Browse Source

Merge branches 'tracing/ftrace', 'tracing/markers', 'tracing/mmiotrace', 'tracing/nmisafe', 'tracing/tracepoints' and 'tracing/urgent' into tracing/core

Ingo Molnar 16 years ago

+ 5 - 0
arch/arm/include/asm/ftrace.h

@@ -1,6 +1,11 @@
 #ifndef _ASM_ARM_FTRACE
 #define _ASM_ARM_FTRACE
 
+#ifndef __ASSEMBLY__
+static inline void ftrace_nmi_enter(void) { }
+static inline void ftrace_nmi_exit(void) { }
+#endif
+
 #ifdef CONFIG_FUNCTION_TRACER
 #define MCOUNT_ADDR		((long)(mcount))
 #define MCOUNT_INSN_SIZE	4 /* sizeof mcount call */

+ 5 - 0
arch/powerpc/include/asm/ftrace.h

@@ -1,6 +1,11 @@
 #ifndef _ASM_POWERPC_FTRACE
 #define _ASM_POWERPC_FTRACE
 
+#ifndef __ASSEMBLY__
+static inline void ftrace_nmi_enter(void) { }
+static inline void ftrace_nmi_exit(void) { }
+#endif
+
 #ifdef CONFIG_FUNCTION_TRACER
 #define MCOUNT_ADDR		((long)(_mcount))
 #define MCOUNT_INSN_SIZE	4 /* sizeof mcount call */

+ 5 - 0
arch/sh/include/asm/ftrace.h

@@ -1,6 +1,11 @@
 #ifndef __ASM_SH_FTRACE_H
 #define __ASM_SH_FTRACE_H
 
+#ifndef __ASSEMBLY__
+static inline void ftrace_nmi_enter(void) { }
+static inline void ftrace_nmi_exit(void) { }
+#endif
+
 #ifndef __ASSEMBLY__
 extern void mcount(void);
 #endif

+ 5 - 0
arch/sparc/include/asm/ftrace.h

@@ -1,6 +1,11 @@
 #ifndef _ASM_SPARC64_FTRACE
 #define _ASM_SPARC64_FTRACE
 
+#ifndef __ASSEMBLY__
+static inline void ftrace_nmi_enter(void) { }
+static inline void ftrace_nmi_exit(void) { }
+#endif
+
 #ifdef CONFIG_MCOUNT
 #define MCOUNT_ADDR		((long)(_mcount))
 #define MCOUNT_INSN_SIZE	4 /* sizeof mcount call */

+ 0 - 4
arch/x86/Kconfig.debug

@@ -186,14 +186,10 @@ config IOMMU_LEAK
 	  Add a simple leak tracer to the IOMMU code. This is useful when you
 	  are debugging a buggy device driver that leaks IOMMU mappings.
 
-config MMIOTRACE_HOOKS
-	bool
-
 config MMIOTRACE
 	bool "Memory mapped IO tracing"
 	depends on DEBUG_KERNEL && PCI
 	select TRACING
-	select MMIOTRACE_HOOKS
 	help
 	  Mmiotrace traces Memory Mapped I/O access and is meant for
 	  debugging and reverse engineering. It is called from the ioremap

+ 15 - 0
arch/x86/include/asm/ftrace.h

@@ -17,6 +17,21 @@ static inline unsigned long ftrace_call_adjust(unsigned long addr)
 	 */
 	return addr - 1;
 }
+
+#ifdef CONFIG_DYNAMIC_FTRACE
+extern void ftrace_nmi_enter(void);
+extern void ftrace_nmi_exit(void);
+#else
+static inline void ftrace_nmi_enter(void) { }
+static inline void ftrace_nmi_exit(void) { }
+#endif
+#endif /* __ASSEMBLY__ */
+
+#else /* CONFIG_FUNCTION_TRACER */
+
+#ifndef __ASSEMBLY__
+static inline void ftrace_nmi_enter(void) { }
+static inline void ftrace_nmi_exit(void) { }
 #endif
 
 #endif /* CONFIG_FUNCTION_TRACER */

+ 128 - 1
arch/x86/kernel/ftrace.c

@@ -56,6 +56,133 @@ unsigned char *ftrace_call_replace(unsigned long ip, unsigned long addr)
 	return calc.code;
 }
 
