On 3/14/19 2:04 PM, Peter Krempa wrote:
On Thu, Mar 14, 2019 at 13:22:35 +0100, Michal Privoznik wrote:
> The aim of this function will be to fix return value of
> qemuMonitorDelDevice() in one specific case. But that is yet to
> come. Right now this is nothing but a plain substitution.
>
> Signed-off-by: Michal Privoznik <mprivozn(a)redhat.com>
> ---
> src/qemu/qemu_hotplug.c | 278 +++++++++++++++-------------------------
> 1 file changed, 103 insertions(+), 175 deletions(-)
>
> diff --git a/src/qemu/qemu_hotplug.c b/src/qemu/qemu_hotplug.c
> index f43f80668c..baa4713cf4 100644
> --- a/src/qemu/qemu_hotplug.c
> +++ b/src/qemu/qemu_hotplug.c
> @@ -67,6 +67,44 @@ VIR_LOG_INIT("qemu.qemu_hotplug");
> unsigned long long qemuDomainRemoveDeviceWaitTime = 1000ull * 5;
>
>
> +/**
> + * qemuDomainDeleteDevice:
> + * @vm: domain object
> + * @alias: device to remove
> + * @enterMonitor: whether do EnterMonitor/ExitMonitor too
I'm not persuaded about usefullnes of this argument.
<snip/>
This means that for the ZPCI extension device does not need this
handling as basically any errors should be ignore. Since the
'enterMonitor' flag results into an utter locking mess in the upcomming
commit which is basically useless you should drop
it, make the ZPCI function ignore error if the device is missing and use
qemuMonitorDelDevice rather than qemuDomainDeleteDevice in it.
Fair enough.
Michal