[PATCH] Notifier chain update: API changes

The kernel's implementation of notifier chains is unsafe.  There is no
protection against entries being added to or removed from a chain while the
chain is in use.  The issues were discussed in this thread:

    http://marc.theaimsgroup.com/?l=linux-kernel&m=113018709002036&w=2

We noticed that notifier chains in the kernel fall into two basic usage
classes:

	"Blocking" chains are always called from a process context
	and the callout routines are allowed to sleep;

	"Atomic" chains can be called from an atomic context and
	the callout routines are not allowed to sleep.

We decided to codify this distinction and make it part of the API.  Therefore
this set of patches introduces three new, parallel APIs: one for blocking
notifiers, one for atomic notifiers, and one for "raw" notifiers (which is
really just the old API under a new name).  New kinds of data structures are
used for the heads of the chains, and new routines are defined for
registration, unregistration, and calling a chain.  The three APIs are
explained in include/linux/notifier.h and their implementation is in
kernel/sys.c.

With atomic and blocking chains, the implementation guarantees that the chain
links will not be corrupted and that chain callers will not get messed up by
entries being added or removed.  For raw chains the implementation provides no
guarantees at all; users of this API must provide their own protections.  (The
idea was that situations may come up where the assumptions of the atomic and
blocking APIs are not appropriate, so it should be possible for users to
handle these things in their own way.)

There are some limitations, which should not be too hard to live with.  For
atomic/blocking chains, registration and unregistration must always be done in
a process context since the chain is protected by a mutex/rwsem.  Also, a
callout routine for a non-raw chain must not try to register or unregister
entries on its own chain.  (This did happen in a couple of places and the code
had to be changed to avoid it.)

Since atomic chains may be called from within an NMI handler, they cannot use
spinlocks for synchronization.  Instead we use RCU.  The overhead falls almost
entirely in the unregister routine, which is okay since unregistration is much
less frequent that calling a chain.

Here is the list of chains that we adjusted and their classifications.  None
of them use the raw API, so for the moment it is only a placeholder.

  ATOMIC CHAINS
  -------------
arch/i386/kernel/traps.c:		i386die_chain
arch/ia64/kernel/traps.c:		ia64die_chain
arch/powerpc/kernel/traps.c:		powerpc_die_chain
arch/sparc64/kernel/traps.c:		sparc64die_chain
arch/x86_64/kernel/traps.c:		die_chain
drivers/char/ipmi/ipmi_si_intf.c:	xaction_notifier_list
kernel/panic.c:				panic_notifier_list
kernel/profile.c:			task_free_notifier
net/bluetooth/hci_core.c:		hci_notifier
net/ipv4/netfilter/ip_conntrack_core.c:	ip_conntrack_chain
net/ipv4/netfilter/ip_conntrack_core.c:	ip_conntrack_expect_chain
net/ipv6/addrconf.c:			inet6addr_chain
net/netfilter/nf_conntrack_core.c:	nf_conntrack_chain
net/netfilter/nf_conntrack_core.c:	nf_conntrack_expect_chain
net/netlink/af_netlink.c:		netlink_chain

  BLOCKING CHAINS
  ---------------
arch/powerpc/platforms/pseries/reconfig.c:	pSeries_reconfig_chain
arch/s390/kernel/process.c:		idle_chain
arch/x86_64/kernel/process.c		idle_notifier
drivers/base/memory.c:			memory_chain
drivers/cpufreq/cpufreq.c		cpufreq_policy_notifier_list
drivers/cpufreq/cpufreq.c		cpufreq_transition_notifier_list
drivers/macintosh/adb.c:		adb_client_list
drivers/macintosh/via-pmu.c		sleep_notifier_list
drivers/macintosh/via-pmu68k.c		sleep_notifier_list
drivers/macintosh/windfarm_core.c	wf_client_list
drivers/usb/core/notify.c		usb_notifier_list
drivers/video/fbmem.c			fb_notifier_list
kernel/cpu.c				cpu_chain
kernel/module.c				module_notify_list
kernel/profile.c			munmap_notifier
kernel/profile.c			task_exit_notifier
kernel/sys.c				reboot_notifier_list
net/core/dev.c				netdev_chain
net/decnet/dn_dev.c:			dnaddr_chain
net/ipv4/devinet.c:			inetaddr_chain

It's possible that some of these classifications are wrong.  If they are,
please let us know or submit a patch to fix them.  Note that any chain that
gets called very frequently should be atomic, because the rwsem read-locking
used for blocking chains is very likely to incur cache misses on SMP systems.
(However, if the chain's callout routines may sleep then the chain cannot be
atomic.)

