Re: [PATCH v10 02/21] acpi/generic_event_device: Update GHES migration to cover hest addr

[Peter Xu <peterx@redhat.com>]

On Tue, Sep 17, 2024 at 11:19:21AM +0200, Igor Mammedov wrote:
> On Sat, 14 Sep 2024 08:13:23 +0200
> Mauro Carvalho Chehab <mchehab+huawei@kernel.org> wrote:
> 
> > The GHES migration logic at GED should now support HEST table
> > location too.
> > 
> > Increase migration version and change needed to check for both
> > ghes_addr_le and hest_addr_le
> But I don't think it will work like this (but I might be easily wrong)
> However I don't know enough to properly review this patch, CCing Peter & Fabiano
> 
> > Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
> > ---
> >  hw/acpi/generic_event_device.c | 11 ++++++-----
> >  1 file changed, 6 insertions(+), 5 deletions(-)
> > 
> > diff --git a/hw/acpi/generic_event_device.c b/hw/acpi/generic_event_device.c
> > index 15b4c3ebbf24..4e5e387ee2df 100644
> > --- a/hw/acpi/generic_event_device.c
> > +++ b/hw/acpi/generic_event_device.c
> > @@ -343,10 +343,11 @@ static const VMStateDescription vmstate_ged_state = {
> >  
> >  static const VMStateDescription vmstate_ghes = {
> >      .name = "acpi-ghes",
> > -    .version_id = 1,
> > -    .minimum_version_id = 1,
> > +    .version_id = 2,
> > +    .minimum_version_id = 2,
> >      .fields = (const VMStateField[]) {
> >          VMSTATE_UINT64(ghes_addr_le, AcpiGhesState),
> > +        VMSTATE_UINT64(hest_addr_le, AcpiGhesState),
> >          VMSTATE_END_OF_LIST()
> >      },
> >  };
> > @@ -354,13 +355,13 @@ static const VMStateDescription vmstate_ghes = {
> >  static bool ghes_needed(void *opaque)
> >  {
> >      AcpiGedState *s = opaque;
> > -    return s->ghes_state.ghes_addr_le;
> > +    return s->ghes_state.ghes_addr_le && s->ghes_state.hest_addr_le;
> >  }
> 
> what I would do:
>   add ghes_needed_v2(): return  s->ghes_state.hest_addr_le;
> 
> and then instead of reusing vmstate_ghes_state would add new
> vmstate_ghes_v2_state subsection that migrates only 
>   VMSTATE_UINT64(hest_addr_le, AcpiGhesState)
> field.
> 
> btw: we probably don't need ghes_addr_le for new code that
> uses HEST to lookup relevant error status block.
> but we should still keep it for 9.1 and older machine types
> as they expect/use it. Separate subsections would work with
> this req just fine.

Right, if we need bi-directional migration we need above and a compat
property (or VMSTATE_UINT64_TEST() would work too, iiuc).

OTOH VMSD versioning only works for forward migration, not backward.

> 
> >  static const VMStateDescription vmstate_ghes_state = {
> >      .name = "acpi-ged/ghes",
> > -    .version_id = 1,
> > -    .minimum_version_id = 1,
> > +    .version_id = 2,
> > +    .minimum_version_id = 2,

(and IIUC if we set min ver=2, even forward migration should fail.. better
 test it with an old binary, migrating back and forth)

> >      .needed = ghes_needed,
> >      .fields = (const VMStateField[]) {
> >          VMSTATE_STRUCT(ghes_state, AcpiGedState, 1,
> 

Thanks,

-- 
Peter Xu