Multicast packet loss

diff --git a/include/linux/interrupt.h b/include/linux/interrupt.h
index 2306920..b79a207 100644
--- a/include/linux/interrupt.h
+++ b/include/linux/interrupt.h
@@ -276,6 +276,24 @@  extern void FASTCALL(raise_softirq_irqoff(unsigned int
nr));
 extern void FASTCALL(raise_softirq(unsigned int nr));


+/*
+ * softirq delayed works : should be delayed at do_softirq() end
+ */
+struct softirq_delay {
+	struct softirq_delay	*next;
+	void 			(*func)(struct softirq_delay *);
+};
+
+int softirq_delay_queue(struct softirq_delay *sdel);
+
+static inline void softirq_delay_init(struct softirq_delay *sdel,
+				      void (*func)(struct softirq_delay *))
+{
+	sdel->next = NULL;
+	sdel->func = func;
+}
+
+
 /* Tasklets --- multithreaded analogue of BHs.

    Main feature differing them of generic softirqs: tasklet
diff --git a/include/linux/irqflags.h b/include/linux/irqflags.h
index 412e025..f7b48a1 100644
--- a/include/linux/irqflags.h
+++ b/include/linux/irqflags.h
@@ -11,19 +11,21 @@ 
 #ifndef _LINUX_TRACE_IRQFLAGS_H
 #define _LINUX_TRACE_IRQFLAGS_H

+#define softirq_enter()	do { current->softirq_context++; } while (0)
+#define softirq_exit()	do { current->softirq_context--; } while (0)
+#define softirq_context(p)	((p)->softirq_context)
+#define running_from_softirq()  (softirq_context(current) > 0)
+
 #ifdef CONFIG_TRACE_IRQFLAGS
   extern void trace_hardirqs_on(void);
   extern void trace_hardirqs_off(void);
   extern void trace_softirqs_on(unsigned long ip);
   extern void trace_softirqs_off(unsigned long ip);
 # define trace_hardirq_context(p)	((p)->hardirq_context)
-# define trace_softirq_context(p)	((p)->softirq_context)
 # define trace_hardirqs_enabled(p)	((p)->hardirqs_enabled)
 # define trace_softirqs_enabled(p)	((p)->softirqs_enabled)
 # define trace_hardirq_enter()	do { current->hardirq_context++; } while
(0)
 # define trace_hardirq_exit()	do { current->hardirq_context--; } while (0)
-# define trace_softirq_enter()	do { current->softirq_context++; } while
(0)
-# define trace_softirq_exit()	do { current->softirq_context--; } while (0)
 # define INIT_TRACE_IRQFLAGS	.softirqs_enabled = 1,
 #else
 # define trace_hardirqs_on()		do { } while (0)
@@ -31,13 +33,10 @@ 
 # define trace_softirqs_on(ip)		do { } while (0)
 # define trace_softirqs_off(ip)		do { } while (0)
 # define trace_hardirq_context(p)	0
-# define trace_softirq_context(p)	0
 # define trace_hardirqs_enabled(p)	0
 # define trace_softirqs_enabled(p)	0
 # define trace_hardirq_enter()		do { } while (0)
 # define trace_hardirq_exit()		do { } while (0)
-# define trace_softirq_enter()		do { } while (0)
-# define trace_softirq_exit()		do { } while (0)
 # define INIT_TRACE_IRQFLAGS
 #endif

diff --git a/include/linux/net.h b/include/linux/net.h
index 596131e..9e762e9 100644
--- a/include/linux/net.h
+++ b/include/linux/net.h
@@ -115,12 +115,16 @@  enum sock_shutdown_cmd {
 struct socket {
 	socket_state		state;
 	unsigned long		flags;
-	const struct proto_ops	*ops;
+	/*
+	 * Please keep fasync_list & wait fields in the same cache line
+	 */
 	struct fasync_struct	*fasync_list;
+	wait_queue_head_t	wait;
+
 	struct file		*file;
 	struct sock		*sk;
-	wait_queue_head_t	wait;
 	short			type;
+	const struct proto_ops	*ops;
 };

 struct vm_area_struct;
diff --git a/include/linux/sched.h b/include/linux/sched.h
index dc2f4fa..bf0ff49 100644
--- a/include/linux/sched.h
+++ b/include/linux/sched.h
@@ -1111,8 +1111,8 @@  struct task_struct {
 	unsigned long softirq_enable_ip;
 	unsigned int softirq_enable_event;
 	int hardirq_context;
-	int softirq_context;
 #endif
+	int softirq_context;
 #ifdef CONFIG_LOCKDEP
 # define MAX_LOCK_DEPTH 30UL
 	u64 curr_chain_key;
diff --git a/include/net/sock.h b/include/net/sock.h
index 6e1542d..fb0f719 100644
--- a/include/net/sock.h
+++ b/include/net/sock.h
@@ -236,6 +236,7 @@  struct sock {
 	unsigned long	        sk_lingertime;
 	struct sk_buff_head	sk_error_queue;
 	struct proto		*sk_prot_creator;
+	struct softirq_delay	sk_delay;
 	rwlock_t		sk_callback_lock;
 	int			sk_err,
 				sk_err_soft;
@@ -859,6 +860,7 @@  extern void *sock_kmalloc(struct sock *sk, int size,
 			  gfp_t priority);
 extern void sock_kfree_s(struct sock *sk, void *mem, int size);
 extern void sk_send_sigurg(struct sock *sk);
+extern void inet_def_readable(struct sock *sk, int len);

