Re: [Qemu-devel] [Qemu-ppc] [PATCH 02/77] ppc: Use split I/D mmu modes to avoid flushes on interrupts

[David Gibson]

On Mon, Nov 16, 2015 at 09:10:24PM +1100, Benjamin Herrenschmidt wrote:
> Thanks for reviewing !
> On Mon, 2015-11-16 at 15:49 +1100, David Gibson wrote:
[snip]
> > > diff --git a/target-ppc/machine.c b/target-ppc/machine.c
> > > index f4ac761..b969492 100644
> > > --- a/target-ppc/machine.c
> > > +++ b/target-ppc/machine.c
> > > @@ -90,9 +90,11 @@ static int cpu_load_old(QEMUFile *f, void
> > > *opaque, int version_id)
> > >      qemu_get_betls(f, &env->nip);
> > >      qemu_get_betls(f, &env->hflags);
> > >      qemu_get_betls(f, &env->hflags_nmsr);
> > > -    qemu_get_sbe32s(f, &env->mmu_idx);
> > 
> > Have I missed something, or do you still need a read here to read the
> > mmux_idx, even though you'll ignore it, otherwise you'll get out of
> > sync and break migration from an old stream.
> 
> I am not completely cognizant of the migration stuff, that's very
> possible yes. Do I need to read into a dummy local ? Or is there
> a way to just say "drop 4 bytes from stream" ?

There's actually an example of this just below :)

> Note that I have generally completely overlooked the migration impact
> of my patches, this is something that I need to do but I wouldn't mind
> your help identifying the parts.

Ok, I'll do what I can.

> > >      qemu_get_sbe32(f); /* Discard unused power_mode */

          ^^^
          example of skipping a redundant value in the migration
          stream

-- 
David Gibson                    | I'll have my music baroque, and my code
david AT gibson.dropbear.id.au  | minimalist, thank you.  NOT _the_ _other_
                                | _way_ _around_!
http://www.ozlabs.org/~dgibson

signature.asc
Description: PGP signature