[libvirt] [PATCH] virnetdaemon: Don't deadlock when talking to D-Bus
https://bugzilla.redhat.com/show_bug.cgi?id=1487322 In ace45e67abbd I've tried to fix a problem that we get the reply to a D-Bus call while we were sleeping. In that case the callback was never set. So I've changed the code that the callback is called directly in this case. However, I hadn't realized that since callback is called out of order it lock the virNetDaemon. Exactly the very same virNetDaemon that we are dealing with right now and have it locked already (in virNetDaemonAddShutdownInhibition()) Signed-off-by: Michal Privoznik <mprivozn@redhat.com> --- src/rpc/virnetdaemon.c | 21 +++++++++++++++------ 1 file changed, 15 insertions(+), 6 deletions(-) diff --git a/src/rpc/virnetdaemon.c b/src/rpc/virnetdaemon.c index 00247cfc3..e3b9390af 100644 --- a/src/rpc/virnetdaemon.c +++ b/src/rpc/virnetdaemon.c @@ -439,14 +439,12 @@ virNetDaemonAutoShutdown(virNetDaemonPtr dmn, #if defined(WITH_DBUS) && defined(DBUS_TYPE_UNIX_FD) static void -virNetDaemonGotInhibitReply(DBusPendingCall *pending, - void *opaque) +virNetDaemonGotInhibitReplyLocked(DBusPendingCall *pending, + virNetDaemonPtr dmn) { - virNetDaemonPtr dmn = opaque; DBusMessage *reply; int fd; - virObjectLock(dmn); dmn->autoShutdownCallingInhibit = false; VIR_DEBUG("dmn=%p", dmn); @@ -470,11 +468,22 @@ virNetDaemonGotInhibitReply(DBusPendingCall *pending, virDBusMessageUnref(reply); cleanup: - virObjectUnlock(dmn); dbus_pending_call_unref(pending); } +static void +virNetDaemonGotInhibitReply(DBusPendingCall *pending, + void *opaque) +{ + virNetDaemonPtr dmn = opaque; + + virObjectLock(dmn); + virNetDaemonGotInhibitReplyLocked(pending, dmn); + virObjectUnlock(dmn); +} + + /* As per: http://www.freedesktop.org/wiki/Software/systemd/inhibit */ static void virNetDaemonCallInhibit(virNetDaemonPtr dmn, @@ -516,7 +525,7 @@ virNetDaemonCallInhibit(virNetDaemonPtr dmn, 25 * 1000) && pendingReply) { if (dbus_pending_call_get_completed(pendingReply)) { - virNetDaemonGotInhibitReply(pendingReply, dmn); + virNetDaemonGotInhibitReplyLocked(pendingReply, dmn); } else { dbus_pending_call_set_notify(pendingReply, virNetDaemonGotInhibitReply, -- 2.13.5
On Fri, Sep 1, 2017 at 10:39 AM, Michal Privoznik <mprivozn@redhat.com> wrote:
https://bugzilla.redhat.com/show_bug.cgi?id=1487322
In ace45e67abbd I've tried to fix a problem that we get the reply to a D-Bus call while we were sleeping. In that case the callback was never set. So I've changed the code that the callback is called directly in this case. However, I hadn't realized that since callback is called out of order it lock the virNetDaemon. Exactly the very same virNetDaemon that we are dealing with right now and have it locked already (in virNetDaemonAddShutdownInhibition())
Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
Change builds fine (on top of 3.6) and seems to fix the issue. Survived 20 minutes in my stress loop, which it never did before. Tested-by: Christian Ehrhardt <christian.ehrhardt@canonical.com>
On Fri, Sep 01, 2017 at 10:39:10AM +0200, Michal Privoznik wrote:
https://bugzilla.redhat.com/show_bug.cgi?id=1487322
In ace45e67abbd I've tried to fix a problem that we get the reply
s/I've/I --> the point in the past is known ;)
to a D-Bus call while we were sleeping. In that case the callback was never set. So I've changed the code that the callback is
s/I've/I --> again, we know exactly when :)
called directly in this case. However, I hadn't realized that since callback is called out of order it lock the virNetDaemon.
s/since callback/since the callback s/lock/locks s/virNetDaemon/virNetDaemon object also I'm not sure about 'hadn't realized' being used in this specific case, but I'd rather not object to that.
Exactly the very same virNetDaemon that we are dealing with right now and have it locked already (in
s/have/having or 'that we have already locked' (preferred)
virNetDaemonAddShutdownInhibition())
Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
ACK Erik
participants (3)
-
Christian Ehrhardt -
Erik Skultety -
Michal Privoznik