Re: [Qemu-devel] [RFC PATCH v2 1/1] s390x/css: unrestrict cssids

[Dong Jia Shi]

* Cornelia Huck <address@hidden> [2017-11-29 12:47:47 +0100]:

[...]

> > With this patch
> > ===============
> > 
> > ------------+---------------+-------------
> >             | squashing off | squashing on
> > ------------+---------------+-------------
> >     auto id |        F      |     F
> > ------------+---------------+-------------
> > explicit id |        S'     |     S
> > ------------+---------------+-------------
> > 
> > T5. squashing off + auto id
> >   qemu-system-s390x: Unknown savevm section or instance 
> > '/fe.0.0003/virtio-rng' 0
> >   qemu-system-s390x: load of migration failed: Invalid argument
> > [Fail due to busid mismatch.]
> > 
> > T6. squashing off + explicit given id
> >   qemu-system-s390x: vmstate: get_nullptr expected VMS_NULLPTR_MARKER
> >   qemu-system-s390x: Failed to load s390_css:css
> >   qemu-system-s390x: error while loading state for instance 0x0 of device 
> > 's390_css'
> >   qemu-system-s390x: load of migration failed: Invalid argument
> > [Setting vfio-ccw.devno=non-fe.x.xxxx. (same as T1)
> >  Fail due to css mismatch - there is no css 0 in the new vm.]
> > 
> >   Succeed.
> > [Setting vfio-ccw.devno=fe.x.xxxx.]
> 
> Don't you need to attach the vfio-ccw device later anyway? You have to
> detach it from the source before you migrate, and I'd expect it to be
> symmetric.
Yes. After migrate, there is no problem to device_add the vfio-ccw
device (id=vfio0,devno=fe.x.xxxx). The result (succeed for this case)
does not change.

> 
> > 
> > T7. squashing on  + auto id
> >   qemu-system-s390x: Unknown savevm section or instance 
> > '/00.0.0003/virtio-rng' 0
> >   qemu-system-s390x: load of migration failed: Invalid argument
> > [Fail due to busid mismatch.]
> > 
> > T8. squashing on  + explicit given id
> >   Succeed.
> > 
> > 
> > Notice:
> > The differences of the test results between w and w/o this patch are in
> > the "squashing off" cases. I think these are things that we can accept.
> 
> Yes, I think that makes sense. If you want reliable migration, you need
> to be specific with your ids. I'd just don't want us to break things
> explicitly.
> 
Fair enough. Got the message.

-- 
Dong Jia Shi