The patch set was written by Alan Stern and Chandra Seetharaman, incorporating
material written by Keith Owens and suggestions from Paul McKenney and Andrew
Morton.

[jes@sgi.com: restructure the notifier chain initialization macros]
Signed-off-by: Alan Stern <stern@rowland.harvard.edu>
Signed-off-by: Chandra Seetharaman <sekharan@us.ibm.com>
Signed-off-by: Jes Sorensen <jes@sgi.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

11 files changed, 134 insertions, 61 deletions

diff --git a/include/asm-i386/kdebug.h b/include/asm-i386/kdebug.h
index 316138e8991..96d0828ce09 100644
--- a/include/asm-i386/kdebug.h
+++ b/include/asm-i386/kdebug.h
@@ -17,11 +17,9 @@ struct die_args {
 	int signr;
 };
 
-/* Note - you should never unregister because that can race with NMIs.
-   If you really want to do it first unregister - then synchronize_sched - then free.
-  */
-int register_die_notifier(struct notifier_block *nb);
-extern struct notifier_block *i386die_chain;
+extern int register_die_notifier(struct notifier_block *);
+extern int unregister_die_notifier(struct notifier_block *);
+extern struct atomic_notifier_head i386die_chain;
 
 
 /* Grossly misnamed. */
@@ -51,7 +49,7 @@ static inline int notify_die(enum die_val val, const char *str,
 		.trapnr = trap,
 		.signr = sig
 	};
-	return notifier_call_chain(&i386die_chain, val, &args);
+	return atomic_notifier_call_chain(&i386die_chain, val, &args);
 }
 
 #endif
diff --git a/include/asm-ia64/kdebug.h b/include/asm-ia64/kdebug.h
index 8b01a083dde..218c458ab60 100644
--- a/include/asm-ia64/kdebug.h
+++ b/include/asm-ia64/kdebug.h
@@ -40,7 +40,7 @@ struct die_args {
 
 extern int register_die_notifier(struct notifier_block *);
 extern int unregister_die_notifier(struct notifier_block *);
-extern struct notifier_block *ia64die_chain;
+extern struct atomic_notifier_head ia64die_chain;
 
 enum die_val {
 	DIE_BREAK = 1,
@@ -81,7 +81,7 @@ static inline int notify_die(enum die_val val, char *str, struct pt_regs *regs,
 		.signr  = sig
 	};
 
-	return notifier_call_chain(&ia64die_chain, val, &args);
+	return atomic_notifier_call_chain(&ia64die_chain, val, &args);
 }
 
 #endif
diff --git a/include/asm-powerpc/kdebug.h b/include/asm-powerpc/kdebug.h
index 7c16265568e..c01786ab5fa 100644
--- a/include/asm-powerpc/kdebug.h
+++ b/include/asm-powerpc/kdebug.h
@@ -16,13 +16,9 @@ struct die_args {
 	int signr;
 };
 
-/*
-   Note - you should never unregister because that can race with NMIs.
-   If you really want to do it first unregister - then synchronize_sched -
-   then free.
- */
-int register_die_notifier(struct notifier_block *nb);
-extern struct notifier_block *powerpc_die_chain;
+extern int register_die_notifier(struct notifier_block *);
+extern int unregister_die_notifier(struct notifier_block *);
+extern struct atomic_notifier_head powerpc_die_chain;
 
