Skip to contentSkip to conversation

kref: remove kref_set

PATCH In review NeilBrown 4 participants 4 messages Mar 16, 2010

Subsystems (4)
lore
Share & tools
lore

NeilBrown

author
#0 lore raw

PATCH 1/1 kref: remove kref_set

Of the three uses of kref_set in the kernel:

One really should be kref_put as the code is letting go of a
reference,
Two really should be kref_init because the kref is being
initialised.

This suggests that making kref_set available encourages bad code. So fix the three uses and remove kref_set completely.

--

The two 'kref_init' calls in ima_iint.c could equally be a single call after the kmem_cache_alloc in ima_inode_alloc, but this version was the minimal code change.

include/linux/kref.h +0 -1
index b0cb0eb..13003ee 100644
@@ -21,7 +21,6 @@ struct kref {
 	atomic_t refcount;
 };
 
-void kref_set(struct kref *kref, int num);
 void kref_init(struct kref *kref);
 void kref_get(struct kref *kref);
 int kref_put(struct kref *kref, void (*release) (struct kref *kref));
kernel/user_namespace.c +2 -2
index 076c7c8..b2d70d3 100644
@@ -54,8 +54,8 @@ int create_user_ns(struct cred *new)
 #endif
 	/* tgcred will be cleared in our caller bc CLONE_THREAD won't be set */
 
-	/* alloc_uid() incremented the userns refcount.  Just set it to 1 */
-	kref_set(&ns->kref, 1);
+	/* root_user holds a reference to ns, our reference can be dropped */
+	put_user_ns(ns);
 
 	return 0;
 }
lib/kref.c +2 -13
index 9ecd6e8..69761d3 100644
@@ -15,23 +15,13 @@
 #include <linux/module.h>
 
 /**
- * kref_set - initialize object and set refcount to requested number.
- * @kref: object in question.
- * @num: initial reference counter
- */
-void kref_set(struct kref *kref, int num)
-{
-	atomic_set(&kref->refcount, num);
-	smp_mb();
-}
-
-/**
  * kref_init - initialize object.
  * @kref: object in question.
  */
 void kref_init(struct kref *kref)
 {
-	kref_set(kref, 1);
+	atomic_set(&kref->refcount, 1);
+	smp_mb();
 }
 
 /**
@@ -71,7 +61,6 @@ int kref_put(struct kref *kref, void (*release)(struct kref *kref))
 	return 0;
 }
 
-EXPORT_SYMBOL(kref_set);
 EXPORT_SYMBOL(kref_init);
 EXPORT_SYMBOL(kref_get);
 EXPORT_SYMBOL(kref_put);
security/integrity/ima/ima_iint.c +2 -2
index 2d4d05d..91c31d8 100644
@@ -93,7 +93,7 @@ void iint_free(struct kref *kref)
 		       iint->opencount);
 		iint->opencount = 0;
 	}
-	kref_set(&iint->refcount, 1);
+	kref_init(&iint->refcount);
 	kmem_cache_free(iint_cache, iint);
 }
 
@@ -132,7 +132,7 @@ static void init_once(void *foo)
 	iint->readcount = 0;
 	iint->writecount = 0;
 	iint->opencount = 0;
-	kref_set(&iint->refcount, 1);
+	kref_init(&iint->refcount);
 }
 
 static int __init ima_iintcache_init(void)
Acked-by on v1 1/1
#1 lore raw
Quoted 9 lines · NeilBrown

Of the three uses of kref_set in the kernel:

One really should be kref_put as the code is letting go of a
reference,
Two really should be kref_init because the kref is being
initialised.

Quoted 4 lines

This suggests that making kref_set available encourages bad code. So fix the three uses and remove kref_set completely.

No objection to the user_namespace.c hunk - haven't tested it but it certainly had better work with what you have.

thanks,
-serge

Quoted 94 lines

--

The two 'kref_init' calls in ima_iint.c could equally be a single call after the kmem_cache_alloc in ima_inode_alloc, but this version was the minimal code change.

include/linux/kref.h +0 -1
index b0cb0eb..13003ee 100644
@@ -21,7 +21,6 @@ struct kref {
 	atomic_t refcount;
 };
 
-void kref_set(struct kref *kref, int num);
 void kref_init(struct kref *kref);
 void kref_get(struct kref *kref);
 int kref_put(struct kref *kref, void (*release) (struct kref *kref));
kernel/user_namespace.c +2 -2
index 076c7c8..b2d70d3 100644
@@ -54,8 +54,8 @@ int create_user_ns(struct cred *new)
 #endif
 	/* tgcred will be cleared in our caller bc CLONE_THREAD won't be set */
 
-	/* alloc_uid() incremented the userns refcount.  Just set it to 1 */
-	kref_set(&ns->kref, 1);
+	/* root_user holds a reference to ns, our reference can be dropped */
+	put_user_ns(ns);
 
 	return 0;
 }
lib/kref.c +2 -13
index 9ecd6e8..69761d3 100644
@@ -15,23 +15,13 @@
 #include <linux/module.h>
 
