[PATCH v11 11/27] ima: Switch to lazy lsm policy updates for better performance

From: Stefan Berger <stefanb@linux.ibm.com>
To: linux-integrity@vger.kernel.org
Cc: zohar@linux.ibm.com, serge@hallyn.com,
	christian.brauner@ubuntu.com, containers@lists.linux.dev,
	dmitry.kasatkin@gmail.com, ebiederm@xmission.com,
	krzysztof.struczynski@huawei.com, roberto.sassu@huawei.com,
	mpeters@redhat.com, lhinds@redhat.com, lsturman@redhat.com,
	puiterwi@redhat.com, jejb@linux.ibm.com, jamjoom@us.ibm.com,
	linux-kernel@vger.kernel.org, paul@paul-moore.com,
	rgb@redhat.com, linux-security-module@vger.kernel.org,
	jmorris@namei.org, Stefan Berger <stefanb@linux.ibm.com>
Subject: [PATCH v11 11/27] ima: Switch to lazy lsm policy updates for better performance
Date: Wed,  2 Mar 2022 08:46:46 -0500	[thread overview]
Message-ID: <20220302134703.1273041-12-stefanb@linux.ibm.com> (raw)
In-Reply-To: <20220302134703.1273041-1-stefanb@linux.ibm.com>

Instead of calling ima_lsm_update_rules() for every namespace upon
invocation of the ima_lsm_policy_change() notification function,
only set a flag in a namespace and defer the call to
ima_lsm_update_rules() to before the policy is accessed the next time,
which is either in ima_policy_start(), when displaying the policy via
the policy file in securityfs, or when calling ima_match_policy().

The performance numbers before this change for a test enabling
and disabling an SELinux module was as follows with a given number
of IMA namespaces that each a have a policy containing 2 rules
with SELinux labels:

2: ~9s
192: ~11s
1920: ~80s

With this change:

2: ~6.5s
192: ~7s
1920: ~8.3s

Signed-off-by: Stefan Berger <stefanb@linux.ibm.com>
---
 security/integrity/ima/ima.h        |  4 ++++
 security/integrity/ima/ima_policy.c | 15 ++++++++++++++-
 2 files changed, 18 insertions(+), 1 deletion(-)

diff --git a/security/integrity/ima/ima.h b/security/integrity/ima/ima.h
index c68b5117d034..5bf7f080c2be 100644
--- a/security/integrity/ima/ima.h
+++ b/security/integrity/ima/ima.h
@@ -123,6 +123,10 @@ struct ima_h_table {
 };
 
 struct ima_namespace {
+	unsigned long ima_ns_flags;
+/* Bit numbers for above flags; use BIT() to get flag */
+#define IMA_NS_LSM_UPDATE_RULES		0
+
 	/* policy rules */
 	struct list_head ima_default_rules; /* Kconfig, builtin & arch rules */
 	struct list_head ima_policy_rules;  /* arch & custom rules */
diff --git a/security/integrity/ima/ima_policy.c b/security/integrity/ima/ima_policy.c
index 94a41572527a..060d45cf35a7 100644
--- a/security/integrity/ima/ima_policy.c
+++ b/security/integrity/ima/ima_policy.c
@@ -229,6 +229,14 @@ static struct ima_rule_entry critical_data_rules[] __ro_after_init = {
 	{.action = MEASURE, .func = CRITICAL_DATA, .flags = IMA_FUNC},
 };
 
+static void ima_lsm_update_rules(struct ima_namespace *ns);
+
+static inline void ima_lazy_lsm_update_rules(struct ima_namespace *ns)
+{
+	if (test_and_clear_bit(IMA_NS_LSM_UPDATE_RULES, &ns->ima_ns_flags))
+		ima_lsm_update_rules(ns);
+}
+
 static int ima_policy __initdata;
 
 static int __init default_measure_policy_setup(char *str)
@@ -479,7 +487,8 @@ int ima_lsm_policy_change(struct notifier_block *nb, unsigned long event,
 		return NOTIFY_DONE;
 
 	ns = container_of(nb, struct ima_namespace, ima_lsm_policy_notifier);
-	ima_lsm_update_rules(ns);
+
+	set_bit(IMA_NS_LSM_UPDATE_RULES, &ns->ima_ns_flags);
 
 	return NOTIFY_OK;
 }
@@ -706,6 +715,8 @@ int ima_match_policy(struct ima_namespace *ns,
 	if (template_desc && !*template_desc)
 		*template_desc = ima_template_desc_current();
 
+	ima_lazy_lsm_update_rules(ns);
+
 	rcu_read_lock();
 	ima_rules_tmp = rcu_dereference(ns->ima_rules);
 	list_for_each_entry_rcu(entry, ima_rules_tmp, list) {
@@ -1908,6 +1919,8 @@ void *ima_policy_start(struct seq_file *m, loff_t *pos)
 	struct ima_rule_entry *entry;
 	struct list_head *ima_rules_tmp;
 
+	ima_lazy_lsm_update_rules(ns);
+
 	rcu_read_lock();
 	ima_rules_tmp = rcu_dereference(ns->ima_rules);
 	list_for_each_entry_rcu(entry, ima_rules_tmp, list) {
-- 
2.31.1

