On 01/05/2018 03:20 AM, Michal Privoznik wrote:
On 01/05/2018 01:15 AM, John Ferlan wrote:
>
>
> On 12/13/2017 09:30 AM, Michal Privoznik wrote:
>> Since we have user aliases it may happen that users want to
>> change it using 'update-device'. Instead of ignoring it silently,
>> error out loudly.
>>
>> Signed-off-by: Michal Privoznik <mprivozn(a)redhat.com>
>> ---
>> src/qemu/qemu_domain.c | 8 ++++++++
>> 1 file changed, 8 insertions(+)
>>
>
> So IIUC this would be adding a check that the libvirt supplied alias
> matches, true? But your goal is to check if a "ua-" prefixed alias was
> being used - so should the check use of the prefix?
I don't think this is limited to user provided aliases only. For
instance, libvirt generates the following aliases for disks: scsi0-0-0,
ide0-1-0, fdc0-0-0.
Now, what should happen if an user provides new disk XML trying to
change nothing but 'scsi0-0-0' to say 'scsi0-0-1' (for live domain)? I
think it should be forbidden - we don't have way to change IDs in qemu,
do we? We certainly lack implementation to support it.
Therefore, I think we don't have to limit the check just for 'ua-' prefixes.
OK - fair enough... Perhaps the commit message is what caught my
attention/focus to ua- in particular. Maybe the extra explanation here
can be worked in.
Reviewed-by: John Ferlan <jferlan(a)redhat.com>
John
>
> There's probably a few more info. fields that could be considered - any
> thoughts on those? Perhaps fairly difficult to keep up with the various
> _virDomainDiskDef and friends alterations for what this function is
> indicating...
Yeah, probably. I'm not sure. Maybe we will find some more.
Michal