+/*
+ * Modifying code must take extra care. On an SMP machine, if
+ * the code being modified is also being executed on another CPU
+ * that CPU will have undefined results and possibly take a GPF.
+ * We use kstop_machine to stop other CPUS from exectuing code.
+ * But this does not stop NMIs from happening. We still need
+ * to protect against that. We separate out the modification of
+ * the code to take care of this.
+ *
+ * Two buffers are added: An IP buffer and a "code" buffer.
+ *
+ * 1) Put the instruction pointer into the IP buffer
+ *    and the new code into the "code" buffer.
+ * 2) Set a flag that says we are modifying code
+ * 3) Wait for any running NMIs to finish.
+ * 4) Write the code
+ * 5) clear the flag.
+ * 6) Wait for any running NMIs to finish.
+ *
+ * If an NMI is executed, the first thing it does is to call
+ * "ftrace_nmi_enter". This will check if the flag is set to write
+ * and if it is, it will write what is in the IP and "code" buffers.
+ *
+ * The trick is, it does not matter if everyone is writing the same
+ * content to the code location. Also, if a CPU is executing code
+ * it is OK to write to that code location if the contents being written
+ * are the same as what exists.
+ */
+
+static atomic_t in_nmi = ATOMIC_INIT(0);
+static int mod_code_status;		/* holds return value of text write */
+static int mod_code_write;		/* set when NMI should do the write */
+static void *mod_code_ip;		/* holds the IP to write to */
+static void *mod_code_newcode;		/* holds the text to write to the IP */
+
+static unsigned nmi_wait_count;
+static atomic_t nmi_update_count = ATOMIC_INIT(0);
+
+int ftrace_arch_read_dyn_info(char *buf, int size)
+{
+	int r;
+
+	r = snprintf(buf, size, "%u %u",
+		     nmi_wait_count,
+		     atomic_read(&nmi_update_count));
+	return r;
+}
+
+static void ftrace_mod_code(void)
+{
+	/*
+	 * Yes, more than one CPU process can be writing to mod_code_status.
+	 *    (and the code itself)
+	 * But if one were to fail, then they all should, and if one were
+	 * to succeed, then they all should.
+	 */
+	mod_code_status = probe_kernel_write(mod_code_ip, mod_code_newcode,
+					     MCOUNT_INSN_SIZE);
+
+}
+
+void ftrace_nmi_enter(void)
+{
+	atomic_inc(&in_nmi);
+	/* Must have in_nmi seen before reading write flag */
+	smp_mb();
+	if (mod_code_write) {
+		ftrace_mod_code();
+		atomic_inc(&nmi_update_count);
+	}
+}
+
+void ftrace_nmi_exit(void)
+{
+	/* Finish all executions before clearing in_nmi */
+	smp_wmb();
+	atomic_dec(&in_nmi);
+}
+
+static void wait_for_nmi(void)
+{
+	int waited = 0;
+
+	while (atomic_read(&in_nmi)) {
+		waited = 1;
+		cpu_relax();
+	}
+
+	if (waited)
+		nmi_wait_count++;
+}
+
+static int
+do_ftrace_mod_code(unsigned long ip, void *new_code)
+{
+	mod_code_ip = (void *)ip;
+	mod_code_newcode = new_code;
+
+	/* The buffers need to be visible before we let NMIs write them */
+	smp_wmb();
+
+	mod_code_write = 1;
+
+	/* Make sure write bit is visible before we wait on NMIs */
+	smp_mb();
+
+	wait_for_nmi();
+
+	/* Make sure all running NMIs have finished before we write the code */
+	smp_mb();
+
+	ftrace_mod_code();
+
+	/* Make sure the write happens before clearing the bit */
+	smp_wmb();
+
+	mod_code_write = 0;
+
+	/* make sure NMIs see the cleared bit */
+	smp_mb();
+
+	wait_for_nmi();
+
+	return mod_code_status;
+}
+
+
 int
 ftrace_modify_code(unsigned long ip, unsigned char *old_code,
 		   unsigned char *new_code)
@@ -81,7 +208,7 @@ ftrace_modify_code(unsigned long ip, unsigned char *old_code,
 		return -EINVAL;
 
 	/* replace the text with the new text */
-	if (probe_kernel_write((void *)ip, new_code, MCOUNT_INSN_SIZE))
+	if (do_ftrace_mod_code(ip, new_code))
 		return -EPERM;
 
 	sync_core();