 /*
  * Functions to fill in entries in struct proto_ops when a protocol
diff --git a/include/net/udplite.h b/include/net/udplite.h
index 635b0ea..1589817 100644
--- a/include/net/udplite.h
+++ b/include/net/udplite.h
@@ -28,6 +28,7 @@  static __inline__ int udplite_getfrag(void *from, char
*to, int  offset,
 /* Designate sk as UDP-Lite socket */
 static inline int udplite_sk_init(struct sock *sk)
 {
+	sk->sk_data_ready = inet_def_readable;
 	udp_sk(sk)->pcflag = UDPLITE_BIT;
 	return 0;
 }
diff --git a/kernel/lockdep.c b/kernel/lockdep.c
index e2c07ec..decb1f7 100644
--- a/kernel/lockdep.c
+++ b/kernel/lockdep.c
@@ -1643,7 +1643,7 @@  print_usage_bug(struct task_struct *curr, struct
held_lock *this,
 	printk("%s/%d [HC%u[%lu]:SC%u[%lu]:HE%u:SE%u] takes:\n",
 		curr->comm, task_pid_nr(curr),
 		trace_hardirq_context(curr), hardirq_count() >> HARDIRQ_SHIFT,
-		trace_softirq_context(curr), softirq_count() >> SOFTIRQ_SHIFT,
+		softirq_context(curr), softirq_count() >> SOFTIRQ_SHIFT,
 		trace_hardirqs_enabled(curr),
 		trace_softirqs_enabled(curr));
 	print_lock(this);
diff --git a/kernel/softirq.c b/kernel/softirq.c
index bd89bc4..fb116ac 100644
--- a/kernel/softirq.c
+++ b/kernel/softirq.c
@@ -194,6 +194,42 @@  void local_bh_enable_ip(unsigned long ip)
 }
 EXPORT_SYMBOL(local_bh_enable_ip);

+
+#define SOFTIRQ_DELAY_END (struct softirq_delay *)1L
+static DEFINE_PER_CPU(struct softirq_delay *, softirq_delay_head) = {
+	SOFTIRQ_DELAY_END
+};
+
+/*
+ * Caller must disable preemption, and take care of appropriate
+ * locking and refcounting
+ */
+int softirq_delay_queue(struct softirq_delay *sdel)
+{
+	if (!sdel->next) {
+		sdel->next = __get_cpu_var(softirq_delay_head);
+		__get_cpu_var(softirq_delay_head) = sdel;
+		return 1;
+	}
+	return 0;
+}
+
+/*
+ * Because locking is provided by subsystem, please note
+ * that sdel->func(sdel) is responsible for setting sdel->next to NULL
+ */
+static void softirq_delay_exec(void)
+{
+	struct softirq_delay *sdel;
+
+	while ((sdel = __get_cpu_var(softirq_delay_head)) != SOFTIRQ_DELAY_END) {
+		__get_cpu_var(softirq_delay_head) = sdel->next;
+		sdel->func(sdel);	/*	sdel->next = NULL;*/
+		}
+}
+
+
+
 /*
  * We restart softirq processing MAX_SOFTIRQ_RESTART times,
  * and we fall back to softirqd after that.
@@ -216,7 +252,7 @@  asmlinkage void __do_softirq(void)
 	account_system_vtime(current);

 	__local_bh_disable((unsigned long)__builtin_return_address(0));
-	trace_softirq_enter();
+	softirq_enter();

 	cpu = smp_processor_id();
 restart:
@@ -236,6 +272,8 @@  restart:
 		pending >>= 1;
 	} while (pending);

+	softirq_delay_exec();
+
 	local_irq_disable();

 	pending = local_softirq_pending();
@@ -245,7 +283,7 @@  restart:
 	if (pending)
 		wakeup_softirqd();

-	trace_softirq_exit();
+	softirq_exit();

 	account_system_vtime(current);
 	_local_bh_enable();
diff --git a/lib/locking-selftest.c b/lib/locking-selftest.c
index 280332c..1aa7351 100644
--- a/lib/locking-selftest.c
+++ b/lib/locking-selftest.c
@@ -157,11 +157,11 @@  static void init_shared_classes(void)
 #define SOFTIRQ_ENTER()				\
 		local_bh_disable();		\
 		local_irq_disable();		\
-		trace_softirq_enter();		\
+		softirq_enter();		\
 		WARN_ON(!in_softirq());

