Skip to contentSkip to conversation

qom-qmp-cmds: Fix another memory leak in qmp_object_add()

PATCH In review Markus Armbruster 4 participants 5 messages Mar 17, 2020

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 qom-qmp-cmds: Fix another memory leak in qmp_object_add()

When user_creatable_add_type() fails, qmp_object_add() returns both its error and the usual empty QDict success value. The QMP core handles the error, and ignores the success value, leaking it. Exposed by qmp-cmd-test case /x86_64/qmp/object-add-without-props, and duly reported both by ASan and valgrind.

To plug the leak, set the success value only on success.

Fixes 5f07c4d60d091320186e7b0edaf9ed2cc16b2d1e
Cc Kevin Wolf <[email protected]>
---
 qom/qom-qmp-cmds.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
qom/qom-qmp-cmds.c +1 -1
index 435193b036..6bd137ccbf 100644
@@ -287,8 +287,8 @@ void qmp_object_add(QDict *qdict, QObject **ret_data, Error **errp)
     visit_free(v);
     if (obj) {
         object_unref(obj);
+        *ret_data = QOBJECT(qdict_new());
     }
-    *ret_data = QOBJECT(qdict_new());
 }
 
 void qmp_object_del(const char *id, Error **errp)
Signature · 1 line
2.21.1
Reviewed-by on v1 1/1
#1 lore raw
Quoted 7 lines · Markus Armbruster

When user_creatable_add_type() fails, qmp_object_add() returns both its error and the usual empty QDict success value. The QMP core handles the error, and ignores the success value, leaking it. Exposed by qmp-cmd-test case /x86_64/qmp/object-add-without-props, and duly reported both by ASan and valgrind.

Quoted 5 lines

To plug the leak, set the success value only on success.

Fixes 5f07c4d60d091320186e7b0edaf9ed2cc16b2d1e
Cc Kevin Wolf <[email protected]>
Quoted 22 lines
---
 qom/qom-qmp-cmds.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
qom/qom-qmp-cmds.c +1 -2
index 435193b036..6bd137ccbf 100644
@@ -287,8 +287,8 @@ void qmp_object_add(QDict *qdict, QObject **ret_data, Error **errp)
     visit_free(v);
     if (obj) {
         object_unref(obj);
+        *ret_data = QOBJECT(qdict_new());
     }
-    *ret_data = QOBJECT(qdict_new());
 }
 
 void qmp_object_del(const char *id, Error **errp)
--

2.21.1

Signature · 1 line
Marc-André Lureau
Reviewed-by on v1 1/1
#2 lore raw
In reply to Markus Armbruster
Quoted 6 lines · Markus Armbruster

When user_creatable_add_type() fails, qmp_object_add() returns both its error and the usual empty QDict success value. The QMP core handles the error, and ignores the success value, leaking it. Exposed by qmp-cmd-test case /x86_64/qmp/object-add-without-props, and duly reported both by ASan and valgrind.

Quoted 8 lines

To plug the leak, set the success value only on success.

Fixes 5f07c4d60d091320186e7b0edaf9ed2cc16b2d1e
Cc Kevin Wolf <[email protected]>
---
 qom/qom-qmp-cmds.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
Quoted 19 lines
qom/qom-qmp-cmds.c +1 -1
index 435193b036..6bd137ccbf 100644
@@ -287,8 +287,8 @@ void qmp_object_add(QDict *qdict, QObject **ret_data, Error **errp)
     visit_free(v);
     if (obj) {
         object_unref(obj);
+        *ret_data = QOBJECT(qdict_new());
     }
-    *ret_data = QOBJECT(qdict_new());
 }
 
 void qmp_object_del(const char *id, Error **errp)
Signature · 1 line
2.21.1

Regards,
Daniel

Signature · 3 lines
|: https://berrange.com      -o-    https://www.flickr.com/photos/dberrange :|
|: https://libvirt.org         -o-            https://fstop138.berrange.com :|
|: https://entangle-photo.org    -o-    https://www.instagram.com/dberrange :|
#3 lore raw
In reply to Markus Armbruster
Quoted 17 lines

-----Original Message-----
From: Qemu-devel [mailto:qemu-devel- [email protected]] On Behalf Of Markus Armbruster
Sent: Tuesday, March 17, 2020 5:23 PM
To: [email protected]
Cc: Kevin Wolf <[email protected]>; [email protected]; [email protected]; [email protected]
Subject: [PATCH] qom-qmp-cmds: Fix another memory leak in qmp_object_add()

When user_creatable_add_type() fails, qmp_object_add() returns both its error and the usual empty QDict success value. The QMP core handles the error, and ignores the success value, leaking it. Exposed by qmp-cmd-test case /x86_64/qmp/object-add-without-props, and duly reported both by ASan and valgrind.