+ 1 - 2
arch/x86/mm/Makefile

@@ -8,9 +8,8 @@ obj-$(CONFIG_X86_PTDUMP)	+= dump_pagetables.o
 
 obj-$(CONFIG_HIGHMEM)		+= highmem_32.o
 
-obj-$(CONFIG_MMIOTRACE_HOOKS)	+= kmmio.o
 obj-$(CONFIG_MMIOTRACE)		+= mmiotrace.o
-mmiotrace-y			:= pf_in.o mmio-mod.o
+mmiotrace-y			:= kmmio.o pf_in.o mmio-mod.o
 obj-$(CONFIG_MMIOTRACE_TEST)	+= testmmiotrace.o
 
 obj-$(CONFIG_NUMA)		+= numa_$(BITS).o

+ 1 - 1
arch/x86/mm/fault.c

@@ -53,7 +53,7 @@
 
 static inline int kmmio_fault(struct pt_regs *regs, unsigned long addr)
 {
-#ifdef CONFIG_MMIOTRACE_HOOKS
+#ifdef CONFIG_MMIOTRACE
 	if (unlikely(is_kmmio_active()))
 		if (kmmio_handler(regs, addr) == 1)
 			return -1;

+ 3 - 0
include/linux/ftrace.h

@@ -74,6 +74,9 @@ extern void ftrace_caller(void);
 extern void ftrace_call(void);
 extern void mcount_call(void);
 
+/* May be defined in arch */
+extern int ftrace_arch_read_dyn_info(char *buf, int size);
+
 /**
  * ftrace_modify_code - modify code segment
  * @ip: the address of the code segment

+ 13 - 2
include/linux/hardirq.h

@@ -5,6 +5,7 @@
 #include <linux/smp_lock.h>
 #include <linux/lockdep.h>
 #include <asm/hardirq.h>
+#include <asm/ftrace.h>
 #include <asm/system.h>
 
 /*
@@ -161,7 +162,17 @@ extern void irq_enter(void);
  */
 extern void irq_exit(void);
 
-#define nmi_enter()		do { lockdep_off(); __irq_enter(); } while (0)
-#define nmi_exit()		do { __irq_exit(); lockdep_on(); } while (0)
+#define nmi_enter()				\
+	do {					\
+		ftrace_nmi_enter();		\
+		lockdep_off();			\
+		__irq_enter();			\
+	} while (0)
+#define nmi_exit()				\
+	do {					\
+		__irq_exit();			\
+		lockdep_on();			\
+		ftrace_nmi_exit();		\
+	} while (0)
 
 #endif /* LINUX_HARDIRQ_H */

+ 0 - 2
include/linux/marker.h

@@ -136,8 +136,6 @@ extern marker_probe_func __mark_empty_function;
 
 extern void marker_probe_cb(const struct marker *mdata,
 	void *call_private, ...);
-extern void marker_probe_cb_noarg(const struct marker *mdata,
-	void *call_private, ...);
 
 /*
  * Connect a probe to a marker.

+ 4 - 0
include/linux/tracepoint.h

@@ -112,6 +112,10 @@ extern int tracepoint_probe_register(const char *name, void *probe);
  */
 extern int tracepoint_probe_unregister(const char *name, void *probe);
 
+extern int tracepoint_probe_register_noupdate(const char *name, void *probe);
+extern int tracepoint_probe_unregister_noupdate(const char *name, void *probe);
+extern void tracepoint_probe_update_all(void);
+
 struct tracepoint_iter {
 	struct module *module;
 	struct tracepoint *tracepoint;

+ 28 - 52
kernel/marker.c

@@ -43,6 +43,7 @@ static DEFINE_MUTEX(markers_mutex);
  */
 #define MARKER_HASH_BITS 6
 #define MARKER_TABLE_SIZE (1 << MARKER_HASH_BITS)
+static struct hlist_head marker_table[MARKER_TABLE_SIZE];
 
 /*
  * Note about RCU :
@@ -64,11 +65,10 @@ struct marker_entry {
 	void *oldptr;
 	int rcu_pending;
 	unsigned char ptype:1;
+	unsigned char format_allocated:1;
 	char name[0];	/* Contains name'\0'format'\0' */
 };
 
-static struct hlist_head marker_table[MARKER_TABLE_SIZE];
-
 /**
  * __mark_empty_function - Empty probe callback
  * @probe_private: probe private data
@@ -157,7 +157,7 @@ EXPORT_SYMBOL_GPL(marker_probe_cb);
  *
  * Should be connected to markers "MARK_NOARGS".
  */
-void marker_probe_cb_noarg(const struct marker *mdata, void *call_private, ...)
+static void marker_probe_cb_noarg(const struct marker *mdata, void *call_private, ...)
 {
 	va_list args;	/* not initialized */
 	char ptype;
@@ -197,7 +197,6 @@ void marker_probe_cb_noarg(const struct marker *mdata, void *call_private, ...)
 	}
 	rcu_read_unlock_sched();
 }
-EXPORT_SYMBOL_GPL(marker_probe_cb_noarg);
 