 /* Grossly misnamed. */
 enum die_val {
@@ -37,7 +33,7 @@ enum die_val {
 static inline int notify_die(enum die_val val,char *str,struct pt_regs *regs,long err,int trap, int sig)
 {
 	struct die_args args = { .regs=regs, .str=str, .err=err, .trapnr=trap,.signr=sig };
-	return notifier_call_chain(&powerpc_die_chain, val, &args);
+	return atomic_notifier_call_chain(&powerpc_die_chain, val, &args);
 }
 
 #endif /* __KERNEL__ */
diff --git a/include/asm-sparc64/kdebug.h b/include/asm-sparc64/kdebug.h
index 6321f5a0198..4040d127ac3 100644
--- a/include/asm-sparc64/kdebug.h
+++ b/include/asm-sparc64/kdebug.h
@@ -15,12 +15,9 @@ struct die_args {
 	int signr;
 };
 
-/* Note - you should never unregister because that can race with NMIs.
- * If you really want to do it first unregister - then synchronize_sched
- * - then free.
- */
-int register_die_notifier(struct notifier_block *nb);
-extern struct notifier_block *sparc64die_chain;
+extern int register_die_notifier(struct notifier_block *);
+extern int unregister_die_notifier(struct notifier_block *);
+extern struct atomic_notifier_head sparc64die_chain;
 
 extern void bad_trap(struct pt_regs *, long);
 
@@ -46,7 +43,7 @@ static inline int notify_die(enum die_val val,char *str, struct pt_regs *regs,
 				 .trapnr	= trap,
 				 .signr		= sig };
 
-	return notifier_call_chain(&sparc64die_chain, val, &args);
+	return atomic_notifier_call_chain(&sparc64die_chain, val, &args);
 }
 
 #endif
diff --git a/include/asm-x86_64/kdebug.h b/include/asm-x86_64/kdebug.h
index b9ed4c0c878..cf795631d9b 100644
--- a/include/asm-x86_64/kdebug.h
+++ b/include/asm-x86_64/kdebug.h
@@ -5,21 +5,20 @@
 
 struct pt_regs;
 
-struct die_args { 
+struct die_args {
 	struct pt_regs *regs;
 	const char *str;
-	long err; 
+	long err;
 	int trapnr;
 	int signr;
-}; 
+};
+
+extern int register_die_notifier(struct notifier_block *);
+extern int unregister_die_notifier(struct notifier_block *);
+extern struct atomic_notifier_head die_chain;
 
-/* Note - you should never unregister because that can race with NMIs.
-   If you really want to do it first unregister - then synchronize_sched - then free.
-  */
-int register_die_notifier(struct notifier_block *nb);
-extern struct notifier_block *die_chain;
 /* Grossly misnamed. */
-enum die_val { 
+enum die_val {
 	DIE_OOPS = 1,
 	DIE_INT3,
 	DIE_DEBUG,
@@ -33,8 +32,8 @@ enum die_val {
 	DIE_CALL,
 	DIE_NMI_IPI,
 	DIE_PAGE_FAULT,
-}; 
-	
+};
+
 static inline int notify_die(enum die_val val, const char *str,
 			struct pt_regs *regs, long err, int trap, int sig)
 {
@@ -45,7 +44,7 @@ static inline int notify_die(enum die_val val, const char *str,
 		.trapnr = trap,
 		.signr = sig
 	};
-	return notifier_call_chain(&die_chain, val, &args); 
+	return atomic_notifier_call_chain(&die_chain, val, &args);
 } 
 
 extern int printk_address(unsigned long address);
diff --git a/include/linux/adb.h b/include/linux/adb.h
index e9fdc63483c..b7305b17827 100644
--- a/include/linux/adb.h
+++ b/include/linux/adb.h
@@ -85,7 +85,7 @@ enum adb_message {
     ADB_MSG_POST_RESET	/* Called after resetting the bus (re-do init & register) */
 };
 extern struct adb_driver *adb_controller;
-extern struct notifier_block *adb_client_list;
+extern struct blocking_notifier_head adb_client_list;
 
 int adb_request(struct adb_request *req, void (*done)(struct adb_request *),
 		int flags, int nbytes, ...);
diff --git a/include/linux/kernel.h b/include/linux/kernel.h
index 03d6cfaa5b8..a3720f973ea 100644
--- a/include/linux/kernel.h
+++ b/include/linux/kernel.h
@@ -87,7 +87,7 @@ extern int cond_resched(void);
 		(__x < 0) ? -__x : __x;		\
 	})
 