 /**
- * kref_set - initialize object and set refcount to requested number.
- * @kref: object in question.
- * @num: initial reference counter
- */
-void kref_set(struct kref *kref, int num)
-{
-	atomic_set(&kref->refcount, num);
-	smp_mb();
-}
-
-/**
  * kref_init - initialize object.
  * @kref: object in question.
  */
 void kref_init(struct kref *kref)
 {
-	kref_set(kref, 1);
+	atomic_set(&kref->refcount, 1);
+	smp_mb();
 }
 
 /**
@@ -71,7 +61,6 @@ int kref_put(struct kref *kref, void (*release)(struct kref *kref))
 	return 0;
 }
 
-EXPORT_SYMBOL(kref_set);
 EXPORT_SYMBOL(kref_init);
 EXPORT_SYMBOL(kref_get);
 EXPORT_SYMBOL(kref_put);
security/integrity/ima/ima_iint.c +2 -2
index 2d4d05d..91c31d8 100644
@@ -93,7 +93,7 @@ void iint_free(struct kref *kref)
 		       iint->opencount);
 		iint->opencount = 0;
 	}
-	kref_set(&iint->refcount, 1);
+	kref_init(&iint->refcount);
 	kmem_cache_free(iint_cache, iint);
 }
 
@@ -132,7 +132,7 @@ static void init_once(void *foo)
 	iint->readcount = 0;
 	iint->writecount = 0;
 	iint->opencount = 0;
-	kref_set(&iint->refcount, 1);
+	kref_init(&iint->refcount);
 }
 
 static int __init ima_iintcache_init(void)
#2 lore raw
In reply to NeilBrown
Quoted 8 lines · NeilBrown

Of the three uses of kref_set in the kernel:

One really should be kref_put as the code is letting go of a
reference,
Two really should be kref_init because the kref is being
initialised.

Quoted 4 lines

This suggests that making kref_set available encourages bad code. So fix the three uses and remove kref_set completely.

Tested the IMA section. Looks good.

Thanks!

Mimi

Quoted 94 lines

--

The two 'kref_init' calls in ima_iint.c could equally be a single call after the kmem_cache_alloc in ima_inode_alloc, but this version was the minimal code change.

include/linux/kref.h +0 -1
index b0cb0eb..13003ee 100644
@@ -21,7 +21,6 @@ struct kref {
 	atomic_t refcount;
 };
 
-void kref_set(struct kref *kref, int num);
 void kref_init(struct kref *kref);
 void kref_get(struct kref *kref);
 int kref_put(struct kref *kref, void (*release) (struct kref *kref));
kernel/user_namespace.c +2 -2
index 076c7c8..b2d70d3 100644
@@ -54,8 +54,8 @@ int create_user_ns(struct cred *new)
 #endif
 	/* tgcred will be cleared in our caller bc CLONE_THREAD won't be set */
 
-	/* alloc_uid() incremented the userns refcount.  Just set it to 1 */
-	kref_set(&ns->kref, 1);
+	/* root_user holds a reference to ns, our reference can be dropped */
+	put_user_ns(ns);
 
 	return 0;
 }
lib/kref.c +2 -13
index 9ecd6e8..69761d3 100644
@@ -15,23 +15,13 @@
 #include <linux/module.h>
 
 /**
- * kref_set - initialize object and set refcount to requested number.
- * @kref: object in question.
- * @num: initial reference counter
- */
-void kref_set(struct kref *kref, int num)
-{
-	atomic_set(&kref->refcount, num);
-	smp_mb();
-}
-
-/**
  * kref_init - initialize object.
  * @kref: object in question.
  */
 void kref_init(struct kref *kref)
 {
-	kref_set(kref, 1);
+	atomic_set(&kref->refcount, 1);
+	smp_mb();
 }
 
 /**
@@ -71,7 +61,6 @@ int kref_put(struct kref *kref, void (*release)(struct kref *kref))
 	return 0;
 }
 
-EXPORT_SYMBOL(kref_set);
 EXPORT_SYMBOL(kref_init);
 EXPORT_SYMBOL(kref_get);
 EXPORT_SYMBOL(kref_put);
security/integrity/ima/ima_iint.c +2 -2
index 2d4d05d..91c31d8 100644
@@ -93,7 +93,7 @@ void iint_free(struct kref *kref)
 		       iint->opencount);
 		iint->opencount = 0;
 	}
-	kref_set(&iint->refcount, 1);
+	kref_init(&iint->refcount);
 	kmem_cache_free(iint_cache, iint);
 }
 
@@ -132,7 +132,7 @@ static void init_once(void *foo)
 	iint->readcount = 0;
 	iint->writecount = 0;
 	iint->opencount = 0;
-	kref_set(&iint->refcount, 1);
+	kref_init(&iint->refcount);
 }
 
 static int __init ima_iintcache_init(void)
#3 lore raw
In reply to NeilBrown
Quoted 11 lines · NeilBrown

Of the three uses of kref_set in the kernel:

One really should be kref_put as the code is letting go of a
reference,
Two really should be kref_init because the kref is being
initialised.

This suggests that making kref_set available encourages bad code. So fix the three uses and remove kref_set completely.

Yeah!

I really didn't like kref_set at all anyway, and I thought that someone "required it" which is why it was added. Oh well, I'm glad it is now gone, thanks so much for this patch.

greg k-h