Skip to contentSkip to conversation

locking/lockdep: Remove unused variable in __lockdep_count*()

PATCH In review wuchi 2 participants 3 messages Oct 30, 2022 – Oct 31, 2022 Locking primitives

lore
Share & tools
lore

wuchi

author
#0 lore raw

PATCH 1/1 locking/lockdep: Remove unused variable in __lockdep_count*()

The target_entry variable will never be referenced because the function noop_count() always returns false in __bfs(). So just remove that.

---
 kernel/locking/lockdep.c | 6 ++----
 1 file changed, 2 insertions(+), 4 deletions(-)
kernel/locking/lockdep.c +2 -4
index e3375bc40dad..04d2ce5d0215 100644
@@ -2075,9 +2075,8 @@ static bool noop_count(struct lock_list *entry, void *data)
 static unsigned long __lockdep_count_forward_deps(struct lock_list *this)
 {
 	unsigned long  count = 0;
-	struct lock_list *target_entry;
 
-	__bfs_forwards(this, (void *)&count, noop_count, NULL, &target_entry);
+	__bfs_forwards(this, (void *)&count, noop_count, NULL, NULL);
 
 	return count;
 }
@@ -2100,9 +2099,8 @@ unsigned long lockdep_count_forward_deps(struct lock_class *class)
 static unsigned long __lockdep_count_backward_deps(struct lock_list *this)
 {
 	unsigned long  count = 0;
-	struct lock_list *target_entry;
 
-	__bfs_backwards(this, (void *)&count, noop_count, NULL, &target_entry);
+	__bfs_backwards(this, (void *)&count, noop_count, NULL, NULL);
 
 	return count;
 }
Signature · 1 line
2.20.1
reviewer
#1 lore raw
Quoted 25 lines · wuchi

The target_entry variable will never be referenced because the function noop_count() always returns false in __bfs(). So just remove that.

---
  kernel/locking/lockdep.c | 6 ++----
  1 file changed, 2 insertions(+), 4 deletions(-)
kernel/locking/lockdep.c +1 -2
index e3375bc40dad..04d2ce5d0215 100644
@@ -2075,9 +2075,8 @@ static bool noop_count(struct lock_list *entry, void *data)
  static unsigned long __lockdep_count_forward_deps(struct lock_list *this)
  {
  	unsigned long  count = 0;
-	struct lock_list *target_entry;
 
-	__bfs_forwards(this, (void *)&count, noop_count, NULL, &target_entry);
+	__bfs_forwards(this, (void *)&count, noop_count, NULL, NULL);
 
  	return count;
  }

kernel/locking/lockdep.c unsigned long lockdep_count_forward_deps(struct lock_class *class) · line 2106

Quoted 10 lines
kernel/locking/lockdep.c +1 -2
@@ -2100,9 +2099,8 @@ unsigned long lockdep_count_forward_deps(struct lock_class *class)
  static unsigned long __lockdep_count_backward_deps(struct lock_list *this)
  {
  	unsigned long  count = 0;
-	struct lock_list *target_entry;
 
-	__bfs_backwards(this, (void *)&count, noop_count, NULL, &target_entry);
+	__bfs_backwards(this, (void *)&count, noop_count, NULL, NULL);
 
  	return count;
  }

There is no target_entry NULL check in __bfs(), so target_entry is always expected to point to a valid memory location. You will need to add the NULL check if you want to remove it from these two functions.

BTW, have you actually exercised the code to make sure that there was no unexpected side effect from this change?

Cheers,
Longman

author
#2 lore raw
Quoted 5 lines · Waiman Long <[email protected]> 于2022年10月30日周日 23:21写道:

There is no target_entry NULL check in __bfs(), so target_entry is always expected to point to a valid memory location. You will need to add the NULL check if you want to remove it from these two functions.

Yes, add code in __bfs() as follow?

if (match(lock, data)) {
    if (target_entry)
        *target_entry = lock;
     return BFS_RMATCH;
}
Quoted 2 lines

BTW, have you actually exercised the code to make sure that there was no unexpected side effect from this change?

Yes, the code "check_irq_usage -> __bfs_backwards" is an example and the primary callers lockdep_stats_show and l_show in lockdep_proc.c work well.

But I'm not sure I missed anything.

Quoted 4 lines

Cheers,
Longman

Thank you for reply