On 08/20/2018 05:07 PM, Daniel P. Berrangé wrote:
On Tue, Aug 14, 2018 at 01:19:42PM +0200, Michal Privoznik wrote:
> In order for our drivers to lock resources for metadata change we
> need set of new APIs. Fortunately, we don't have to care about
> every possible device a domain can have. We care only about those
> which can live on a network filesystem and hence can be accessed
> by multiple daemons at the same time. These devices are covered
> in virDomainLockMetadataLock() and only a small fraction of
> those can be hotplugged (covered in the rest of the introduced
> APIs).
I'm not sure I understand the rationale behind saying we only care
about resources on network filesystems.
If I have 2 locally running guests, and both have a serial port
backed by a physical serial port, eg
<serial type="dev">
<source path="/dev/ttyS0"/>
<target port="1"/>
</serial>
we *do* care about locking /dev/ttyS0, as libvirtd isn't doing
mutual exclusion checks anywhere else for the /dev/ttyS0 device
node.
Ah you mean that the system wide daemon and session daemon could clash
when relabeling /dev/ttyS0? Well, we don't do relabeling for session
daemons and running two system daemons is not supported (you're gonna
hit more serious problems when trying that anyway).
In general I think we need to lock every single file resource
that is labelled for a guest, regardless of whether its local
or remote.
Well this might be feasible iff locking would be done from security driver.
Michal