[Qemu-devel] Re: [PATCH] savevm: unbreak register_savevm_live()/vmstate_register_with_alias_id()

[Isaku Yamahata <yamahata@valinux.co.jp>]

I sent the patch out too early.
I found the issue is in pcibus_get_dev_path(). Sorry for noise.

On Wed, Jan 26, 2011 at 07:04:51AM -0700, Alex Williamson wrote:
> On Wed, 2011-01-26 at 18:45 +0900, Isaku Yamahata wrote:
> > This patch unbreaks 7685ee6abcb939104801f84b3fe9645412528088.
> > With the changeset, more than one instances of same device on bus
> > that provides get_dev_path method can't be created because it hits
> > the assertion.
> > This patch removes the assertion whose assumption isn't correct.
> 
> Can you give an example of where this is needed?  The point of
> get_dev_path is that when it's available, it's supposed to provide a
> unique string per device.  Since get_dev_path is really only implemented
> for PCI, does this mean you have two independent devices at the same PCI
> address?  Thanks,
> 
> Alex
> 
> > Cc: Alex Williamson <alex.williamson@redhat.com>
> > Signed-off-by: Isaku Yamahata <yamahata@valinux.co.jp>
> > ---
> >  savevm.c |    2 --
> >  1 files changed, 0 insertions(+), 2 deletions(-)
> > 
> > diff --git a/savevm.c b/savevm.c
> > index fcd8db4..cd29eb8 100644
> > --- a/savevm.c
> > +++ b/savevm.c
> > @@ -1126,7 +1126,6 @@ int register_savevm_live(DeviceState *dev,
> >      } else {
> >          se->instance_id = instance_id;
> >      }
> > -    assert(!se->compat || se->instance_id == 0);
> >      /* add at the end of list */
> >      QTAILQ_INSERT_TAIL(&savevm_handlers, se, entry);
> >      return 0;
> > @@ -1236,7 +1235,6 @@ int vmstate_register_with_alias_id(DeviceState *dev, int instance_id,
> >      } else {
> >          se->instance_id = instance_id;
> >      }
> > -    assert(!se->compat || se->instance_id == 0);
> >      /* add at the end of list */
> >      QTAILQ_INSERT_TAIL(&savevm_handlers, se, entry);
> >      return 0;
> 
> 
> 

-- 
yamahata