On 07.12.2012, at 16:14, Christian Borntraeger wrote: > Alex, > > here is were the IPL device code would move into. Some rough edges are still > there, but it can ipl almost anything that was zipled under LPAR/VM.
Please try and check how nicely this can be done if we would map a firmware blob into the guest address space. I still believe that we can handle all boot strapping cases from within guest context. The great advantage would be that we get user interaction and consistent device state. And less code that runs under host privileges. Alex
