Re: [libvirt] PATCH: 8/25: Concurrent dispatch of RPC methods

On Thu, Jan 15, 2009 at 05:07:01PM +0000, Richard W.M. Jones wrote: > On Tue, Jan 13, 2009 at 05:41:48PM +0000, Daniel P. Berrange wrote: > > + /* Get a unique serial number for this message. */ > > + rv->serial = priv->counter++; > > This (and similar) are safe because we hold the remoteDriverLock until > we enter the actual call? That is correct - the first thing all the public API entry points do is grab the lock. The lock is only then released, when the main thread is waiting on I/O in the poll() call: + remoteDriverUnlock(priv); + + repoll: + ret = poll(fds, ARRAY_CARDINALITY(fds), -1); + if (ret < 0 && errno == EINTR) + goto repoll; + remoteDriverLock(priv); Or when additional threads are sleeping on the condition variable.