[PATCH] failover: unregister ram on unplug

[PATCH] failover: unregister ram on unplug

[Laurent Vivier]

This simple change allows to test failover with a simulated device
like e1000e rather than a vfio device.

This is interesting to developers that want to test failover on
a system with no vfio device. Moreover it simplifies host networking
configuration as we can use the same bridge for virtio-net and
the other failover networking device.

Without this change the migration of a system configured with failover
fails with:

  Unknown ramblock "0000:00:01.1:00.0/e1000e.rom", cannot accept migration
  error while loading state for instance 0x0 of device 'ram'
  load of migration failed: Invalid argument

Signed-off-by: Laurent Vivier <lvivier@redhat.com>
---
 hw/net/virtio-net.c | 3 +++
 1 file changed, 3 insertions(+)

diff --git a/hw/net/virtio-net.c b/hw/net/virtio-net.c
index 16d20cdee52a..8f7735bad4f2 100644
--- a/hw/net/virtio-net.c
+++ b/hw/net/virtio-net.c
@@ -3256,6 +3256,9 @@ static void virtio_net_handle_migration_primary(VirtIONet *n, MigrationState *s)
     if (migration_in_setup(s) && !should_be_hidden) {
         if (failover_unplug_primary(n, dev)) {
             vmstate_unregister(VMSTATE_IF(dev), qdev_get_vmsd(dev), dev);
+            if (PCI_DEVICE(dev)->has_rom) {
+                vmstate_unregister_ram(&PCI_DEVICE(dev)->rom , dev);
+            }
             qapi_event_send_unplug_primary(dev->id);
             qatomic_set(&n->failover_primary_hidden, true);
         } else {
-- 
2.31.1

Re: [PATCH] failover: unregister ram on unplug

[Igor Mammedov]

On Tue, 20 Jul 2021 20:16:44 +0200
Laurent Vivier <lvivier@redhat.com> wrote:

> This simple change allows to test failover with a simulated device
> like e1000e rather than a vfio device.
> 
> This is interesting to developers that want to test failover on
> a system with no vfio device. Moreover it simplifies host networking
> configuration as we can use the same bridge for virtio-net and
> the other failover networking device.
> 
> Without this change the migration of a system configured with failover
> fails with:
> 
>   Unknown ramblock "0000:00:01.1:00.0/e1000e.rom", cannot accept migration
>   error while loading state for instance 0x0 of device 'ram'
>   load of migration failed: Invalid argument
> 
> Signed-off-by: Laurent Vivier <lvivier@redhat.com>

Reviewed-by: Igor Mammedov <imammedo@redhat.com>

> ---
>  hw/net/virtio-net.c | 3 +++
>  1 file changed, 3 insertions(+)
> 
> diff --git a/hw/net/virtio-net.c b/hw/net/virtio-net.c
> index 16d20cdee52a..8f7735bad4f2 100644
> --- a/hw/net/virtio-net.c
> +++ b/hw/net/virtio-net.c
> @@ -3256,6 +3256,9 @@ static void virtio_net_handle_migration_primary(VirtIONet *n, MigrationState *s)
>      if (migration_in_setup(s) && !should_be_hidden) {
>          if (failover_unplug_primary(n, dev)) {
>              vmstate_unregister(VMSTATE_IF(dev), qdev_get_vmsd(dev), dev);
> +            if (PCI_DEVICE(dev)->has_rom) {
> +                vmstate_unregister_ram(&PCI_DEVICE(dev)->rom , dev);
> +            }
>              qapi_event_send_unplug_primary(dev->id);
>              qatomic_set(&n->failover_primary_hidden, true);
>          } else {

Re: [PATCH] failover: unregister ram on unplug

[Laurent Vivier]

