Re: [RESEND][PATCH] multifd/tls: fix memoryleak of the QIOChannelSocket object when canceling migration

["Daniel P. Berrangé" <berrange@redhat.com>]

On Wed, Nov 11, 2020 at 03:14:18PM +0800, Chuan Zheng wrote:
> When creating new tls client, the tioc->master will be referenced, we need dereferenced
> it after tls handshake.
> 
> Signed-off-by: Chuan Zheng <zhengchuan@huawei.com>
> ---
>  migration/multifd.c | 4 +++-
>  1 file changed, 3 insertions(+), 1 deletion(-)
> 
> diff --git a/migration/multifd.c b/migration/multifd.c
> index 68b171f..df76a8e 100644
> --- a/migration/multifd.c
> +++ b/migration/multifd.c
> @@ -728,7 +728,8 @@ static void multifd_tls_outgoing_handshake(QIOTask *task,
>                                             gpointer opaque)
>  {
>      MultiFDSendParams *p = opaque;
> -    QIOChannel *ioc = QIO_CHANNEL(qio_task_get_source(task));
> +    QIOChannelTLS *tioc = QIO_CHANNEL_TLS(qio_task_get_source(task));
> +    QIOChannel *ioc = QIO_CHANNEL(tioc);
>      Error *err = NULL;
>  
>      if (qio_task_propagate_error(task, &err)) {
> @@ -737,6 +738,7 @@ static void multifd_tls_outgoing_handshake(QIOTask *task,
>          trace_multifd_tls_outgoing_handshake_complete(ioc);
>      }
>      multifd_channel_connect(p, ioc, err);
> +    object_unref(OBJECT(tioc->master));
>  }

I think this should be done earlier, at the time where we wrap the channel.
eg in multifd_tls_channel_connect(), immediately after the call to
migration_tls_client_create, as that's where the double reference starts
from.


Regards,
Daniel
-- 
|: 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 :|