Paolo Bonzini <pbonz...@redhat.com> wrote: > Il 21/09/2012 10:47, Juan Quintela ha scritto: >> Signed-off-by: Juan Quintela <quint...@redhat.com> > > Consider making the save_live functions return the expected downtime in > an int64_t* argument. The loop then can sum all the expected downtimes > and store them in s->expected_downtime, removing the need for patch 6.
If you look at the migration thread patches, I change the functions to return the "pending amount". And the migration thread does the bandwidth calculations. It is just impossible to do the right bandwidth calculation locally. Later, Juan.
