Re: [libvirt] [PATCHv2.5 08/10] qemu: conf: Add support for memory device cold(un)plug
> +
> +int
> +virDomainMemoryInsert(virDomainDefPtr def,
> + virDomainMemoryDefPtr mem)
> +{
> + int id = def->nmems;
> +
> + if (mem->info.type != VIR_DOMAIN_DEVICE_ADDRESS_TYPE_NONE &&
> + virDomainDefHasDeviceAddress(def, &mem->info)) {
Hmm... so if our incoming mem has an address defined - it could be
anything - we're just failing declaring that the domain already contains
a device with the same address? Doesn't seem right.
And again - we have this problem of TYPE_NONE, TYPE_DIMM, TYPE_LAST and
who knows what in the future.
Perhaps I was thinking too much about the various Find* API's when I
first read this, but upon further reflection and looking at the code
paths using it I see what the goal is. So safely ignore this particular
comment.
John
> + virReportError(VIR_ERR_OPERATION_INVALID,
"%s",
> + _("Domain already contains a device with the same
"
> + "address"));
> + return -1;
> + }
> +
> + if (VIR_APPEND_ELEMENT(def->mems, def->nmems, mem) < 0)
> + return -1;
> +
> + return id;
> +}
> +