 static void free_old_closure(struct rcu_head *head)
 {
@@ -416,6 +415,7 @@ static struct marker_entry *add_marker(const char *name, const char *format)
 	e->single.probe_private = NULL;
 	e->multi = NULL;
 	e->ptype = 0;
+	e->format_allocated = 0;
 	e->refcount = 0;
 	e->rcu_pending = 0;
 	hlist_add_head(&e->hlist, head);
@@ -447,6 +447,8 @@ static int remove_marker(const char *name)
 	if (e->single.func != __mark_empty_function)
 		return -EBUSY;
 	hlist_del(&e->hlist);
+	if (e->format_allocated)
+		kfree(e->format);
 	/* Make sure the call_rcu has been executed */
 	if (e->rcu_pending)
 		rcu_barrier_sched();
@@ -457,57 +459,34 @@ static int remove_marker(const char *name)
 /*
  * Set the mark_entry format to the format found in the element.
  */
-static int marker_set_format(struct marker_entry **entry, const char *format)
+static int marker_set_format(struct marker_entry *entry, const char *format)
 {
-	struct marker_entry *e;
-	size_t name_len = strlen((*entry)->name) + 1;
-	size_t format_len = strlen(format) + 1;
-
-
-	e = kmalloc(sizeof(struct marker_entry) + name_len + format_len,
-			GFP_KERNEL);
-	if (!e)
+	entry->format = kstrdup(format, GFP_KERNEL);
+	if (!entry->format)
 		return -ENOMEM;
-	memcpy(&e->name[0], (*entry)->name, name_len);
-	e->format = &e->name[name_len];
-	memcpy(e->format, format, format_len);
-	if (strcmp(e->format, MARK_NOARGS) == 0)
-		e->call = marker_probe_cb_noarg;
-	else
-		e->call = marker_probe_cb;
-	e->single = (*entry)->single;
-	e->multi = (*entry)->multi;
-	e->ptype = (*entry)->ptype;
-	e->refcount = (*entry)->refcount;
-	e->rcu_pending = 0;
-	hlist_add_before(&e->hlist, &(*entry)->hlist);
-	hlist_del(&(*entry)->hlist);
-	/* Make sure the call_rcu has been executed */
-	if ((*entry)->rcu_pending)
-		rcu_barrier_sched();
-	kfree(*entry);
-	*entry = e;
+	entry->format_allocated = 1;
+
 	trace_mark(core_marker_format, "name %s format %s",
-			e->name, e->format);
+			entry->name, entry->format);
 	return 0;
 }
 
 /*
  * Sets the probe callback corresponding to one marker.
  */
-static int set_marker(struct marker_entry **entry, struct marker *elem,
+static int set_marker(struct marker_entry *entry, struct marker *elem,
 		int active)
 {
 	int ret;
-	WARN_ON(strcmp((*entry)->name, elem->name) != 0);
+	WARN_ON(strcmp(entry->name, elem->name) != 0);
 
-	if ((*entry)->format) {
-		if (strcmp((*entry)->format, elem->format) != 0) {
+	if (entry->format) {
+		if (strcmp(entry->format, elem->format) != 0) {
 			printk(KERN_NOTICE
 				"Format mismatch for probe %s "
 				"(%s), marker (%s)\n",
-				(*entry)->name,
-				(*entry)->format,
+				entry->name,
+				entry->format,
 				elem->format);
 			return -EPERM;
 		}
@@ -523,34 +502,33 @@ static int set_marker(struct marker_entry **entry, struct marker *elem,
 	 * pass from a "safe" callback (with argument) to an "unsafe"
 	 * callback (does not set arguments).
 	 */
-	elem->call = (*entry)->call;
+	elem->call = entry->call;
 	/*
 	 * Sanity check :
 	 * We only update the single probe private data when the ptr is
 	 * set to a _non_ single probe! (0 -> 1 and N -> 1, N != 1)
 	 */
 	WARN_ON(elem->single.func != __mark_empty_function
-		&& elem->single.probe_private
-		!= (*entry)->single.probe_private &&
-		!elem->ptype);
-	elem->single.probe_private = (*entry)->single.probe_private;
+		&& elem->single.probe_private != entry->single.probe_private
+		&& !elem->ptype);
+	elem->single.probe_private = entry->single.probe_private;
 	/*
 	 * Make sure the private data is valid when we update the
 	 * single probe ptr.
 	 */
 	smp_wmb();
-	elem->single.func = (*entry)->single.func;
+	elem->single.func = entry->single.func;
 	/*
 	 * We also make sure that the new probe callbacks array is consistent
 	 * before setting a pointer to it.
 	 */
-	rcu_assign_pointer(elem->multi, (*entry)->multi);
+	rcu_assign_pointer(elem->multi, entry->multi);
 	/*
 	 * Update the function or multi probe array pointer before setting the
 	 * ptype.
 	 */
 	smp_wmb();
-	elem->ptype = (*entry)->ptype;
+	elem->ptype = entry->ptype;
 	elem->state = active;
 
