Re: [Xen-devel] [PATCH 03/28] libxl: Provide libxl__dm_support_*

[Ian Campbell <ian.campbell@xxxxxxxxxx>]

On Thu, 2016-01-14 at 11:31 -0700, Jim Fehlig wrote:
> Ian Campbell wrote:
> > On Mon, 2016-01-11 at 10:00 -0700, Jim Fehlig wrote:
> > > On 01/07/2016 10:13 AM, Ian Campbell wrote:
> > > > On Tue, 2015-12-22 at 18:44 +0000, Ian Jackson wrote:
> > > > > This allows code elsewhere in libxl to find out what options a
> > > > > device
> > > > > model executable supports.ÂÂThis is done by searching the usage
> > > > > message for fixed strings.
> > > > Has anyone (ever, not necessarily a Xen person nor in this context)
> > > > approached upstream QEMU about a machine readable output of some
> > > > sort?
> > > > 
> > > > I know libvirt does something similar to this, but they want to
> > > > support
> > > > older versions, whereas we at least have the luxury of not caring
> > > > about
> > > > versions before the point this code lands.
> > > Since qemu 1.2.0, libvirt has been using the various QMP commands to
> > > probe for
> > > qemu capabilities, instead of parsing help output.
> > 
> > As in it spawns a qemu specifically to ask the questions and then kills
> > it
> > and starts what it needs _or_ it starts the qemu with minimal command
> > line
> > cfg and then dynamically pokes in the full config via qmp?
> 
> The latter.

From the description I think you meant former?

>  When the qemu driver loads, it starts qemu, pokes it for
> capabilities via qmp, caches what it finds, then terminates it. The cached
> capabilities are used until the associated qemu binary is changed/updated.
> 
> If interested in peeking at the code, see virQEMUCapsInitQMP() and the 
> functions
> it calls in $libvirt_src/src/qemu/qemu_capabilities.c. E.g.
> 
> virQEMUCapsInitQMP()
> Â -> virQEMUCapsInitQMPMonitor()
> ÂÂÂÂ-> virQEMUCapsInitQMPBasic()

Thanks.

When I spoke to Ian J yesterday we decided doing this properly (via QMP as
above) would be nice it was out of scope for this series for now. It would
make a nice future bit of cleanup though for sure.

Ian.

> 
> Regards,
> Jim
> 
> 
> _______________________________________________
> Xen-devel mailing list
> Xen-devel@xxxxxxxxxxxxx
> http://lists.xen.org/xen-devel

_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxx
http://lists.xen.org/xen-devel