On 21/07/2021 10:58, Igor Mammedov wrote:
> On Tue, 20 Jul 2021 20:16:44 +0200
> Laurent Vivier <lvivier@redhat.com> wrote:
> 
>> This simple change allows to test failover with a simulated device
>> like e1000e rather than a vfio device.
>>
>> This is interesting to developers that want to test failover on
>> a system with no vfio device. Moreover it simplifies host networking
>> configuration as we can use the same bridge for virtio-net and
>> the other failover networking device.
>>
>> Without this change the migration of a system configured with failover
>> fails with:
>>
>>   Unknown ramblock "0000:00:01.1:00.0/e1000e.rom", cannot accept migration
>>   error while loading state for instance 0x0 of device 'ram'
>>   load of migration failed: Invalid argument
>>
>> Signed-off-by: Laurent Vivier <lvivier@redhat.com>
> 
> Reviewed-by: Igor Mammedov <imammedo@redhat.com>
> 

I've sent a v2 of this patch to reset has_rom to false.

I've updated the commit log message and the subject of the email to "failover: unregister
ROM on unplug" (that is more correct).

Thanks,
Laurent

Re: [PATCH] failover: unregister ram on unplug

[Michael S. Tsirkin]

On Wed, Jul 21, 2021 at 10:58:17AM +0200, Igor Mammedov wrote:
> On Tue, 20 Jul 2021 20:16:44 +0200
> Laurent Vivier <lvivier@redhat.com> wrote:
> 
> > This simple change allows to test failover with a simulated device
> > like e1000e rather than a vfio device.
> > 
> > This is interesting to developers that want to test failover on
> > a system with no vfio device. Moreover it simplifies host networking
> > configuration as we can use the same bridge for virtio-net and
> > the other failover networking device.
> > 
> > Without this change the migration of a system configured with failover
> > fails with:
> > 
> >   Unknown ramblock "0000:00:01.1:00.0/e1000e.rom", cannot accept migration
> >   error while loading state for instance 0x0 of device 'ram'
> >   load of migration failed: Invalid argument
> > 
> > Signed-off-by: Laurent Vivier <lvivier@redhat.com>
> 
> Reviewed-by: Igor Mammedov <imammedo@redhat.com>
> 
> > ---
> >  hw/net/virtio-net.c | 3 +++
> >  1 file changed, 3 insertions(+)
> > 
> > diff --git a/hw/net/virtio-net.c b/hw/net/virtio-net.c
> > index 16d20cdee52a..8f7735bad4f2 100644
> > --- a/hw/net/virtio-net.c
> > +++ b/hw/net/virtio-net.c
> > @@ -3256,6 +3256,9 @@ static void virtio_net_handle_migration_primary(VirtIONet *n, MigrationState *s)
> >      if (migration_in_setup(s) && !should_be_hidden) {
> >          if (failover_unplug_primary(n, dev)) {
> >              vmstate_unregister(VMSTATE_IF(dev), qdev_get_vmsd(dev), dev);
> > +            if (PCI_DEVICE(dev)->has_rom) {


Hmm. Any way to hide this behind an interface so
we don't need to poke at pci device internals?

> > +                vmstate_unregister_ram(&PCI_DEVICE(dev)->rom , dev);
> > +            }
> >              qapi_event_send_unplug_primary(dev->id);
> >              qatomic_set(&n->failover_primary_hidden, true);
> >          } else {

Re: [PATCH] failover: unregister ram on unplug

[Laurent Vivier]