 	return 0;
@@ -594,8 +572,7 @@ void marker_update_probe_range(struct marker *begin,
 	for (iter = begin; iter < end; iter++) {
 		mark_entry = get_marker(iter->name);
 		if (mark_entry) {
-			set_marker(&mark_entry, iter,
-					!!mark_entry->refcount);
+			set_marker(mark_entry, iter, !!mark_entry->refcount);
 			/*
 			 * ignore error, continue
 			 */
@@ -657,7 +634,7 @@ int marker_probe_register(const char *name, const char *format,
 			ret = PTR_ERR(entry);
 	} else if (format) {
 		if (!entry->format)
-			ret = marker_set_format(&entry, format);
+			ret = marker_set_format(entry, format);
 		else if (strcmp(entry->format, format))
 			ret = -EPERM;
 	}
@@ -848,8 +825,6 @@ void *marker_get_private_data(const char *name, marker_probe_func *probe,
 			if (!e->ptype) {
 				if (num == 0 && e->single.func == probe)
 					return e->single.probe_private;
-				else
-					break;
 			} else {
 				struct marker_probe_closure *closure;
 				int match = 0;
@@ -861,6 +836,7 @@ void *marker_get_private_data(const char *name, marker_probe_func *probe,
 						return closure[i].probe_private;
 				}
 			}
+			break;
 		}
 	}
 	return ERR_PTR(-ENOENT);

+ 1 - 1
kernel/trace/Kconfig

@@ -25,7 +25,7 @@ config TRACING
 	bool
 	select DEBUG_FS
 	select RING_BUFFER
-	select STACKTRACE
+	select STACKTRACE if STACKTRACE_SUPPORT
 	select TRACEPOINTS
 	select NOP_TRACER
 

+ 25 - 7
kernel/trace/trace.c

@@ -736,6 +736,7 @@ static void ftrace_trace_stack(struct trace_array *tr,
 			       unsigned long flags,
 			       int skip, int pc)
 {
+#ifdef CONFIG_STACKTRACE
 	struct ring_buffer_event *event;
 	struct stack_entry *entry;
 	struct stack_trace trace;
@@ -761,6 +762,7 @@ static void ftrace_trace_stack(struct trace_array *tr,
 
 	save_stack_trace(&trace);
 	ring_buffer_unlock_commit(tr->buffer, event, irq_flags);
+#endif
 }
 
 void __trace_stack(struct trace_array *tr,
@@ -2860,22 +2862,38 @@ static struct file_operations tracing_mark_fops = {
 
 #ifdef CONFIG_DYNAMIC_FTRACE
 
+int __weak ftrace_arch_read_dyn_info(char *buf, int size)
+{
+	return 0;
+}
+
 static ssize_t
-tracing_read_long(struct file *filp, char __user *ubuf,
+tracing_read_dyn_info(struct file *filp, char __user *ubuf,
 		  size_t cnt, loff_t *ppos)
 {
+	static char ftrace_dyn_info_buffer[1024];
+	static DEFINE_MUTEX(dyn_info_mutex);
 	unsigned long *p = filp->private_data;
-	char buf[64];
+	char *buf = ftrace_dyn_info_buffer;
+	int size = ARRAY_SIZE(ftrace_dyn_info_buffer);
 	int r;
 
-	r = sprintf(buf, "%ld\n", *p);
+	mutex_lock(&dyn_info_mutex);
+	r = sprintf(buf, "%ld ", *p);
 
-	return simple_read_from_buffer(ubuf, cnt, ppos, buf, r);
+	r += ftrace_arch_read_dyn_info(buf+r, (size-1)-r);
+	buf[r++] = '\n';
+
+	r = simple_read_from_buffer(ubuf, cnt, ppos, buf, r);
+
+	mutex_unlock(&dyn_info_mutex);
+
+	return r;
 }
 
-static struct file_operations tracing_read_long_fops = {
+static struct file_operations tracing_dyn_info_fops = {
 	.open		= tracing_open_generic,
-	.read		= tracing_read_long,
+	.read		= tracing_read_dyn_info,
 };
 #endif
 
