qemu-devel
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [PATCH v1 06/13] migrate/ram: Discard new RAM when growing RAM block


From: Dr. David Alan Gilbert
Subject: Re: [PATCH v1 06/13] migrate/ram: Discard new RAM when growing RAM blocks and the VM is stopped
Date: Fri, 21 Feb 2020 15:51:13 +0000
User-agent: Mutt/1.13.3 (2020-01-12)

* David Hildenbrand (address@hidden) wrote:
> On 19.02.20 17:17, David Hildenbrand wrote:
> > In case we grow our RAM after ram_postcopy_incoming_init() (e.g., when
> > synchronizing the RAM block state with the migration source), the resized
> > part would not get discarded. Let's perform that when being notified
> > about a resize while postcopy has been advised and the guest is not
> > running yet.
> > 
> > Cc: "Dr. David Alan Gilbert" <address@hidden>
> > Cc: Juan Quintela <address@hidden>
> > Cc: Peter Xu <address@hidden>
> > Signed-off-by: David Hildenbrand <address@hidden>
> > ---
> >  migration/ram.c | 19 +++++++++++++++++++
> >  1 file changed, 19 insertions(+)
> > 
> > diff --git a/migration/ram.c b/migration/ram.c
> > index 57f32011a3..cbd54947fb 100644
> > --- a/migration/ram.c
> > +++ b/migration/ram.c
> > @@ -3722,6 +3722,25 @@ static void 
> > ram_mig_ram_block_resized(RAMBlockNotifier *n, void *host,
> >          return;
> >      }
> >  
> > +    /*
> > +     * Especially at the start of precopy on the migration target, before
> > +     * starting postcopy, we synchronize the RAM block sizes. Let's make 
> > sure
> > +     * that any resizes before starting the guest are properly handled by
> > +     * postcopy. Note: All other postcopy handling (e.g., registering 
> > handlers,
> > +     * disabling THP) happens after all resizes (e.g., during precopy) were
> > +     * performed.
> > +     */
> 
> I think it would be clearer to do a
> 
> ps >= POSTCOPY_INCOMING_ADVISE && ps < POSTCOPY_INCOMING_RUNNING
> 
> We really only want to do something when psotcopy has been advised but
> the guest is not running yet.
> 
> Will look into that as I find ways to actually test this :)

Should that be < POSTCOPY_INCOMING_LISTENING - i.e. before the
userfaultfd has been enabled on the region?

Dave

> 
> -- 
> Thanks,
> 
> David / dhildenb
--
Dr. David Alan Gilbert / address@hidden / Manchester, UK




reply via email to

[Prev in Thread] Current Thread [Next in Thread]