On 07/31/2010 01:52 AM, Daniel Veillard wrote:
@@ -1416,7 +1416,6 @@ qemuConnectMonitor(struct qemud_driver *driver, virDomainObjPtr vm) ret = qemuMonitorSetCapabilities(priv->mon); qemuDomainObjExitMonitorWithDriver(driver, vm);
- ret = 0; error: if (ret < 0) qemuMonitorClose(priv->mon);
Hum, if we do this we change the behaviour in case of errors in qemuMonitorSetCapabilities(), I wonder if this wasn't there to avoid failing in that case, and if this was the case then it's the first affectation we need to remove. I would wait until this gets clarified (qemuMonitorSetCapabilities failure may not be a blocking factor to starting a guest ...)
This bug was was introduced in commit e72cc3c, with dwalsh's patch on May 27. We were doing the correct thing before then, so only 0.8.2 contains a problem where failure to connect to the monitor is undetected.
okidoc, ACK, sorry for the burden :-)
No problem - it was worth it to make me trawl through 'git gui blame', since I learned more about how to effectively use it. Now pushed. -- Eric Blake eblake@redhat.com +1-801-349-2682 Libvirt virtualization library http://libvirt.org