@@ -2984,7 +3002,7 @@ static __init int tracer_init_debugfs(void)
 #ifdef CONFIG_DYNAMIC_FTRACE
 	entry = debugfs_create_file("dyn_ftrace_total_info", 0444, d_tracer,
 				    &ftrace_update_tot_cnt,
-				    &tracing_read_long_fops);
+				    &tracing_dyn_info_fops);
 	if (!entry)
 		pr_warning("Could not create debugfs "
 			   "'dyn_ftrace_total_info' entry\n");

+ 159 - 102
kernel/tracepoint.c

@@ -43,6 +43,7 @@ static DEFINE_MUTEX(tracepoints_mutex);
  */
 #define TRACEPOINT_HASH_BITS 6
 #define TRACEPOINT_TABLE_SIZE (1 << TRACEPOINT_HASH_BITS)
+static struct hlist_head tracepoint_table[TRACEPOINT_TABLE_SIZE];
 
 /*
  * Note about RCU :
@@ -54,40 +55,43 @@ struct tracepoint_entry {
 	struct hlist_node hlist;
 	void **funcs;
 	int refcount;	/* Number of times armed. 0 if disarmed. */
-	struct rcu_head rcu;
-	void *oldptr;
-	unsigned char rcu_pending:1;
 	char name[0];
 };
 
-static struct hlist_head tracepoint_table[TRACEPOINT_TABLE_SIZE];
+struct tp_probes {
+	union {
+		struct rcu_head rcu;
+		struct list_head list;
+	} u;
+	void *probes[0];
+};
 
-static void free_old_closure(struct rcu_head *head)
+static inline void *allocate_probes(int count)
 {
-	struct tracepoint_entry *entry = container_of(head,
-		struct tracepoint_entry, rcu);
-	kfree(entry->oldptr);
-	/* Make sure we free the data before setting the pending flag to 0 */
-	smp_wmb();
-	entry->rcu_pending = 0;
+	struct tp_probes *p  = kmalloc(count * sizeof(void *)
+			+ sizeof(struct tp_probes), GFP_KERNEL);
+	return p == NULL ? NULL : p->probes;
 }
 
-static void tracepoint_entry_free_old(struct tracepoint_entry *entry, void *old)
+static void rcu_free_old_probes(struct rcu_head *head)
 {
-	if (!old)
-		return;
-	entry->oldptr = old;
-	entry->rcu_pending = 1;
-	/* write rcu_pending before calling the RCU callback */
-	smp_wmb();
-	call_rcu_sched(&entry->rcu, free_old_closure);
+	kfree(container_of(head, struct tp_probes, u.rcu));
+}
+
+static inline void release_probes(void *old)
+{
+	if (old) {
+		struct tp_probes *tp_probes = container_of(old,
+			struct tp_probes, probes[0]);
+		call_rcu_sched(&tp_probes->u.rcu, rcu_free_old_probes);
+	}
 }
 
 static void debug_print_probes(struct tracepoint_entry *entry)
 {
 	int i;
 
-	if (!tracepoint_debug)
+	if (!tracepoint_debug || !entry->funcs)
 		return;
 
 	for (i = 0; entry->funcs[i]; i++)
@@ -111,12 +115,13 @@ tracepoint_entry_add_probe(struct tracepoint_entry *entry, void *probe)
 				return ERR_PTR(-EEXIST);
 	}
 	/* + 2 : one for new probe, one for NULL func */
-	new = kzalloc((nr_probes + 2) * sizeof(void *), GFP_KERNEL);
+	new = allocate_probes(nr_probes + 2);
 	if (new == NULL)
 		return ERR_PTR(-ENOMEM);
 	if (old)
 		memcpy(new, old, nr_probes * sizeof(void *));
 	new[nr_probes] = probe;
+	new[nr_probes + 1] = NULL;
 	entry->refcount = nr_probes + 1;
 	entry->funcs = new;
 	debug_print_probes(entry);
@@ -132,7 +137,7 @@ tracepoint_entry_remove_probe(struct tracepoint_entry *entry, void *probe)
 	old = entry->funcs;
 
 	if (!old)
-		return NULL;
+		return ERR_PTR(-ENOENT);
 
