Re: [PATCH v3 10/10] vfio: unplug failover primary device before migration

[Cornelia Huck <cohuck@redhat.com>]

On Fri, 11 Oct 2019 13:20:15 +0200
Jens Freimann <jfreimann@redhat.com> wrote:

> As usual block all vfio-pci devices from being migrated, but make an
> exception for failover primary devices. This is achieved by setting
> unmigratable to 0 but also add a migration blocker for all vfio-pci
> devices except failover primary devices. These will be unplugged before
> migration happens by the migration handler of the corresponding
> virtio-net standby device.
> 
> Signed-off-by: Jens Freimann <jfreimann@redhat.com>
> ---
>  hw/vfio/pci.c | 35 ++++++++++++++++++++++++++++++++++-
>  hw/vfio/pci.h |  2 ++
>  2 files changed, 36 insertions(+), 1 deletion(-)
> 
> diff --git a/hw/vfio/pci.c b/hw/vfio/pci.c
> index c5e6fe61cb..64cf8e07d9 100644
> --- a/hw/vfio/pci.c
> +++ b/hw/vfio/pci.c
> @@ -40,6 +40,9 @@
>  #include "pci.h"
>  #include "trace.h"
>  #include "qapi/error.h"
> +#include "migration/blocker.h"
> +#include "qemu/option.h"
> +#include "qemu/option_int.h"
>  
>  #define TYPE_VFIO_PCI "vfio-pci"
>  #define PCI_VFIO(obj)    OBJECT_CHECK(VFIOPCIDevice, obj, TYPE_VFIO_PCI)
> @@ -2698,6 +2701,12 @@ static void vfio_unregister_req_notifier(VFIOPCIDevice *vdev)
>      vdev->req_enabled = false;
>  }
>  
> +static int has_net_failover_arg(void *opaque, const char *name,
> +                           const char *value, Error **errp)
> +{
> +    return (strcmp(name, "net_failover_pair_id") == 0);
> +}
> +
>  static void vfio_realize(PCIDevice *pdev, Error **errp)
>  {
>      VFIOPCIDevice *vdev = PCI_VFIO(pdev);
> @@ -2710,6 +2719,20 @@ static void vfio_realize(PCIDevice *pdev, Error **errp)
>      int groupid;
>      int i, ret;
>      bool is_mdev;
> +    uint16_t class_id;
> +
> +    if (qemu_opt_foreach(pdev->qdev.opts, has_net_failover_arg,
> +                         (void *) pdev->qdev.opts, &err) == 0) {
> +        error_setg(&vdev->migration_blocker,
> +                "VFIO device doesn't support migration");
> +        ret = migrate_add_blocker(vdev->migration_blocker, &err);
> +        if (err) {
> +            error_propagate(errp, err);
> +            error_free(vdev->migration_blocker);
> +        }

Do we really want to continue here if adding the migration blocker
failed? Previously, adding the vmsd with unmigratable=1 would have
prevented realization of the device if --only-migratable was specified;
I guess we should keep that behaviour and just bail out here?

> +    } else {
> +        pdev->qdev.allow_unplug_during_migration = true;
> +    }
>  
>      if (!vdev->vbasedev.sysfsdev) {
>          if (!(~vdev->host.domain || ~vdev->host.bus ||
> @@ -2812,6 +2835,14 @@ static void vfio_realize(PCIDevice *pdev, Error **errp)
>          goto error;
>      }
>  
> +    if (vdev->net_failover_pair_id != NULL) {
> +        class_id = pci_get_word(pdev->config + PCI_CLASS_DEVICE);
> +        if (class_id != PCI_CLASS_NETWORK_ETHERNET) {
> +            error_setg(errp, "failover device is not an Ethernet device");
> +            goto error;
> +        }
> +    }
> +
>      /* vfio emulates a lot for us, but some bits need extra love */
>      vdev->emulated_config_bits = g_malloc0(vdev->config_size);
>  
> @@ -3110,6 +3141,8 @@ static Property vfio_pci_dev_properties[] = {
>                              display, ON_OFF_AUTO_OFF),
>      DEFINE_PROP_UINT32("xres", VFIOPCIDevice, display_xres, 0),
>      DEFINE_PROP_UINT32("yres", VFIOPCIDevice, display_yres, 0),
> +    DEFINE_PROP_STRING("net_failover_pair_id", VFIOPCIDevice,
> +            net_failover_pair_id),
>      DEFINE_PROP_UINT32("x-intx-mmap-timeout-ms", VFIOPCIDevice,
>                         intx.mmap_timeout, 1100),
>      DEFINE_PROP_BIT("x-vga", VFIOPCIDevice, features,
> @@ -3152,7 +3185,7 @@ static Property vfio_pci_dev_properties[] = {
>  
>  static const VMStateDescription vfio_pci_vmstate = {
>      .name = "vfio-pci",
> -    .unmigratable = 1,
> +    .unmigratable = 0,

Is this vmsd useful in any way with that change anymore?

>  };
>  

(...)