[libvirt] [PATCH v3 3/3] qemu: hot-unplug of watchdog

[John Ferlan]

On 09/27/2017 08:12 AM, Michal Privoznik wrote:
> https://bugzilla.redhat.com/show_bug.cgi?id=1447169
> 
> Signed-off-by: Michal Privoznik <mprivozn at redhat.com>
> ---
>  src/qemu/qemu_driver.c                             |  4 +-
>  src/qemu/qemu_hotplug.c                            | 67 ++++++++++++++++++++++
>  src/qemu/qemu_hotplug.h                            |  3 +
>  tests/qemuhotplugtest.c                            |  7 ++-
>  .../qemuhotplug-watchdog-full.xml                  |  4 ++
>  5 files changed, 83 insertions(+), 2 deletions(-)
>  create mode 100644 tests/qemuhotplugtestdevices/qemuhotplug-watchdog-full.xml
> 
[...]

>  
> +int
> +qemuDomainDetachWatchdog(virQEMUDriverPtr driver,
> +                         virDomainObjPtr vm,
> +                         virDomainWatchdogDefPtr dev)
> +{
> +    int ret = -1;
> +    virDomainWatchdogDefPtr watchdog = vm->def->watchdog;
> +    qemuDomainObjPrivatePtr priv = vm->privateData;
> +

Similar to the hot-plug side - does hot unplug work for older qemu's?

> +    /* While domains can have up to one watchdog, the one supplied by the user
> +     * doesn't necessarily match the one domain has. Refuse to detach in such
> +     * case. */
> +    if (!(watchdog &&
> +          watchdog->model == dev->model &&
> +          watchdog->action == dev->action &&
> +          virDomainDeviceInfoAddressIsEqual(&dev->info, &watchdog->info))) {
> +        virReportError(VIR_ERR_OPERATION_INVALID, "%s",
> +                       _("watchdog device not present in domain configuration"));
> +        return -1;
> +    }
> +
> +    if (watchdog->model != VIR_DOMAIN_WATCHDOG_MODEL_I6300ESB) {
> +        virReportError(VIR_ERR_OPERATION_UNSUPPORTED,
> +                       _("hotunplug of watchdog of model %s is not supported"),

"hot unplug"

> +                       virDomainWatchdogModelTypeToString(watchdog->model));
> +        return -1;
> +    }
> +
> +    qemuDomainMarkDeviceForRemoval(vm, &watchdog->info);
> +    qemuDomainObjEnterMonitor(driver, vm);
> +

Not a problem per se, just a question - is there a need to do any sort
of "watchdog-set-action" to say WATCHDOG_ACTION_NONE or RESET (since
RESET is the default if not provided).  I guess I'm just thinking
outside the box of how someone could add a watchdog device (not w/
libvirt) without an action if they knew how and the old action would
then conceivably take place. I can think of one such team that would try
something like that ;-)

In general - based on of course the qemuCaps question answer...

Reviewed-by: John Ferlan <jferlan at redhat.com>

John

> +    ret = qemuMonitorDelDevice(priv->mon, watchdog->info.alias);
> +
> +    if (qemuDomainObjExitMonitor(driver, vm) < 0)
> +        ret = -1;
> +
> +    if (ret == 0) {
> +        if ((ret = qemuDomainWaitForDeviceRemoval(vm)) == 1) {
> +            qemuDomainReleaseDeviceAddress(vm, &watchdog->info, NULL);
> +            ret = qemuDomainRemoveWatchdog(driver, vm, watchdog);
> +        }
> +    }
> +    qemuDomainResetDeviceRemoval(vm);
> +
> +    return ret;
> +}
> +
> +

[..]