 	debug_print_probes(entry);
 	/* (N -> M), (N > 1, M >= 0) probes */
@@ -151,13 +156,13 @@ tracepoint_entry_remove_probe(struct tracepoint_entry *entry, void *probe)
 		int j = 0;
 		/* N -> M, (N > 1, M > 0) */
 		/* + 1 for NULL */
-		new = kzalloc((nr_probes - nr_del + 1)
-			* sizeof(void *), GFP_KERNEL);
+		new = allocate_probes(nr_probes - nr_del + 1);
 		if (new == NULL)
 			return ERR_PTR(-ENOMEM);
 		for (i = 0; old[i]; i++)
 			if ((probe && old[i] != probe))
 				new[j++] = old[i];
+		new[nr_probes - nr_del] = NULL;
 		entry->refcount = nr_probes - nr_del;
 		entry->funcs = new;
 	}
@@ -215,7 +220,6 @@ static struct tracepoint_entry *add_tracepoint(const char *name)
 	memcpy(&e->name[0], name, name_len);
 	e->funcs = NULL;
 	e->refcount = 0;
-	e->rcu_pending = 0;
 	hlist_add_head(&e->hlist, head);
 	return e;
 }
@@ -224,32 +228,10 @@ static struct tracepoint_entry *add_tracepoint(const char *name)
  * Remove the tracepoint from the tracepoint hash table. Must be called with
  * mutex_lock held.
  */
-static int remove_tracepoint(const char *name)
+static inline void remove_tracepoint(struct tracepoint_entry *e)
 {
-	struct hlist_head *head;
-	struct hlist_node *node;
-	struct tracepoint_entry *e;
-	int found = 0;
-	size_t len = strlen(name) + 1;
-	u32 hash = jhash(name, len-1, 0);
-
-	head = &tracepoint_table[hash & (TRACEPOINT_TABLE_SIZE - 1)];
-	hlist_for_each_entry(e, node, head, hlist) {
-		if (!strcmp(name, e->name)) {
-			found = 1;
-			break;
-		}
-	}
-	if (!found)
-		return -ENOENT;
-	if (e->refcount)
-		return -EBUSY;
 	hlist_del(&e->hlist);
-	/* Make sure the call_rcu_sched has been executed */
-	if (e->rcu_pending)
-		rcu_barrier_sched();
 	kfree(e);
-	return 0;
 }
 
 /*
@@ -320,6 +302,23 @@ static void tracepoint_update_probes(void)
 	module_update_tracepoints();
 }
 
+static void *tracepoint_add_probe(const char *name, void *probe)
+{
+	struct tracepoint_entry *entry;
+	void *old;
+
+	entry = get_tracepoint(name);
+	if (!entry) {
+		entry = add_tracepoint(name);
+		if (IS_ERR(entry))
+			return entry;
+	}
+	old = tracepoint_entry_add_probe(entry, probe);
+	if (IS_ERR(old) && !entry->refcount)
+		remove_tracepoint(entry);
+	return old;
+}
+
 /**
  * tracepoint_probe_register -  Connect a probe to a tracepoint
  * @name: tracepoint name
@@ -330,44 +329,36 @@ static void tracepoint_update_probes(void)
  */
 int tracepoint_probe_register(const char *name, void *probe)
 {
-	struct tracepoint_entry *entry;
-	int ret = 0;
 	void *old;
 
 	mutex_lock(&tracepoints_mutex);
-	entry = get_tracepoint(name);
-	if (!entry) {
-		entry = add_tracepoint(name);
-		if (IS_ERR(entry)) {
-			ret = PTR_ERR(entry);
-			goto end;
-		}
-	}
-	/*
-	 * If we detect that a call_rcu_sched is pending for this tracepoint,
-	 * make sure it's executed now.
-	 */
-	if (entry->rcu_pending)
-		rcu_barrier_sched();
-	old = tracepoint_entry_add_probe(entry, probe);
-	if (IS_ERR(old)) {
-		ret = PTR_ERR(old);
-		goto end;
-	}
+	old = tracepoint_add_probe(name, probe);
 	mutex_unlock(&tracepoints_mutex);
+	if (IS_ERR(old))
+		return PTR_ERR(old);
+
 	tracepoint_update_probes();		/* may update entry */
-	mutex_lock(&tracepoints_mutex);
-	entry = get_tracepoint(name);
-	WARN_ON(!entry);
-	if (entry->rcu_pending)
-		rcu_barrier_sched();
-	tracepoint_entry_free_old(entry, old);
-end:
-	mutex_unlock(&tracepoints_mutex);
-	return ret;
+	release_probes(old);
+	return 0;
 }
 EXPORT_SYMBOL_GPL(tracepoint_probe_register);
 
