Re: [PATCH v1 2/2] qemu_process: continue to process fakereboot after restarting libvirtd

Hi, Michal. Can you take the time to help me revivew the '[PATCH v1 1/2] qemu_process: set fakereboot flags false after processing fakereboot over'? Best wishes! Bihong On 2021/11/10 21:41, Michal Prívozník wrote: > On 11/10/21 2:25 PM, Bihong Yu wrote: >> >> >> On 2021/11/10 17:32, Michal Prívozník wrote: >>> On 10/25/21 11:04 AM, Bihong Yu wrote: >>>> During the vm rebooting, the vm could be paused if the libvirtd is >>>> restarted for some reason, which is not expected. We need continue >>>> fakereboot process if fakereboot flags is true and the vm is in >>>> paused-user status. >>>> >>>> Signed-off-by: Bihong Yu <yubihong(a)huawei.com&gt; >>>> --- >>>> src/qemu/qemu_process.c | 10 ++++++---- >>>> 1 file changed, 6 insertions(+), 4 deletions(-) >>>> >>>> diff --git a/src/qemu/qemu_process.c b/src/qemu/qemu_process.c >>>> index 832ce164fb..a758b96fa6 100644 >>>> --- a/src/qemu/qemu_process.c >>>> +++ b/src/qemu/qemu_process.c >>>> @@ -8742,13 +8742,15 @@ qemuProcessReconnect(void *opaque) >>>> goto error; >>>> } >>>> >>>> - /* In case the domain shutdown while we were not running, >>>> - * we need to finish the shutdown process. And we need to do it after >>>> - * we have virQEMUCaps filled in. >>>> + /* In case the domain shutdown or fake reboot while we were not running, >>>> + * we need to finish the shutdown or fake reboot process. And we need to >>>> + * do it after we have virQEMUCaps filled in. >>>> */ >>>> if (state == VIR_DOMAIN_SHUTDOWN || >>>> (state == VIR_DOMAIN_PAUSED && >>>> - reason == VIR_DOMAIN_PAUSED_SHUTTING_DOWN)) { >>>> + reason == VIR_DOMAIN_PAUSED_SHUTTING_DOWN) || >>>> + (priv->fakeReboot && state == VIR_DOMAIN_PAUSED && >>>> + reason == VIR_DOMAIN_PAUSED_USER)) { >>> >>> One thing that I don't quite understand is why this new condition checks >>> for state or reason. I could understand the reason a bit (because domain >>> is paused after SHUTDOWN event), but the reason? Can you elaborate please? >> >> Hi, Michal. Thank you for your reply. >> The reason is that: while libvirt reboot vm with ACPI mode, the vm will >> undergo the following state changes: >> running -> shutdown -> prelaunch -> running >> If libvirtd reboot after vm prelaunch status and before vm running status, >> the qemuProcessUpdateState() will update the vm status to 'VIR_DOMAIN_PAUSED' >> and reason to 'VIR_DOMAIN_PAUSED_USER' according to the qemuMonitorGetStatus() >> returning result in qemuProcessReconnect(). >> So, we need to recognize this scenario and continue to finish rebooting >> process. > > Ah, you are correct. I've missed that. > > Reviewed-by: Michal Privoznik <mprivozn(a)redhat.com&gt; > > Michal > > . >