Re: [PATCH v3 1/6] migration: Allow immutable device state to be migrated early (i.e., before RAM)

[Michael S. Tsirkin]

On Fri, Dec 23, 2022 at 10:34:36AM +0100, David Hildenbrand wrote:
> On 22.12.22 12:02, David Hildenbrand wrote:
> > For virtio-mem, we want to have the plugged/unplugged state of memory
> > blocks available before migrating any actual RAM content. This
> > information is immutable on the migration source while migration is active,
> > 
> > For example, we want to use this information for proper preallocation
> > support with migration: currently, we don't preallocate memory on the
> > migration target, and especially with hugetlb, we can easily run out of
> > hugetlb pages during RAM migration and will crash (SIGBUS) instead of
> > catching this gracefully via preallocation.
> > 
> > Migrating device state before we start iterating is currently impossible.
> > Introduce and use qemu_savevm_state_start_precopy(), and use
> > a new special migration priority -- MIG_PRI_POST_SETUP -- to decide whether
> > state will be saved in qemu_savevm_state_start_precopy() or in
> > qemu_savevm_state_complete_precopy_*().
> > 
> > We have to take care of properly including the early device state in the
> > vmdesc. Relying on migrate_get_current() to temporarily store the vmdesc is
> > a bit sub-optimal, but we use that explicitly or implicitly all over the
> > place already, so this barely matters in practice.
> > 
> > Note that only very selected devices (i.e., ones seriously messing with
> > RAM setup) are supposed to make use of that.
> > 
> > Signed-off-by: David Hildenbrand <david@redhat.com>
> 
> [...]
> 
> >       if (inactivate_disks) {
> > @@ -1427,6 +1474,10 @@ int 
> > qemu_savevm_state_complete_precopy_non_iterable(QEMUFile *f,
> >           qemu_put_buffer(f, (uint8_t *)json_writer_get(vmdesc), 
> > vmdesc_len);
> >       }
> > +    /* Free it now to detect any inconsistencies. */
> > +    g_free(vmdesc);
> 
> Missed to convert that to a json_writer_free().


I get it you will post v4?

> -- 
> Thanks,
> 
> David / dhildenb