+static void *tracepoint_remove_probe(const char *name, void *probe)
+{
+	struct tracepoint_entry *entry;
+	void *old;
+
+	entry = get_tracepoint(name);
+	if (!entry)
+		return ERR_PTR(-ENOENT);
+	old = tracepoint_entry_remove_probe(entry, probe);
+	if (IS_ERR(old))
+		return old;
+	if (!entry->refcount)
+		remove_tracepoint(entry);
+	return old;
+}
+
 /**
  * tracepoint_probe_unregister -  Disconnect a probe from a tracepoint
  * @name: tracepoint name
@@ -380,38 +371,104 @@ EXPORT_SYMBOL_GPL(tracepoint_probe_register);
  */
 int tracepoint_probe_unregister(const char *name, void *probe)
 {
-	struct tracepoint_entry *entry;
 	void *old;
-	int ret = -ENOENT;
 
 	mutex_lock(&tracepoints_mutex);
-	entry = get_tracepoint(name);
-	if (!entry)
-		goto end;
-	if (entry->rcu_pending)
-		rcu_barrier_sched();
-	old = tracepoint_entry_remove_probe(entry, probe);
-	if (!old) {
-		printk(KERN_WARNING "Warning: Trying to unregister a probe"
-				    "that doesn't exist\n");
-		goto end;
-	}
+	old = tracepoint_remove_probe(name, probe);
 	mutex_unlock(&tracepoints_mutex);
+	if (IS_ERR(old))
+		return PTR_ERR(old);
+
 	tracepoint_update_probes();		/* may update entry */
+	release_probes(old);
+	return 0;
+}
+EXPORT_SYMBOL_GPL(tracepoint_probe_unregister);
+
+static LIST_HEAD(old_probes);
+static int need_update;
+
+static void tracepoint_add_old_probes(void *old)
+{
+	need_update = 1;
+	if (old) {
+		struct tp_probes *tp_probes = container_of(old,
+			struct tp_probes, probes[0]);
+		list_add(&tp_probes->u.list, &old_probes);
+	}
+}
+
+/**
+ * tracepoint_probe_register_noupdate -  register a probe but not connect
+ * @name: tracepoint name
+ * @probe: probe handler
+ *
+ * caller must call tracepoint_probe_update_all()
+ */
+int tracepoint_probe_register_noupdate(const char *name, void *probe)
+{
+	void *old;
+
 	mutex_lock(&tracepoints_mutex);
-	entry = get_tracepoint(name);
-	if (!entry)
-		goto end;
-	if (entry->rcu_pending)
-		rcu_barrier_sched();
-	tracepoint_entry_free_old(entry, old);
-	remove_tracepoint(name);	/* Ignore busy error message */
-	ret = 0;
-end:
+	old = tracepoint_add_probe(name, probe);
+	if (IS_ERR(old)) {
+		mutex_unlock(&tracepoints_mutex);
+		return PTR_ERR(old);
+	}
+	tracepoint_add_old_probes(old);
 	mutex_unlock(&tracepoints_mutex);
-	return ret;
+	return 0;
 }
-EXPORT_SYMBOL_GPL(tracepoint_probe_unregister);
+EXPORT_SYMBOL_GPL(tracepoint_probe_register_noupdate);
+
+/**
+ * tracepoint_probe_unregister_noupdate -  remove a probe but not disconnect
+ * @name: tracepoint name
+ * @probe: probe function pointer
+ *
+ * caller must call tracepoint_probe_update_all()
+ */
+int tracepoint_probe_unregister_noupdate(const char *name, void *probe)
+{
+	void *old;
+
+	mutex_lock(&tracepoints_mutex);
+	old = tracepoint_remove_probe(name, probe);
+	if (IS_ERR(old)) {
+		mutex_unlock(&tracepoints_mutex);
+		return PTR_ERR(old);
+	}
+	tracepoint_add_old_probes(old);
+	mutex_unlock(&tracepoints_mutex);
+	return 0;
+}
+EXPORT_SYMBOL_GPL(tracepoint_probe_unregister_noupdate);
+
+/**
+ * tracepoint_probe_update_all -  update tracepoints
+ */
+void tracepoint_probe_update_all(void)
+{
+	LIST_HEAD(release_probes);
+	struct tp_probes *pos, *next;
+
+	mutex_lock(&tracepoints_mutex);
+	if (!need_update) {
+		mutex_unlock(&tracepoints_mutex);
+		return;
+	}
+	if (!list_empty(&old_probes))
+		list_replace_init(&old_probes, &release_probes);
+	need_update = 0;
+	mutex_unlock(&tracepoints_mutex);
+
+	tracepoint_update_probes();
+	list_for_each_entry_safe(pos, next, &release_probes, u.list) {
+		list_del(&pos->u.list);
+		call_rcu_sched(&pos->u.rcu, rcu_free_old_probes);
+	}
+}
+EXPORT_SYMBOL_GPL(tracepoint_probe_update_all);
 
 /**
  * tracepoint_get_iter_range - Get a next tracepoint iterator given a range.