qemu_agent: Bring back single sync

Historically, we had no idea whether the qemu-ga running inside
the guest was running or not. Or whether it crashed in the middle
of reading of a command. That's why we issued guest-sync prior
any intended command, to make the agent flush any partially read
JSON and reset its state machine.

But with VSERPORT_CHANGE event we know when the guest agent
(dis-)connects and thus can issue the sync command just once for
each 'connection'. Whether the agent is synced is tracked in
agent->inSync member, which used to be set to true upon
successful sync. But after rework in v8.0.0-rc1~361 that line is
gone, leaving us with using the historic approach basically.

Fixes: cad84fd51eaac5e3bfdf441f9986e1f2639a0828
Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>

diff --git a/src/qemu/qemu_agent.c b/src/qemu/qemu_agent.c
index 0d77a2f90d9ccdf907384bc0843f6ee4534847eb..94782f2bc9ad29e47846e937b772179035e0ba82 100644 (file)
--- a/src/qemu/qemu_agent.c
+++ b/src/qemu/qemu_agent.c
@@ -869,16 +869,20 @@ qemuAgentGuestSync(qemuAgent *agent)
         return -1;
 
     /* successfully sync'd */
-    if (rc == 1)
+    if (rc == 1) {
+        agent->inSync = true;
         return 0;
+    }
 
     /* send another sync */
     if ((rc = qemuAgentGuestSyncSend(agent, timeout, false)) < 0)
         return -1;
 
     /* successfully sync'd */
-    if (rc == 1)
+    if (rc == 1) {
+        agent->inSync = true;
         return 0;
+    }
 
     if (agent->running)
         virReportError(VIR_ERR_INTERNAL_ERROR, "%s",