On Sat, 2006-07-15 at 12:39 -0700, Zachary Amsden wrote: > Nivedita Singhvi wrote: > > > > Rusty won't be at the Kernel Summit Session itself, only > > you and Zach have invites, and Chris W, from those working on > > this. Also we figured 30 mins wouldn't be enough. Hence the > > Virtualization API BoF session that I'm proposing - really > > just so that you, Jeremy, Ian, ChrisW, Zach and Rusty get to > > sync up. Rusty should be in Ottawa by Monday night. > > > > Still would be nice to sync up and make sure we're all on the same > page. Rusty, will you have time to meet before the kernel summit > session on Tuesday or should we sync up over email? I arrive Ottawa late Monday night. I can crash the KS party. I'll be jetlagged out of my little mind, but maybe that will help 8) BTW, re: previous discussion of unified entry point for hypervisors. Someone mentioned how difficult it is to share a single entry for native and paravirt, and I agree. But it was also mooted that the current Xen setup is fine for others to use, too. AFAICT, the current CONFIG_XEN patches take over startup_32, so native can no longer use it; I thought we wanted a single kernel which can boot native and paravirt? Hmm, it seems we *could* use startup_32 with %esi == 0 (currently impossible?) as a new entry point. The first instruction would be a jump if it's zero (which should be OK since segment regs would be fine in the case this branch is taken). Out of my depth in x86-land... Rusty. -- Help! Save Australia from the worst of the DMCA: http://linux.org.au/law
