[Qemu-devel] [PATCH for-2.12] iothread: workaround glib bug which hangs qmp-test

[Peter Xu <peterx@redhat.com>]

Free the AIO context earlier than the GMainContext (if we have) to
workaround a possible Glib bug.  No functional change at all.

We encountered a qmp-test hang with oob:

  #0  0x00007f35ffe45334 in __lll_lock_wait () from /lib64/libpthread.so.0
  #1  0x00007f35ffe405d8 in _L_lock_854 () from /lib64/libpthread.so.0
  #2  0x00007f35ffe404a7 in pthread_mutex_lock () from /lib64/libpthread.so.0
  #3  0x00007f35fc5b9c9d in g_source_unref_internal (source=0x24f0600, context=0x7f35f0000960, have_lock=0) at gmain.c:1685
  #4  0x0000000000aa6672 in aio_context_unref (ctx=0x24f0600) at /root/qemu/util/async.c:497
  #5  0x000000000065851c in iothread_instance_finalize (obj=0x24f0380) at /root/qemu/iothread.c:129
  #6  0x0000000000962d79 in object_deinit (obj=0x24f0380, type=0x242e960) at /root/qemu/qom/object.c:462
  #7  0x0000000000962e0d in object_finalize (data=0x24f0380) at /root/qemu/qom/object.c:476
  #8  0x0000000000964146 in object_unref (obj=0x24f0380) at /root/qemu/qom/object.c:924
  #9  0x0000000000965880 in object_finalize_child_property (obj=0x24ec640, name=0x24efca0 "mon_iothread", opaque=0x24f0380) at /root/qemu/qom/object.c:1436
  #10 0x0000000000962c33 in object_property_del_child (obj=0x24ec640, child=0x24f0380, errp=0x0) at /root/qemu/qom/object.c:436
  #11 0x0000000000962d26 in object_unparent (obj=0x24f0380) at /root/qemu/qom/object.c:455
  #12 0x0000000000658f00 in iothread_destroy (iothread=0x24f0380) at /root/qemu/iothread.c:365
  #13 0x00000000004c67a8 in monitor_cleanup () at /root/qemu/monitor.c:4663
  #14 0x0000000000669e27 in main (argc=16, argv=0x7ffc8b1ae2f8, envp=0x7ffc8b1ae380) at /root/qemu/vl.c:4749

With glib version 2.28.8-9 (current default version on centos6) we might
encounter above with the old code. It is verified that glib version
2.50.3-3 won't trigger that bug again, but since we are still supporting
glib 2.28.8-9, we may want this workaround.

Signed-off-by: Peter Xu <peterx@redhat.com>
---
 iothread.c | 18 ++++++++++++++----
 1 file changed, 14 insertions(+), 4 deletions(-)

diff --git a/iothread.c b/iothread.c
index e675c38442..d41d661cdf 100644
--- a/iothread.c
+++ b/iothread.c
@@ -117,16 +117,26 @@ static void iothread_instance_finalize(Object *obj)
     IOThread *iothread = IOTHREAD(obj);
 
     iothread_stop(iothread);
+    /*
+     * With glib version 2.28.8-9 (current default version on centos6)
+     * we might encounter problem of qmp-test OOB hang if we unref the
+     * AIO context later than the GMainContext below.  Let's free the
+     * AIO context earlier to bypass that possible glib bug.
+     *
+     * It is verified that glib version 2.50.3-3 (or even earlier)
+     * won't trigger that bug again, but since we are still supporting
+     * glib 2.28.8-9, we need this workaround.
+     */
+    if (iothread->ctx) {
+        aio_context_unref(iothread->ctx);
+        iothread->ctx = NULL;
+    }
     if (iothread->worker_context) {
         g_main_context_unref(iothread->worker_context);
         iothread->worker_context = NULL;
     }
     qemu_cond_destroy(&iothread->init_done_cond);
     qemu_mutex_destroy(&iothread->init_done_lock);
-    if (!iothread->ctx) {
-        return;
-    }
-    aio_context_unref(iothread->ctx);
 }
 
 static void iothread_complete(UserCreatable *obj, Error **errp)
-- 
2.14.3