[libvirt] [PATCH v2 0/4] Couple of seclabels improvements

[John Ferlan]

On 07/10/2014 10:04 AM, Michal Privoznik wrote:
> diff to v1:
> - rework the 3rd patch
> - introduce one more bugfix
> 
> Michal Privoznik (4):
>   virSecurityLabelDef: substitute 'norelabel' with 'relabel'
>   virSecurityDeviceLabelDef: substitute 'norelabel' with 'relabel'
>   conf: Always format seclabel's model
>   conf: Don't allow multiple seclabels for same model
> 
>  src/conf/domain_conf.c                             | 67 ++++++++++++----------
>  src/security/security_apparmor.c                   | 10 ++--
>  src/security/security_dac.c                        | 22 +++----
>  src/security/security_manager.c                    |  2 +-
>  src/security/security_selinux.c                    | 32 +++++------
>  src/util/virseclabel.c                             |  2 +-
>  src/util/virseclabel.h                             |  4 +-
>  .../qemuxml2argv-seclabel-dynamic-none.xml         | 28 +++++++++
>  .../qemuxml2argv-seclabel-multiple.xml             | 40 +++++++++++++
>  tests/qemuxml2argvtest.c                           |  1 +
>  tests/qemuxml2xmltest.c                            |  1 +
>  11 files changed, 142 insertions(+), 67 deletions(-)
>  create mode 100644 tests/qemuxml2argvdata/qemuxml2argv-seclabel-dynamic-none.xml
>  create mode 100644 tests/qemuxml2argvdata/qemuxml2argv-seclabel-multiple.xml
> 

There's a Coverity issue from these patches - it looks like perhaps
patch 1&2 were combined when submitted into commit id '13adf1b' which has:

virSecurityLabelDefPtr
 virSecurityLabelDefNew(const char *model)
 {
     virSecurityLabelDefPtr seclabel = NULL;

     if (VIR_ALLOC(seclabel) < 0 ||
         VIR_STRDUP(seclabel->model, model) < 0) {
         virSecurityLabelDefFree(seclabel);
         seclabel = NULL;
     }

+    seclabel->relabel = true;
+
     return seclabel;
 }


See the problem at all?  It's a FORWARD_NULL on 'seclabel'.

John