 #define SOFTIRQ_EXIT()				\
-		trace_softirq_exit();		\
+		softirq_exit();		\
 		local_irq_enable();		\
 		local_bh_enable();

diff --git a/net/core/sock.c b/net/core/sock.c
index c519b43..cb70343 100644
--- a/net/core/sock.c
+++ b/net/core/sock.c
@@ -213,6 +213,8 @@  __u32 sysctl_rmem_default __read_mostly = SK_RMEM_MAX;
 /* Maximal space eaten by iovec or ancilliary data plus some space */
 int sysctl_optmem_max __read_mostly = sizeof(unsigned
long)*(2*UIO_MAXIOV+512);

+static void sock_readable_defer(struct softirq_delay *sdel);
+
 static int sock_set_timeout(long *timeo_p, char __user *optval, int
optlen)
 {
 	struct timeval tv;
@@ -996,6 +998,7 @@  struct sock *sk_clone(const struct sock *sk, const
gfp_t priority)
 #endif

 		rwlock_init(&newsk->sk_dst_lock);
+		softirq_delay_init(&newsk->sk_delay, sock_readable_defer);
 		rwlock_init(&newsk->sk_callback_lock);
 		lockdep_set_class_and_name(&newsk->sk_callback_lock,
 				af_callback_keys + newsk->sk_family,
@@ -1509,6 +1512,45 @@  static void sock_def_readable(struct sock *sk, int
len)
 	read_unlock(&sk->sk_callback_lock);
 }

+/*
+ * helper function called by softirq_delay_exec(),
+ * if inet_def_readable() queued us.
+ */
+static void sock_readable_defer(struct softirq_delay *sdel)
+{
+	struct sock *sk = container_of(sdel, struct sock, sk_delay);
+
+	sdel->next = NULL;
+	/*
+	 * At this point, we dont own a lock on socket, only a reference.
+	 * We must commit above write, or another cpu could miss a wakeup
+	 */
+	smp_wmb();
+	sock_def_readable(sk, 0);
+	sock_put(sk);
+}
+
+/*
+ * Custom version of sock_def_readable()
+ * We want to defer scheduler processing at the end of do_softirq()
+ * Called with socket locked.
+ */
+void inet_def_readable(struct sock *sk, int len)
+{
+	if (running_from_softirq()) {
+		if (softirq_delay_queue(&sk->sk_delay))
+			/*
+			 * If we queued this socket, take a reference on it
+			 * Caller owns socket lock, so write to sk_delay.next
+			 * will be committed before unlock.
+			 */
+			sock_hold(sk);
+	} else
+		sock_def_readable(sk, len);
+}
+
+EXPORT_SYMBOL(inet_def_readable);
+
 static void sock_def_write_space(struct sock *sk)
 {
 	read_lock(&sk->sk_callback_lock);
@@ -1586,6 +1628,7 @@  void sock_init_data(struct socket *sock, struct sock
*sk)
 		sk->sk_sleep	=	NULL;

 	rwlock_init(&sk->sk_dst_lock);
+	softirq_delay_init(&sk->sk_delay, sock_readable_defer);
 	rwlock_init(&sk->sk_callback_lock);
 	lockdep_set_class_and_name(&sk->sk_callback_lock,
 			af_callback_keys + sk->sk_family,
diff --git a/net/ipv4/udp.c b/net/ipv4/udp.c
index 03c400c..cfeb051 100644
--- a/net/ipv4/udp.c
+++ b/net/ipv4/udp.c
@@ -1226,6 +1226,12 @@  int udp_destroy_sock(struct sock *sk)
 	return 0;
 }

+static int udp_init_sock(struct sock *sk)
+{
+	sk->sk_data_ready = inet_def_readable;
+	return 0;
+}
+
 /*
  *	Socket option code for UDP
  */
@@ -1439,6 +1445,7 @@  struct proto udp_prot = {
 	.connect	   = ip4_datagram_connect,
 	.disconnect	   = udp_disconnect,
 	.ioctl		   = udp_ioctl,
+	.init		   = udp_init_sock,
 	.destroy	   = udp_destroy_sock,
 	.setsockopt	   = udp_setsockopt,
 	.getsockopt	   = udp_getsockopt,
diff --git a/net/ipv6/udp.c b/net/ipv6/udp.c
index ee1cc3f..fa9ce73 100644
--- a/net/ipv6/udp.c
+++ b/net/ipv6/udp.c
@@ -856,6 +856,12 @@  int udpv6_destroy_sock(struct sock *sk)
 	return 0;
 }

+static int udpv6_init_sock(struct sock *sk)
+{
+	sk->sk_data_ready = inet_def_readable;
+	return 0;
+}
+
 /*
  *	Socket option code for UDP
  */
@@ -979,6 +985,7 @@  struct proto udpv6_prot = {
 	.connect	   = ip6_datagram_connect,
 	.disconnect	   = udp_disconnect,
 	.ioctl		   = udp_ioctl,
+	.init 		   = udpv6_init_sock,
 	.destroy	   = udpv6_destroy_sock,
 	.setsockopt	   = udpv6_setsockopt,
 	.getsockopt	   = udpv6_getsockopt,