On 18/02/16 15:06, Erik Skultety wrote:

When virt-admin is run with valgrind, this kind of output can be obtained:

HEAP SUMMARY: in use at exit: 134,589 bytes in 1,031 blocks total heap usage: 2,667 allocs, 1,636 frees, 496,755 bytes allocated

88 bytes in 1 blocks are definitely lost in loss record 82 of 128 at 0x4C2A9C7: calloc (in /usr/lib64/valgrind/vgpreload_memcheck-amd64-linux.so) by 0x52F6D1F: virAllocVar (viralloc.c:560) by 0x5350268: virObjectNew (virobject.c:193) by 0x53503E0: virObjectLockableNew (virobject.c:219) by 0x4E3BBCB: virAdmConnectNew (datatypes.c:832) by 0x4E38495: virAdmConnectOpen (libvirt-admin.c:209) by 0x10C541: vshAdmConnect (virt-admin.c:107) by 0x10C7B2: vshAdmReconnect (virt-admin.c:163) by 0x10CC7C: cmdConnect (virt-admin.c:298) by 0x110838: vshCommandRun (vsh.c:1224) by 0x10DFD8: main (virt-admin.c:862)

LEAK SUMMARY: definitely lost: 88 bytes in 1 blocks indirectly lost: 0 bytes in 0 blocks possibly lost: 0 bytes in 0 blocks still reachable: 134,501 bytes in 1,030 blocks suppressed: 0 bytes in 0 blocks

This is because virNetClientSetCloseCallback was being reinitialized incorrectly. By resetting the callbacks in a proper way, the leak itself is fixed, however we should still dispose of the client object properly by issuing virObjectUnref on it. --- src/admin/admin_remote.c | 6 +++++- 1 file changed, 5 insertions(+), 1 deletion(-)

diff --git a/src/admin/admin_remote.c b/src/admin/admin_remote.c index e634fa7..e174e88 100644 --- a/src/admin/admin_remote.c +++ b/src/admin/admin_remote.c @@ -180,7 +180,11 @@ remoteAdminConnectClose(virAdmConnectPtr conn) goto done; }

- virNetClientSetCloseCallback(priv->client, NULL, NULL, NULL); + virNetClientSetCloseCallback(priv->client, NULL, conn->closeCallback, + virObjectFreeCallback); + virNetClientClose(priv->client); + virObjectUnref(priv->client); + priv->client = NULL;

rv = 0;

NACK, actually, the unref is already preformed in remoteAdminPrivDispose, so it's not needed here. I'll fix it in v2 Erik