[PATCH] rcu: avoid passing an argument to the callback function

From: Dipankar Sarma <dipankar@in.ibm.com>

This patch changes the call_rcu() API and avoids passing an argument to the
callback function as suggested by Rusty.  Instead, it is assumed that the
user has embedded the rcu head into a structure that is useful in the
callback and the rcu_head pointer is passed to the callback.  The callback
can use container_of() to get the pointer to its structure and work with
it.  Together with the rcu-singly-link patch, it reduces the rcu_head size
by 50%.  Considering that we use these in things like struct dentry and
struct dst_entry, this is good savings in space.

An example :

struct my_struct {
	struct rcu_head rcu;
	int x;
	int y;
};

void my_rcu_callback(struct rcu_head *head)
{
	struct my_struct *p = container_of(head, struct my_struct, rcu);
	free(p);
}

void my_delete(struct my_struct *p)
{
	...
	call_rcu(&p->rcu, my_rcu_callback);
	...
}
Signed-Off-By: Dipankar Sarma <dipankar@in.ibm.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

arch/ppc64/mm/tlb.c

@@ -158,9 +158,10 @@ void pte_free_now(struct page *ptepage)
 	pte_free(ptepage);
 }
 
-static void pte_free_rcu_callback(void *arg)
+static void pte_free_rcu_callback(struct rcu_head *head)
 {
-	struct pte_freelist_batch *batch = arg;
+	struct pte_freelist_batch *batch =
+		container_of(head, struct pte_freelist_batch, rcu);
 	unsigned int i;
 
 	for (i = 0; i < batch->index; i++)
@@ -171,7 +172,7 @@ static void pte_free_rcu_callback(void *arg)
 void pte_free_submit(struct pte_freelist_batch *batch)
 {
 	INIT_RCU_HEAD(&batch->rcu);
-	call_rcu(&batch->rcu, pte_free_rcu_callback, batch);
+	call_rcu(&batch->rcu, pte_free_rcu_callback);
 }
 
 void pte_free_finish(void)

fs/dcache.c

@@ -65,9 +65,9 @@ struct dentry_stat_t dentry_stat = {
 	.age_limit = 45,
 };
 
-static void d_callback(void *arg)
+static void d_callback(struct rcu_head *head)
 {
-	struct dentry * dentry = (struct dentry *)arg;
+	struct dentry * dentry = container_of(head, struct dentry, d_rcu);
 
 	if (dname_external(dentry))
 		kfree(dentry->d_name.name);
@@ -82,7 +82,7 @@ static void d_free(struct dentry *dentry)
 {
 	if (dentry->d_op && dentry->d_op->d_release)
 		dentry->d_op->d_release(dentry);
- 	call_rcu(&dentry->d_rcu, d_callback, dentry);
+ 	call_rcu(&dentry->d_rcu, d_callback);
 }
 
 /*

include/linux/rcupdate.h

@@ -46,18 +46,16 @@
  * struct rcu_head - callback structure for use with RCU
  * @next: next update requests in a list
  * @func: actual update function to call after the grace period.
- * @arg: argument to be passed to the actual update function.
  */
 struct rcu_head {
 	struct rcu_head *next;
-	void (*func)(void *obj);
-	void *arg;
+	void (*func)(struct rcu_head *head);
 };
 
-#define RCU_HEAD_INIT(head) { .next = NULL, .func = NULL, .arg = NULL }
+#define RCU_HEAD_INIT(head) { .next = NULL, .func = NULL }
 #define RCU_HEAD(head) struct rcu_head head = RCU_HEAD_INIT(head)
 #define INIT_RCU_HEAD(ptr) do { \
-       (ptr)->next = NULL; (ptr)->func = NULL; (ptr)->arg = NULL; \
+       (ptr)->next = NULL; (ptr)->func = NULL; \
 } while (0)
 
 
@@ -144,7 +142,7 @@ extern void rcu_restart_cpu(int cpu);
 
 /* Exported interfaces */
 extern void FASTCALL(call_rcu(struct rcu_head *head, 
-                          void (*func)(void *arg), void *arg));
+				void (*func)(struct rcu_head *head)));
 extern void synchronize_kernel(void);
 
 #endif /* __KERNEL__ */

include/net/dst.h

@@ -183,6 +183,12 @@ static inline void dst_free(struct dst_entry * dst)
 	__dst_free(dst);
 }
 
+static inline void dst_rcu_free(struct rcu_head *head)
+{
+	struct dst_entry *dst = container_of(head, struct dst_entry, rcu_head);
+	dst_free(dst);
+}
+
 static inline void dst_confirm(struct dst_entry *dst)
 {
 	if (dst)

ipc/util.c

@@ -333,25 +333,40 @@ void* ipc_rcu_alloc(int size)
  * Since RCU callback function is called in bh,
  * we need to defer the vfree to schedule_work
  */
-static void ipc_schedule_free(void* arg)
+static void ipc_schedule_free(struct rcu_head *head)
 {
-	struct ipc_rcu_vmalloc *free = arg;
+	struct ipc_rcu_vmalloc *free =
+		container_of(head, struct ipc_rcu_vmalloc, rcu);
 
