Michal Privoznik <mprivozn(a)redhat.com> [2019-03-27, 02:05PM +0100]:
On 3/25/19 9:04 AM, Bjoern Walk wrote:
> @@ -14069,6 +14072,39 @@
qemuDomainStatePanicInfoFormatMsg(qemuDomainStatePanicInfoPtr info)
> }
> +void
> +qemuDomainStatePanicInfoSet(virDomainObjPtr vm,
> + qemuDomainStatePanicInfoPtr info)
> +{
> + qemuDomainObjPrivatePtr priv = vm->privateData;
> +
> + if (!priv->panicInfo && VIR_ALLOC(priv->panicInfo) < 0)
> + return;
> +
> + priv->panicInfo->type = info->type;
> +
> + switch (info->type) {
> + case QEMU_DOMAIN_STATE_PANIC_INFO_TYPE_HYPERV:
> + priv->panicInfo->data.hyperv.arg1 = info->data.hyperv.arg1;
> + priv->panicInfo->data.hyperv.arg2 = info->data.hyperv.arg2;
> + priv->panicInfo->data.hyperv.arg3 = info->data.hyperv.arg3;
> + priv->panicInfo->data.hyperv.arg4 = info->data.hyperv.arg4;
> + priv->panicInfo->data.hyperv.arg5 = info->data.hyperv.arg5;
> + break;
> + case QEMU_DOMAIN_STATE_PANIC_INFO_TYPE_S390:
> + priv->panicInfo->data.s390.core = info->data.s390.core;
> + priv->panicInfo->data.s390.psw_mask = info->data.s390.psw_mask;
> + priv->panicInfo->data.s390.psw_addr = info->data.s390.psw_addr;
> + ignore_value(VIR_STRDUP(priv->panicInfo->data.s390.reason,
> + info->data.s390.reason));
> + break;
> + case QEMU_DOMAIN_STATE_PANIC_INFO_TYPE_NONE:
> + case QEMU_DOMAIN_STATE_PANIC_INFO_TYPE_LAST:
> + break;
> + }
How about memcpy() and then just STRDUP() in the one place we need to?
No need to send v3 for that, just asking and if so, I can fix that before
pushing.
My reasoning is that it might be easy to forget update this function if
qemuDomainStatePanicInfo struct gets a new member.
Yes, I agree, that's much better.
Michal