[Spice-devel] [Users] oVirt console plans

[Hans de Goede]

Hi,

On 02/02/2012 09:58 AM, Itamar Heim wrote:
> On 02/01/2012 07:04 PM, Brown, Chris (GE Healthcare) wrote:
>> Not a problem I am happy to help in any way that I can.
>> To Clarify I executed the same tests with the same results using both RHEV and ovirt.
>> I am using:
>> RHEV 3.0.1
>> oVirt (FC16 + oVirt Nightly builds)
>
> Hans - is there any downside/conflict if usbtablet is passed when local mouse/agent is used?
> if not, we could pass usbtablet when booting a guest not known to have the agent installed from previous run?

Yes, for most guests (*) this will cause a significant CPU load even when fully idle (USB emulation
causes a 1000 vm-exits / sec). We just need to figure out why the relative mouse mode is working
so poorly for Chris and fix it.

* For guests that don't do USB powermanagement for USB HID devices, which is almost all of them
since many many real world HID devices are just too buggy. In recent Linux versions there are
udev rules to recognize the qemu tablet, and in that case enable USB powermanagement, but AFAIK
all other guests pay a 1000 vm-exits / sec price, which causes a couple of percents of CPU usage
on the host per vm.

Regards,

Hans