On 2012-01-06 13:58, Peter Maydell wrote: > On 6 January 2012 13:30, Jan Kiszka <jan.kis...@web.de> wrote: >> The third point indicates that there is rather more generic room for >> improvements: Why should qemu reset device models before restore at all? > > Commit 5a8a49d7aa says: > # if we load from a snapshot, the machine can be in any state. That can > # cause troubles if loading an older image which does not carry all state > # information the executing QEMU requires. Hardly any device takes care of > # this scenario. > > I think it's also easier to reason about devices if you know that the > device before a load was in a known clean state rather than being > anything at all.
True, that's why we do this ATM. But issuing the reset globally without giving devices chance to handle this more precisely is just a suboptimal workaround, one that we may want to overcome at some point. Jan
signature.asc
Description: OpenPGP digital signature