-extern struct notifier_block *panic_notifier_list;
+extern struct atomic_notifier_head panic_notifier_list;
 extern long (*panic_blink)(long time);
 NORET_TYPE void panic(const char * fmt, ...)
 	__attribute__ ((NORET_AND format (printf, 1, 2)));
diff --git a/include/linux/memory.h b/include/linux/memory.h
index e251dc43d0f..8f04143ca36 100644
--- a/include/linux/memory.h
+++ b/include/linux/memory.h
@@ -77,7 +77,6 @@ extern int remove_memory_block(unsigned long, struct mem_section *, int);
 
 #define CONFIG_MEM_BLOCK_SIZE	(PAGES_PER_SECTION<<PAGE_SHIFT)
 
-struct notifier_block;
 
 #endif /* CONFIG_MEMORY_HOTPLUG */
 
diff --git a/include/linux/netfilter_ipv4/ip_conntrack.h b/include/linux/netfilter_ipv4/ip_conntrack.h
index f32d75c4f4c..d54d7b278e9 100644
--- a/include/linux/netfilter_ipv4/ip_conntrack.h
+++ b/include/linux/netfilter_ipv4/ip_conntrack.h
@@ -308,29 +308,30 @@ DECLARE_PER_CPU(struct ip_conntrack_ecache, ip_conntrack_ecache);
 
 #define CONNTRACK_ECACHE(x)	(__get_cpu_var(ip_conntrack_ecache).x)
  
-extern struct notifier_block *ip_conntrack_chain;
-extern struct notifier_block *ip_conntrack_expect_chain;
+extern struct atomic_notifier_head ip_conntrack_chain;
+extern struct atomic_notifier_head ip_conntrack_expect_chain;
 
 static inline int ip_conntrack_register_notifier(struct notifier_block *nb)
 {
-	return notifier_chain_register(&ip_conntrack_chain, nb);
+	return atomic_notifier_chain_register(&ip_conntrack_chain, nb);
 }
 
 static inline int ip_conntrack_unregister_notifier(struct notifier_block *nb)
 {
-	return notifier_chain_unregister(&ip_conntrack_chain, nb);
+	return atomic_notifier_chain_unregister(&ip_conntrack_chain, nb);
 }
 
 static inline int 
 ip_conntrack_expect_register_notifier(struct notifier_block *nb)
 {
-	return notifier_chain_register(&ip_conntrack_expect_chain, nb);
+	return atomic_notifier_chain_register(&ip_conntrack_expect_chain, nb);
 }
 
 static inline int
 ip_conntrack_expect_unregister_notifier(struct notifier_block *nb)
 {
-	return notifier_chain_unregister(&ip_conntrack_expect_chain, nb);
+	return atomic_notifier_chain_unregister(&ip_conntrack_expect_chain,
+			nb);
 }
 
 extern void ip_ct_deliver_cached_events(const struct ip_conntrack *ct);
@@ -355,14 +356,14 @@ static inline void ip_conntrack_event(enum ip_conntrack_events event,
 				      struct ip_conntrack *ct)
 {
 	if (is_confirmed(ct) && !is_dying(ct))
-		notifier_call_chain(&ip_conntrack_chain, event, ct);
+		atomic_notifier_call_chain(&ip_conntrack_chain, event, ct);
 }
 
 static inline void 
 ip_conntrack_expect_event(enum ip_conntrack_expect_events event,
 			  struct ip_conntrack_expect *exp)
 {
-	notifier_call_chain(&ip_conntrack_expect_chain, event, exp);
+	atomic_notifier_call_chain(&ip_conntrack_expect_chain, event, exp);
 }
 #else /* CONFIG_IP_NF_CONNTRACK_EVENTS */
 static inline void ip_conntrack_event_cache(enum ip_conntrack_events event, 
diff --git a/include/linux/notifier.h b/include/linux/notifier.h
index 5937dd6053c..51dbab9710c 100644
--- a/include/linux/notifier.h
+++ b/include/linux/notifier.h
@@ -10,25 +10,107 @@
 #ifndef _LINUX_NOTIFIER_H
 #define _LINUX_NOTIFIER_H
 #include <linux/errno.h>
+#include <linux/mutex.h>
+#include <linux/rwsem.h>
 
