Don't sleep in poll() if there is existing SASL decoded data

In the SASL codepath we typically read far more data off the
wire than we immediately need. When using a connection from a
single thread this isn't a problem, since only our reply will
be pending (or an event we can handle directly). When using a
connection from multiple threads though, we may read the data
from replies from other threads. If those replies occur after
our own reply, they'll not be processed. The other thread will
then go into poll() and wait for its reply which has already
been received and decoded. The solution is to set poll() timeout
to 0 if there is pending SASL data.

* src/remote/remote_driver.c: Don't sleep in poll() if SASL
  data exists

src/remote/remote_driver.c

@@ -10180,6 +10180,14 @@ remoteIOEventLoop(virConnectPtr conn,
 #ifdef HAVE_PTHREAD_SIGMASK
         sigset_t oldmask, blockedsigs;
 #endif
+        int timeout = -1;
+
+        /* If we have existing SASL decoded data we
+         * don't want to sleep in the poll(), just
+         * check if any other FDs are also ready
+         */
+        if (priv->saslDecoded)
+            timeout = 0;
 
         fds[0].events = fds[0].revents = 0;
         fds[1].events = fds[1].revents = 0;
@@ -10215,7 +10223,7 @@ remoteIOEventLoop(virConnectPtr conn,
 #endif
 
     repoll:
-        ret = poll(fds, ARRAY_CARDINALITY(fds), -1);
+        ret = poll(fds, ARRAY_CARDINALITY(fds), timeout);
         if (ret < 0 && errno == EAGAIN)
             goto repoll;
 
@@ -10225,6 +10233,12 @@ remoteIOEventLoop(virConnectPtr conn,
 
         remoteDriverLock(priv);
 
+        /* If we have existing SASL decoded data, pretend
+         * the socket became readable so we consume it
+         */
+        if (priv->saslDecoded)
+            fds[0].revents |= POLLIN;
+
         if (fds[1].revents) {
             ssize_t s;
             DEBUG0("Woken up from poll by other thread");