ucounts: In set_cred_ucounts assume new->ucounts is non-NULL
authorEric W. Biederman <ebiederm@xmission.com>
Sat, 16 Oct 2021 17:51:58 +0000 (12:51 -0500)
committerEric W. Biederman <ebiederm@xmission.com>
Wed, 20 Oct 2021 15:45:34 +0000 (10:45 -0500)
Any cred that is destined for use by commit_creds must have a non-NULL
cred->ucounts field.  Only curing credential construction is a NULL
cred->ucounts valid.  Only abort_creds, put_cred, and put_cred_rcu
needs to deal with a cred with a NULL ucount.  As set_cred_ucounts is
non of those case don't confuse people by handling something that can
not happen.

Link: https://lkml.kernel.org/r/871r4irzds.fsf_-_@disp2133
Tested-by: Yu Zhao <yuzhao@google.com>
Reviewed-by: Alexey Gladkov <legion@kernel.org>
Signed-off-by: "Eric W. Biederman" <ebiederm@xmission.com>
kernel/cred.c

index 1ae0b4948a5a820d4ea3f758c733418dcb3f286f..473d17c431f3a8e7dc10cf66c15b8ff526bf551d 100644 (file)
@@ -676,15 +676,14 @@ int set_cred_ucounts(struct cred *new)
         * This optimization is needed because alloc_ucounts() uses locks
         * for table lookups.
         */
-       if (old_ucounts && old_ucounts->ns == new->user_ns && uid_eq(old_ucounts->uid, new->euid))
+       if (old_ucounts->ns == new->user_ns && uid_eq(old_ucounts->uid, new->euid))
                return 0;
 
        if (!(new_ucounts = alloc_ucounts(new->user_ns, new->euid)))
                return -EAGAIN;
 
        new->ucounts = new_ucounts;
-       if (old_ucounts)
-               put_ucounts(old_ucounts);
+       put_ucounts(old_ucounts);
 
        return 0;
 }