-struct notifier_block
-{
-	int (*notifier_call)(struct notifier_block *self, unsigned long, void *);
+/*
+ * Notifier chains are of three types:
+ *
+ *	Atomic notifier chains: Chain callbacks run in interrupt/atomic
+ *		context. Callouts are not allowed to block.
+ *	Blocking notifier chains: Chain callbacks run in process context.
+ *		Callouts are allowed to block.
+ *	Raw notifier chains: There are no restrictions on callbacks,
+ *		registration, or unregistration.  All locking and protection
+ *		must be provided by the caller.
+ *
+ * atomic_notifier_chain_register() may be called from an atomic context,
+ * but blocking_notifier_chain_register() must be called from a process
+ * context.  Ditto for the corresponding _unregister() routines.
+ *
+ * atomic_notifier_chain_unregister() and blocking_notifier_chain_unregister()
+ * _must not_ be called from within the call chain.
+ */
+
+struct notifier_block {
+	int (*notifier_call)(struct notifier_block *, unsigned long, void *);
 	struct notifier_block *next;
 	int priority;
 };
 
+struct atomic_notifier_head {
+	spinlock_t lock;
+	struct notifier_block *head;
+};
+
+struct blocking_notifier_head {
+	struct rw_semaphore rwsem;
+	struct notifier_block *head;
+};
+
+struct raw_notifier_head {
+	struct notifier_block *head;
+};
+
+#define ATOMIC_INIT_NOTIFIER_HEAD(name) do {	\
+		spin_lock_init(&(name)->lock);	\
+		(name)->head = NULL;		\
+	} while (0)
+#define BLOCKING_INIT_NOTIFIER_HEAD(name) do {	\
+		init_rwsem(&(name)->rwsem);	\
+		(name)->head = NULL;		\
+	} while (0)
+#define RAW_INIT_NOTIFIER_HEAD(name) do {	\
+		(name)->head = NULL;		\
+	} while (0)
+
+#define ATOMIC_NOTIFIER_INIT(name) {				\
+		.lock = SPIN_LOCK_UNLOCKED,			\
+		.head = NULL }
+#define BLOCKING_NOTIFIER_INIT(name) {				\
+		.rwsem = __RWSEM_INITIALIZER((name).rwsem),	\
+		.head = NULL }
+#define RAW_NOTIFIER_INIT(name)	{				\
+		.head = NULL }
+
+#define ATOMIC_NOTIFIER_HEAD(name)				\
+	struct atomic_notifier_head name =			\
+		ATOMIC_NOTIFIER_INIT(name)
+#define BLOCKING_NOTIFIER_HEAD(name)				\
+	struct blocking_notifier_head name =			\
+		BLOCKING_NOTIFIER_INIT(name)
+#define RAW_NOTIFIER_HEAD(name)					\
+	struct raw_notifier_head name =				\
+		RAW_NOTIFIER_INIT(name)
 