Quoted 6 lines

To plug the leak, set the success value only on success.

Fixes 5f07c4d60d091320186e7b0edaf9ed2cc16b2d1e
Cc Kevin Wolf <[email protected]>

---

Hi, Markus

Looks like the same patch that has been reported already here:
https://lists.gnu.org/archive/html/qemu-devel/2020-03/msg03928.html

Maybe we should initialize ret_data in xen-block to avoid a possible uninitialized error ?

Thanks.

Quoted 21 lines

qom/qom-qmp-cmds.c | 2 +-

1 file changed, 1 insertion(+), 1 deletion(-)
qom/qom-qmp-cmds.c index renamed from qom/qom-qmp-cmds.c +0 -0

435193b036..6bd137ccbf 100644

qom/qom-qmp-cmds.c +1 -1
@@ -287,8 +287,8 @@ void qmp_object_add(QDict *qdict, QObject
 **ret_data, Error **errp)
    visit_free(v);
    if (obj) {
        object_unref(obj);
+        *ret_data = QOBJECT(qdict_new());
    }
-    *ret_data = QOBJECT(qdict_new());
 }
 

void qmp_object_del(const char *id, Error **errp)
--
2.21.1

#4 lore raw
Quoted 19 lines · Chenqun (kuhn)
Quoted 19 lines

-----Original Message-----
From: Qemu-devel [mailto:qemu-devel- [email protected]] On Behalf Of Markus Armbruster
Sent: Tuesday, March 17, 2020 5:23 PM
To: [email protected]
Cc: Kevin Wolf <[email protected]>; [email protected]; [email protected]; [email protected]
Subject: [PATCH] qom-qmp-cmds: Fix another memory leak in qmp_object_add()

When user_creatable_add_type() fails, qmp_object_add() returns both its error and the usual empty QDict success value. The QMP core handles the error, and ignores the success value, leaking it. Exposed by qmp-cmd-test case /x86_64/qmp/object-add-without-props, and duly reported both by ASan and valgrind.

To plug the leak, set the success value only on success.

Quoted 8 lines
Quoted 4 lines
Fixes 5f07c4d60d091320186e7b0edaf9ed2cc16b2d1e
Cc Kevin Wolf <[email protected]>

---

Hi, Markus

Looks like the same patch that has been reported already here:
https://lists.gnu.org/archive/html/qemu-devel/2020-03/msg03928.html

That patch has an additional hunk:

  diff --git a/hw/block/xen-block.c b/hw/block/xen-block.c
  index 3885464513..041866b846 100644
  --- a/hw/block/xen-block.c
  +++ b/hw/block/xen-block.c
  @@ -860,7 +860,7 @@ static XenBlockIOThread *xen_block_iothread_create(const 
  char *id,
       XenBlockIOThread *iothread = g_new(XenBlockIOThread, 1);
       Error *local_err = NULL;
       QDict *opts;
  -    QObject *ret_data;
  +    QObject *ret_data = NULL;
       iothread->id = g_strdup(id);

       opts = qdict_new();
       qdict_put_str(opts, "qom-type", TYPE_IOTHREAD);
       qdict_put_str(opts, "id", id);
       qmp_object_add(opts, &ret_data, &local_err);
       qobject_unref(opts);
       qobject_unref(ret_data);

       if (local_err) {
           error_propagate(errp, local_err);

           g_free(iothread->id);
           g_free(iothread);
           return NULL;
       }

       return iothread;
Quoted 1 line

Maybe we should initialize ret_data in xen-block to avoid a possible uninitialized error ?

Yes, because xen_block_iothread_create() passes @ret_data to qobject_unref() even after qmp_object_add() failed.

In short, Pan Nengyuan's patch is more complete.