[PATCH 2/3] LSM: SafeSetID: Stop releasing uninitialized ruleset

Tyler Hicks tyhicks at canonical.com
Wed Sep 25 21:43:53 UTC 2019

From: Micah Morton <mortonm at chromium.org>

BugLink: https://launchpad.net/bugs/1845391

The first time a rule set is configured for SafeSetID, we shouldn't be
trying to release the previously configured ruleset, since there isn't
one. Currently, the pointer that would point to a previously configured
ruleset is uninitialized on first rule set configuration, leading to a
crash when we try to call release_ruleset with that pointer.

Acked-by: Jann Horn <jannh at google.com>
Signed-off-by: Micah Morton <mortonm at chromium.org>

(cherry picked from commit 21ab8580b383f27b7f59b84ac1699cb26d6c3d69)
Signed-off-by: Tyler Hicks <tyhicks at canonical.com>
 security/safesetid/securityfs.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/security/safesetid/securityfs.c b/security/safesetid/securityfs.c
index d568e17dd773..74a13d432ed8 100644
--- a/security/safesetid/securityfs.c
+++ b/security/safesetid/securityfs.c
@@ -187,7 +187,8 @@ static ssize_t handle_policy_update(struct file *file,
-	release_ruleset(pol);
+	if (pol)
+                release_ruleset(pol);
 	return err;

More information about the kernel-team mailing list