Skip to contentSkip to conversation

locking/lockdep: Test no new string literal is created in lockdep_set_subclass()

PATCH In review Ahmed Ehab 2 participants 2 messages Sep 4, 2024 – Sep 10, 2024

Subsystems (2)
lore
Share & tools
lore

Ahmed Ehab

author
#0 lore raw

PATCH v8 2/2 locking/lockdep: Test no new string literal is created in lockdep_set_subclass()

Add a test case to ensure that no new name string literal will be created in lockdep_set_subclass(), otherwise a warning will be triggered in look_up_lock_class(). Add this to catch the problem in the future.

---
 lib/locking-selftest.c | 40 ++++++++++++++++++++++++++++++++++++++++
 1 file changed, 40 insertions(+)
lib/locking-selftest.c +40 -0
index 6f6a5fc85b42..0783ee97c971 100644
@@ -2710,6 +2710,44 @@ static void local_lock_3B(void)
 
 }
 
+#if CONFIG_DEBUG_LOCK_ALLOC
+static inline const char *rw_semaphore_lockdep_name(struct rw_semaphore *rwsem)
+{
+	return rwsem->dep_map.name;
+}
+#else
+static inline const char *rw_semaphore_lockdep_name(struct rw_semaphore *rwsem)
+{
+	return NULL;
+}
+#endif
+
+static void lock_class_subclass_X1(void)
+{
+	const char *name_before_setting_subclass = rw_semaphore_lockdep_name(&rwsem_X1);
+	const char *name_after_setting_subclass;
+
+	lockdep_set_subclass(&rwsem_X1, 1);
+	name_after_setting_subclass = rw_semaphore_lockdep_name(&rwsem_X1);
+	DEBUG_LOCKS_WARN_ON(name_before_setting_subclass != name_after_setting_subclass);
+}
+
+/*
+ * after setting the subclass the lockdep_map.name changes
+ * if we initialize a new string literal for the subclass
+ * we will have a new name pointer
+ */
+static void class_subclass_X1_name_test(void)
+{
+	printk("  --------------------------------------------------------------------------\n");
+	printk("  | class and subclass name test|\n");
+	printk("  ---------------------\n");
+
+	print_testname("lock class and subclass same name");
+	dotest(lock_class_subclass_X1, SUCCESS, LOCKTYPE_RWSEM);
+	pr_cont("\n");
+}
+
 static void local_lock_tests(void)
 {
 	printk("  --------------------------------------------------------------------------\n");
@@ -2920,6 +2958,8 @@ void locking_selftest(void)
 	dotest(hardirq_deadlock_softirq_not_deadlock, FAILURE, LOCKTYPE_SPECIAL);
 	pr_cont("\n");
 
+	class_subclass_X1_name_test();
+
 	if (unexpected_testcase_failures) {
 		printk("-----------------------------------------------------------------\n");
 		debug_locks = 0;
Signature · 1 line
2.46.0
#1 lore raw

Hi Ahmed,

Quoted 4 lines · Ahmed Ehab

Add a test case to ensure that no new name string literal will be created in lockdep_set_subclass(), otherwise a warning will be triggered in look_up_lock_class(). Add this to catch the problem in the future.

This overall looks good to me, I'm going to take it and create a PR for tip in next release. However, please note a few things:

* Since you only send one of the patch from your original series, you should avoid use "2/2" in the title, because it could be confusing whether there is "1/2" lost in sending. If you want to make sure people aware that this is a continued work of the patch #2 in your original series, you can put some description after the following "---"

* You need also to put changes between patch versions after "---" so that people can know the context, for example, I have no idea why you send a v8 after v7 and what's the delta here. Here is an example of how to document the delta:

https://lore.kernel.org/rust-for-linux/[email protected]/

Regards,
Boqun

Quoted 66 lines
---
 lib/locking-selftest.c | 40 ++++++++++++++++++++++++++++++++++++++++
 1 file changed, 40 insertions(+)
lib/locking-selftest.c +40 -0
index 6f6a5fc85b42..0783ee97c971 100644
@@ -2710,6 +2710,44 @@ static void local_lock_3B(void)
 
 }
 
+#if CONFIG_DEBUG_LOCK_ALLOC
+static inline const char *rw_semaphore_lockdep_name(struct rw_semaphore *rwsem)
+{
+	return rwsem->dep_map.name;
+}
+#else
+static inline const char *rw_semaphore_lockdep_name(struct rw_semaphore *rwsem)
+{
+	return NULL;
+}
+#endif
+
+static void lock_class_subclass_X1(void)
+{
+	const char *name_before_setting_subclass = rw_semaphore_lockdep_name(&rwsem_X1);
+	const char *name_after_setting_subclass;
+
+	lockdep_set_subclass(&rwsem_X1, 1);
+	name_after_setting_subclass = rw_semaphore_lockdep_name(&rwsem_X1);
+	DEBUG_LOCKS_WARN_ON(name_before_setting_subclass != name_after_setting_subclass);
+}
+
+/*
+ * after setting the subclass the lockdep_map.name changes
+ * if we initialize a new string literal for the subclass
+ * we will have a new name pointer
+ */
+static void class_subclass_X1_name_test(void)
+{
+	printk("  --------------------------------------------------------------------------\n");
+	printk("  | class and subclass name test|\n");
+	printk("  ---------------------\n");
+
+	print_testname("lock class and subclass same name");
+	dotest(lock_class_subclass_X1, SUCCESS, LOCKTYPE_RWSEM);
+	pr_cont("\n");
+}
+
 static void local_lock_tests(void)
 {
 	printk("  --------------------------------------------------------------------------\n");
@@ -2920,6 +2958,8 @@ void locking_selftest(void)
 	dotest(hardirq_deadlock_softirq_not_deadlock, FAILURE, LOCKTYPE_SPECIAL);
 	pr_cont("\n");
 
+	class_subclass_X1_name_test();
+
 	if (unexpected_testcase_failures) {
 		printk("-----------------------------------------------------------------\n");
 		debug_locks = 0;
Signature · 1 line
2.46.0