* Jason Gunthorpe (jgg@xxxxxxxxxx) wrote: > On Mon, Oct 25, 2021 at 07:47:29PM +0100, Dr. David Alan Gilbert wrote: > > > It may need some further refinement; for example in that quiesed state > > do counters still tick? will a NIC still respond to packets that don't > > get forwarded to the host? > > At least for the mlx5 NIC the two states are 'able to issue outbound > DMA' and 'all internal memories and state are frozen and unchanging'. Yeh, so my point was just that if you're adding a new state to this process, you need to define the details like that. Dave > The later means it should not respond/count/etc to network packets > either. > > Roughly it is > 'able to mutate external state' / 'able to mutate internal state' > > The invariant should be that successive calls to serialize the > internal state while frozen should return the same serialization. > > We may find that migratable PCI functions must have some limited > functionality. Ie anything with a realtime compoment - for instance a > sync-ethernet clock synchronization control loop, may become > challenging to migrate without creating a serious functional > distortion. > > Jason > -- Dr. David Alan Gilbert / dgilbert@xxxxxxxxxx / Manchester, UK
