On 11/24/2010 03:08 PM, Matthias Bolte wrote:
2010/11/24 Eric Blake <eblake(a)redhat.com>:
> * src/storage/storage_backend.c
> (virStorageBackendUpdateVolTargetInfoFD): Avoid leak.
> ---
>
> Unlikely to bite in real life, but still a leak.
>
> src/storage/storage_backend.c | 2 +-
> 1 files changed, 1 insertions(+), 1 deletions(-)
>
> diff --git a/src/storage/storage_backend.c b/src/storage/storage_backend.c
> index a6ee564..10ea33c 100644
> --- a/src/storage/storage_backend.c
> +++ b/src/storage/storage_backend.c
> @@ -1148,11 +1148,11 @@ virStorageBackendUpdateVolTargetInfoFD(virStorageVolTargetPtr
target,
> }
> } else {
> target->perms.label = strdup(filecon);
> + freecon(filecon);
> if (target->perms.label == NULL) {
> virReportOOMError();
> return -1;
> }
> - freecon(filecon);
> }
> #else
> target->perms.label = NULL;
> --
> 1.7.3.2
>
ACK.
I've pushed the series through 5/n, although I may continue to extend it
further as I review more valgrind logs (by the way, libnl is a beast for
how frequently it leaks small strings).
Meanwhile, as discussed with Matthew on IRC, I squashed 4 and 5 together
before pushing, since 5 had the better commit message and those are the
only two files to use security_context_t.
--
Eric Blake eblake(a)redhat.com +1-801-349-2682
Libvirt virtualization library
http://libvirt.org