 	INIT_WORK(&free->work, vfree, free);
 	schedule_work(&free->work);
 }
 
+/**
+ *	ipc_immediate_free	- free ipc + rcu space
+ *
+ *	Free from the RCU callback context
+ *
+ */
+static void ipc_immediate_free(struct rcu_head *head)
+{
+	struct ipc_rcu_kmalloc *free =
+		container_of(head, struct ipc_rcu_kmalloc, rcu);
+	kfree(free);
+}
+
+
+
 void ipc_rcu_free(void* ptr, int size)
 {
 	if (rcu_use_vmalloc(size)) {
 		struct ipc_rcu_vmalloc *free;
 		free = ptr - sizeof(*free);
-		call_rcu(&free->rcu, ipc_schedule_free, free);
+		call_rcu(&free->rcu, ipc_schedule_free);
 	} else {
 		struct ipc_rcu_kmalloc *free;
 		free = ptr - sizeof(*free);
-		/* kfree takes a "const void *" so gcc warns.  So we cast. */
-		call_rcu(&free->rcu, (void (*)(void *))kfree, free);
+		call_rcu(&free->rcu, ipc_immediate_free);
 	}
 
 }

kernel/auditsc.c

@@ -177,9 +177,10 @@ static inline int audit_add_rule(struct audit_entry *entry,
 	return 0;
 }
 
-static void audit_free_rule(void *arg)
+static void audit_free_rule(struct rcu_head *head)
 {
-	kfree(arg);
+	struct audit_entry *e = container_of(head, struct audit_entry, rcu);
+	kfree(e);
 }
 
 /* Note that audit_add_rule and audit_del_rule are called via
@@ -195,7 +196,7 @@ static inline int audit_del_rule(struct audit_rule *rule,
 	list_for_each_entry(e, list, list) {
 		if (!audit_compare_rule(rule, &e->rule)) {
 			list_del_rcu(&e->list);
-			call_rcu(&e->rcu, audit_free_rule, e);
+			call_rcu(&e->rcu, audit_free_rule);
 			return 0;
 		}
 	}

kernel/rcupdate.c

@@ -68,20 +68,19 @@ static DEFINE_PER_CPU(struct tasklet_struct, rcu_tasklet) = {NULL};
  * call_rcu - Queue an RCU update request.
  * @head: structure to be used for queueing the RCU updates.
  * @func: actual update function to be invoked after the grace period
- * @arg: argument to be passed to the update function
  *
  * The update function will be invoked as soon as all CPUs have performed 
  * a context switch or been seen in the idle loop or in a user process. 
  * The read-side of critical section that use call_rcu() for updation must 
  * be protected by rcu_read_lock()/rcu_read_unlock().
  */
-void fastcall call_rcu(struct rcu_head *head, void (*func)(void *arg), void *arg)
+void fastcall call_rcu(struct rcu_head *head,
+				void (*func)(struct rcu_head *rcu))
 {
 	int cpu;
 	unsigned long flags;
 
 	head->func = func;
-	head->arg = arg;
 	head->next = NULL;
 	local_irq_save(flags);
 	cpu = smp_processor_id();
@@ -100,7 +99,7 @@ static void rcu_do_batch(struct rcu_head *list)
 
 	while (list) {
 		next = list->next;
-		list->func(list->arg);
+		list->func(list);
 		list = next;
 	}
 }
@@ -358,11 +357,18 @@ void __init rcu_init(void)
 	register_cpu_notifier(&rcu_nb);
 }
 
+struct rcu_synchronize {
+	struct rcu_head head;
+	struct completion completion;
+};
 
 /* Because of FASTCALL declaration of complete, we use this wrapper */
-static void wakeme_after_rcu(void *completion)
+static void wakeme_after_rcu(struct rcu_head  *head)
 {
-	complete(completion);
+	struct rcu_synchronize *rcu;
+
+	rcu = container_of(head, struct rcu_synchronize, head);
+	complete(&rcu->completion);
 }
 
 /**
@@ -371,14 +377,14 @@ static void wakeme_after_rcu(void *completion)
  */
 void synchronize_kernel(void)
 {
-	struct rcu_head rcu;
-	DECLARE_COMPLETION(completion);
+	struct rcu_synchronize rcu;
 
+	init_completion(&rcu.completion);
 	/* Will wake me after RCU finished */
-	call_rcu(&rcu, wakeme_after_rcu, &completion);
+	call_rcu(&rcu.head, wakeme_after_rcu);
 
