[PATCH] selinux: skip bounded transition processing if the policy isn't loaded

[Paul Moore]

From: Paul Moore <***@paul-moore.com>

We can't do anything reasonable in security_bounded_transition() if we
don't have a policy loaded, and in fact we could run into problems
with some of the code inside expecting a policy. Fix these problems
like we do many others in security/selinux/ss/services.c by checking
to see if the policy is loaded (ss_initialized) and returning quickly
if it isn't.

Reported-by: syzbot <syzkaller-***@googlegroups.com>
Signed-off-by: Paul Moore <***@paul-moore.com>
Acked-by: Stephen Smalley <***@tycho.nsa.gov>
---
security/selinux/ss/services.c | 3 +++
1 file changed, 3 insertions(+)

diff --git a/security/selinux/ss/services.c b/security/selinux/ss/services.c
index d05496deb229..8900ea5cbabf 100644
--- a/security/selinux/ss/services.c
+++ b/security/selinux/ss/services.c
@@ -867,6 +867,9 @@ int security_bounded_transition(u32 old_sid, u32 new_sid)
int index;
int rc;

+ if (!ss_initialized)
+ return 0;
+
read_lock(&policy_rwlock);

rc = -EINVAL;

[James Morris]

Post by Paul Moore
We can't do anything reasonable in security_bounded_transition() if we
don't have a policy loaded, and in fact we could run into problems
with some of the code inside expecting a policy. Fix these problems
like we do many others in security/selinux/ss/services.c by checking
to see if the policy is loaded (ss_initialized) and returning quickly
if it isn't.

Reviewed-by: James Morris <***@oracle.com>

[Paul Moore]

Post by Paul Moore
We can't do anything reasonable in security_bounded_transition() if we
don't have a policy loaded, and in fact we could run into problems
with some of the code inside expecting a policy. Fix these problems
like we do many others in security/selinux/ss/services.c by checking
to see if the policy is loaded (ss_initialized) and returning quickly
if it isn't.

Thanks. I just merged the patch to selinux/next.