On 21/07/2021 12:41, Michael S. Tsirkin wrote:
> On Wed, Jul 21, 2021 at 10:58:17AM +0200, Igor Mammedov wrote:
>> On Tue, 20 Jul 2021 20:16:44 +0200
>> Laurent Vivier <lvivier@redhat.com> wrote:
>>
>>> This simple change allows to test failover with a simulated device
>>> like e1000e rather than a vfio device.
>>>
>>> This is interesting to developers that want to test failover on
>>> a system with no vfio device. Moreover it simplifies host networking
>>> configuration as we can use the same bridge for virtio-net and
>>> the other failover networking device.
>>>
>>> Without this change the migration of a system configured with failover
>>> fails with:
>>>
>>>   Unknown ramblock "0000:00:01.1:00.0/e1000e.rom", cannot accept migration
>>>   error while loading state for instance 0x0 of device 'ram'
>>>   load of migration failed: Invalid argument
>>>
>>> Signed-off-by: Laurent Vivier <lvivier@redhat.com>
>>
>> Reviewed-by: Igor Mammedov <imammedo@redhat.com>
>>
>>> ---
>>>  hw/net/virtio-net.c | 3 +++
>>>  1 file changed, 3 insertions(+)
>>>
>>> diff --git a/hw/net/virtio-net.c b/hw/net/virtio-net.c
>>> index 16d20cdee52a..8f7735bad4f2 100644
>>> --- a/hw/net/virtio-net.c
>>> +++ b/hw/net/virtio-net.c
>>> @@ -3256,6 +3256,9 @@ static void virtio_net_handle_migration_primary(VirtIONet *n, MigrationState *s)
>>>      if (migration_in_setup(s) && !should_be_hidden) {
>>>          if (failover_unplug_primary(n, dev)) {
>>>              vmstate_unregister(VMSTATE_IF(dev), qdev_get_vmsd(dev), dev);
>>> +            if (PCI_DEVICE(dev)->has_rom) {
> 
> 
> Hmm. Any way to hide this behind an interface so
> we don't need to poke at pci device internals?

There is the pci_del_option_rom() but it's not exported.

Do you want I export and use it?

Thanks,
Laurent

> 
>>> +                vmstate_unregister_ram(&PCI_DEVICE(dev)->rom , dev);
>>> +            }
>>>              qapi_event_send_unplug_primary(dev->id);
>>>              qatomic_set(&n->failover_primary_hidden, true);
>>>          } else {
> 

Re: [PATCH] failover: unregister ram on unplug

[Philippe Mathieu-Daudé]

On 7/21/21 12:49 PM, Laurent Vivier wrote:
> On 21/07/2021 12:41, Michael S. Tsirkin wrote:
>> On Wed, Jul 21, 2021 at 10:58:17AM +0200, Igor Mammedov wrote:
>>> On Tue, 20 Jul 2021 20:16:44 +0200
>>> Laurent Vivier <lvivier@redhat.com> wrote:
>>>
>>>> This simple change allows to test failover with a simulated device
>>>> like e1000e rather than a vfio device.
>>>>
>>>> This is interesting to developers that want to test failover on
>>>> a system with no vfio device. Moreover it simplifies host networking
>>>> configuration as we can use the same bridge for virtio-net and
>>>> the other failover networking device.
>>>>
>>>> Without this change the migration of a system configured with failover
>>>> fails with:
>>>>
>>>>   Unknown ramblock "0000:00:01.1:00.0/e1000e.rom", cannot accept migration
>>>>   error while loading state for instance 0x0 of device 'ram'
>>>>   load of migration failed: Invalid argument
>>>>
>>>> Signed-off-by: Laurent Vivier <lvivier@redhat.com>
>>>
>>> Reviewed-by: Igor Mammedov <imammedo@redhat.com>
>>>
>>>> ---
>>>>  hw/net/virtio-net.c | 3 +++
>>>>  1 file changed, 3 insertions(+)
>>>>
>>>> diff --git a/hw/net/virtio-net.c b/hw/net/virtio-net.c
>>>> index 16d20cdee52a..8f7735bad4f2 100644
>>>> --- a/hw/net/virtio-net.c
>>>> +++ b/hw/net/virtio-net.c
>>>> @@ -3256,6 +3256,9 @@ static void virtio_net_handle_migration_primary(VirtIONet *n, MigrationState *s)
>>>>      if (migration_in_setup(s) && !should_be_hidden) {
>>>>          if (failover_unplug_primary(n, dev)) {
>>>>              vmstate_unregister(VMSTATE_IF(dev), qdev_get_vmsd(dev), dev);
>>>> +            if (PCI_DEVICE(dev)->has_rom) {
>>
>>
>> Hmm. Any way to hide this behind an interface so
>> we don't need to poke at pci device internals?
> 
> There is the pci_del_option_rom() but it's not exported.
> 
> Do you want I export and use it?

Looks cleaner indeed.