 	/* Wait for it */
-	wait_for_completion(&completion);
+	wait_for_completion(&rcu.completion);
 }
 
 

net/bridge/br_if.c

@@ -75,9 +75,8 @@ static int br_initial_port_cost(struct net_device *dev)
 	return 100;	/* assume old 10Mbps */
 }
 
-static void destroy_nbp(void *arg)
+static void destroy_nbp(struct net_bridge_port *p)
 {
-	struct net_bridge_port *p = arg;
 	struct net_device *dev = p->dev;
 
 	dev->br_port = NULL;
@@ -88,6 +87,13 @@ static void destroy_nbp(void *arg)
 	br_sysfs_freeif(p);
 }
 
+static void destroy_nbp_rcu(struct rcu_head *head)
+{
+	struct net_bridge_port *p =
+			container_of(head, struct net_bridge_port, rcu);
+	destroy_nbp(p);
+}
+
 /* called with RTNL */
 static void del_nbp(struct net_bridge_port *p)
 {
@@ -108,7 +114,7 @@ static void del_nbp(struct net_bridge_port *p)
 	del_timer_sync(&p->forward_delay_timer);
 	del_timer_sync(&p->hold_timer);
 	
-	call_rcu(&p->rcu, destroy_nbp, p);
+	call_rcu(&p->rcu, destroy_nbp_rcu);
 }
 
 /* called with RTNL */

net/decnet/dn_route.c

@@ -146,14 +146,14 @@ static __inline__ unsigned dn_hash(unsigned short src, unsigned short dst)
 
 static inline void dnrt_free(struct dn_route *rt)
 {
-	call_rcu(&rt->u.dst.rcu_head, (void (*)(void *))dst_free, &rt->u.dst);
+	call_rcu(&rt->u.dst.rcu_head, dst_rcu_free);
 }
 
 static inline void dnrt_drop(struct dn_route *rt)
 {
 	if (rt)
 		dst_release(&rt->u.dst);
-	call_rcu(&rt->u.dst.rcu_head, (void (*)(void *))dst_free, &rt->u.dst);
+	call_rcu(&rt->u.dst.rcu_head, dst_rcu_free);
 }
 
 static void dn_dst_check_expire(unsigned long dummy)

net/ipv4/route.c

@@ -439,13 +439,13 @@ static struct file_operations rt_cpu_seq_fops = {
   
 static __inline__ void rt_free(struct rtable *rt)
 {
-	call_rcu(&rt->u.dst.rcu_head, (void (*)(void *))dst_free, &rt->u.dst);
+	call_rcu(&rt->u.dst.rcu_head, dst_rcu_free);
 }
 
 static __inline__ void rt_drop(struct rtable *rt)
 {
 	ip_rt_put(rt);
-	call_rcu(&rt->u.dst.rcu_head, (void (*)(void *))dst_free, &rt->u.dst);
+	call_rcu(&rt->u.dst.rcu_head, dst_rcu_free);
 }
 
 static __inline__ int rt_fast_clean(struct rtable *rth)

net/sched/sch_generic.c

@@ -411,9 +411,9 @@ void qdisc_reset(struct Qdisc *qdisc)
 /* this is the rcu callback function to clean up a qdisc when there 
  * are no further references to it */
 
-static void __qdisc_destroy (void * arg) 
+static void __qdisc_destroy(struct rcu_head *head)
 {
-	struct Qdisc    *qdisc = (struct Qdisc *) arg;
+	struct Qdisc *qdisc = container_of(head, struct Qdisc, q_rcu);
 	struct Qdisc_ops  *ops = qdisc->ops;
 
 #ifdef CONFIG_NET_ESTIMATOR
@@ -448,7 +448,7 @@ void qdisc_destroy(struct Qdisc *qdisc)
 		}
 	}
 
-	call_rcu(&qdisc->q_rcu, __qdisc_destroy, qdisc);
+	call_rcu(&qdisc->q_rcu, __qdisc_destroy);
 
 }
 

security/selinux/netif.c

@@ -134,9 +134,9 @@ struct sel_netif *sel_netif_lookup(struct net_device *dev)
 	return netif;
 }
 
-static void sel_netif_free(void *p)
+static void sel_netif_free(struct rcu_head *p)
 {
-	struct sel_netif *netif = p;
+	struct sel_netif *netif = container_of(p, struct sel_netif, rcu_head);
 	
 	DEBUGP("%s: %s\n", __FUNCTION__, netif->nsec.dev->name);
 	kfree(netif);
@@ -151,7 +151,7 @@ static void sel_netif_destroy(struct sel_netif *netif)
 	sel_netif_total--;
 	spin_unlock_bh(&sel_netif_lock);
 
-	call_rcu(&netif->rcu_head, sel_netif_free, netif);
+	call_rcu(&netif->rcu_head, sel_netif_free);
 }
 
 void sel_netif_put(struct sel_netif *netif)