Re: [Qemu-devel] [PATCH v3 5/6] s390x/css: activate ChannelSubSys migration

[Thomas Huth <thuth@redhat.com>]

On 11.07.2017 16:54, Halil Pasic wrote:
> Turn on migration for the channel subsystem for the next machine.  For
> legacy machines we still have to do things the old way.
> 
> Signed-off-by: Halil Pasic <pasic@linux.vnet.ibm.com>
> Reviewed-by: Cornelia Huck <cornelia.huck@de.ibm.com>
> ---
>  hw/s390x/css.c             | 5 +++++
>  hw/s390x/s390-virtio-ccw.c | 9 ++++-----
>  include/hw/s390x/css.h     | 4 ++++
>  3 files changed, 13 insertions(+), 5 deletions(-)
> 
> diff --git a/hw/s390x/css.c b/hw/s390x/css.c
> index 20b5fbed68..b89b60751a 100644
> --- a/hw/s390x/css.c
> +++ b/hw/s390x/css.c
> @@ -407,6 +407,11 @@ static int subch_dev_post_load(void *opaque, int version_id)
>      return 0;
>  }
>  
> +void css_register_vmstate(void)
> +{
> +    vmstate_register(NULL, 0, &vmstate_css, &channel_subsys);
> +}
> +
>  IndAddr *get_indicator(hwaddr ind_addr, int len)
>  {
>      IndAddr *indicator;
> diff --git a/hw/s390x/s390-virtio-ccw.c b/hw/s390x/s390-virtio-ccw.c
> index 751febb87a..20e3f06519 100644
> --- a/hw/s390x/s390-virtio-ccw.c
> +++ b/hw/s390x/s390-virtio-ccw.c
> @@ -206,7 +206,7 @@ static void ccw_machine_class_init(ObjectClass *oc, void *data)
>  
>      s390mc->ri_allowed = true;
>      s390mc->cpu_model_allowed = true;
> -    s390mc->css_migration_enabled = false; /* TODO: set to true */
> +    s390mc->css_migration_enabled = true;
>      mc->init = ccw_init;
>      mc->reset = s390_machine_reset;
>      mc->hot_add_cpu = s390_hot_add_cpu;
> @@ -482,10 +482,9 @@ bool css_migration_enabled(void)
>  
>  static void ccw_machine_2_10_instance_options(MachineState *machine)
>  {
> -    /*
> -     * TODO Once preparations are done register vmstate for the css if
> -     * css_migration_enabled().
> -     */
> +    if (css_migration_enabled()) {
> +        css_register_vmstate();
> +    }
>  }

The location for the above hunk seems to be quite unfortunate: For the
s390-ccw-virtio-2.11 machine, we now do not call css_register_vmstate()
anymore.

Maybe this should rather go into ccw_init() instead?

 Thomas