 #ifdef __KERNEL__
 
-extern int notifier_chain_register(struct notifier_block **list, struct notifier_block *n);
-extern int notifier_chain_unregister(struct notifier_block **nl, struct notifier_block *n);
-extern int notifier_call_chain(struct notifier_block **n, unsigned long val, void *v);
+extern int atomic_notifier_chain_register(struct atomic_notifier_head *,
+		struct notifier_block *);
+extern int blocking_notifier_chain_register(struct blocking_notifier_head *,
+		struct notifier_block *);
+extern int raw_notifier_chain_register(struct raw_notifier_head *,
+		struct notifier_block *);
+
+extern int atomic_notifier_chain_unregister(struct atomic_notifier_head *,
+		struct notifier_block *);
+extern int blocking_notifier_chain_unregister(struct blocking_notifier_head *,
+		struct notifier_block *);
+extern int raw_notifier_chain_unregister(struct raw_notifier_head *,
+		struct notifier_block *);
+
+extern int atomic_notifier_call_chain(struct atomic_notifier_head *,
+		unsigned long val, void *v);
+extern int blocking_notifier_call_chain(struct blocking_notifier_head *,
+		unsigned long val, void *v);
+extern int raw_notifier_call_chain(struct raw_notifier_head *,
+		unsigned long val, void *v);
 
 #define NOTIFY_DONE		0x0000		/* Don't care */
 #define NOTIFY_OK		0x0001		/* Suits me */
 #define NOTIFY_STOP_MASK	0x8000		/* Don't call further */
-#define NOTIFY_BAD		(NOTIFY_STOP_MASK|0x0002)	/* Bad/Veto action	*/
+#define NOTIFY_BAD		(NOTIFY_STOP_MASK|0x0002)
+						/* Bad/Veto action */
 /*
  * Clean way to return from the notifier and stop further calls.
  */
diff --git a/include/net/netfilter/nf_conntrack.h b/include/net/netfilter/nf_conntrack.h
index b6f0905a4ee..916013ca4a5 100644
--- a/include/net/netfilter/nf_conntrack.h
+++ b/include/net/netfilter/nf_conntrack.h
@@ -300,29 +300,30 @@ DECLARE_PER_CPU(struct nf_conntrack_ecache, nf_conntrack_ecache);
 
 #define CONNTRACK_ECACHE(x)	(__get_cpu_var(nf_conntrack_ecache).x)
 
-extern struct notifier_block *nf_conntrack_chain;
-extern struct notifier_block *nf_conntrack_expect_chain;
+extern struct atomic_notifier_head nf_conntrack_chain;
+extern struct atomic_notifier_head nf_conntrack_expect_chain;
 
 static inline int nf_conntrack_register_notifier(struct notifier_block *nb)
 {
-	return notifier_chain_register(&nf_conntrack_chain, nb);
+	return atomic_notifier_chain_register(&nf_conntrack_chain, nb);
 }
 
 static inline int nf_conntrack_unregister_notifier(struct notifier_block *nb)
 {
-	return notifier_chain_unregister(&nf_conntrack_chain, nb);
+	return atomic_notifier_chain_unregister(&nf_conntrack_chain, nb);
 }
 
 static inline int
 nf_conntrack_expect_register_notifier(struct notifier_block *nb)
 {
-	return notifier_chain_register(&nf_conntrack_expect_chain, nb);
+	return atomic_notifier_chain_register(&nf_conntrack_expect_chain, nb);
 }
 
 static inline int
 nf_conntrack_expect_unregister_notifier(struct notifier_block *nb)
 {
-	return notifier_chain_unregister(&nf_conntrack_expect_chain, nb);
+	return atomic_notifier_chain_unregister(&nf_conntrack_expect_chain,
+			nb);
 }
 
 extern void nf_ct_deliver_cached_events(const struct nf_conn *ct);
@@ -347,14 +348,14 @@ static inline void nf_conntrack_event(enum ip_conntrack_events event,
 				      struct nf_conn *ct)
 {
 	if (nf_ct_is_confirmed(ct) && !nf_ct_is_dying(ct))
-		notifier_call_chain(&nf_conntrack_chain, event, ct);
+		atomic_notifier_call_chain(&nf_conntrack_chain, event, ct);
 }
 
 static inline void
 nf_conntrack_expect_event(enum ip_conntrack_expect_events event,
 			  struct nf_conntrack_expect *exp)
 {
-	notifier_call_chain(&nf_conntrack_expect_chain, event, exp);
+	atomic_notifier_call_chain(&nf_conntrack_expect_chain, event, exp);
 }
 #else /* CONFIG_NF_CONNTRACK_EVENTS */
 static inline void nf_conntrack_event_cache(enum ip_conntrack_events event,