xensource.com/xen-devel
Yes, tmem is compatible with migration (and save/restore) and was tested with localhost migration as well as cross-host migration. BUT, I don''t know if I tested it when there is no free (but only freeable) memory, which might, as suggested in the separate thread, fail due to shadow memory requiring order=2 allocations. Or are you seeing a different problem? Thanks, Dan> -----Original Message----- > From: Jan Beulich [mailto:JBeulich@novell.com] > Sent: Monday, February 08, 2010 8:49 AM > To: Dan Magenheimer > Cc: xen-devel@lists.xensource.com > Subject: tmem and localhost migration > > Dan, > > is tmem supposed to be usable with localhost migration? Was it ever > tested that way? > > Thanks, Jan >_______________________________________________ Xen-devel mailing list Xen-devel@lists.xensource.com http://lists.xensource.com/xen-devel
>>> Dan Magenheimer <dan.magenheimer@oracle.com> 08.02.10 18:24 >>> >Yes, tmem is compatible with migration (and save/restore) >and was tested with localhost migration as well as >cross-host migration. BUT, I don''t know if I tested >it when there is no free (but only freeable) memory, >which might, as suggested in the separate thread, >fail due to shadow memory requiring order=2 allocations. > >Or are you seeing a different problem?Unfortunately we seem to: For one there is an apparent deadlock (DomU having read-locked tmem_lock thus preventing Dom0 to obtain a write lock thereof, DomU spinning to acquire a read lock in obj_find()). Also, I think numerous domain references get leaked, but I think I have a patch ready to address that (will send out later today, after having it at least build-validated). Jan _______________________________________________ Xen-devel mailing list Xen-devel@lists.xensource.com http://lists.xensource.com/xen-devel