On 10/27/2014 09:51 AM, Matthias Gatto wrote:
On Sat, Oct 25, 2014 at 6:30 AM, Eric Blake <eblake(a)redhat.com>
wrote:
> On 10/07/2014 05:14 AM, Matthias Gatto wrote:
>> Detect if the the qemu binary currently in use suport the bps_max option,
s/suport/supports/
>> If yes add it to the command, if not, just ignore the
options.
>
> Ignoring options is wrong. If a user explicitly requested something in
> XML but the hypervisor cannot honor it, we want to error out.
>
> --
> Eric Blake eblake redhat com +1-919-301-3266
> Libvirt virtualization library
http://libvirt.org
>
But I've already check this in qemu_driver.c
We can check it twice, but will it be really useful ?
We have to check in two places: when constructing the command line
(qemu_command.c) and when hotplugging a new value (qemu_driver.c calling
into qemu_monitor.c). In general, if the user has requested the XML,
but qemu can't support it, then we must error out.
I haven't fully read the series to see if you adequately filtered things
out earlier; but if the filter happened earlier, then a comment in the
code pointing to the other location that did the filtering would help
(not just the commit message).
Maybe I should write the check has already been made in the commit
message ?
--
Eric Blake eblake redhat com +1-919-301-3266
Libvirt virtualization library
http://libvirt.org