Merge branch 'tracing-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip

* 'tracing-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip:
  kernel/profile.c: Switch /proc/irq/prof_cpu_mask to seq_file
  tracing: Export trace_profile_buf symbols
  tracing/events: use list_for_entry_continue
  tracing: remove max_tracer_type_len
  function-graph: use ftrace_graph_funcs directly
  tracing: Remove markers
  tracing: Allocate the ftrace event profile buffer dynamically
  tracing: Factorize the events profile accounting

6 files changed, 193 insertions, 108 deletions

diff --git a/kernel/trace/ftrace.c b/kernel/trace/ftrace.c
index cc615f84751..c71e91bf737 100644
--- a/kernel/trace/ftrace.c
+++ b/kernel/trace/ftrace.c
@@ -2414,11 +2414,9 @@ unsigned long ftrace_graph_funcs[FTRACE_GRAPH_MAX_FUNCS] __read_mostly;
 static void *
 __g_next(struct seq_file *m, loff_t *pos)
 {
-	unsigned long *array = m->private;
-
 	if (*pos >= ftrace_graph_count)
 		return NULL;
-	return &array[*pos];
+	return &ftrace_graph_funcs[*pos];
 }
 
 static void *
@@ -2482,16 +2480,10 @@ ftrace_graph_open(struct inode *inode, struct file *file)
 		ftrace_graph_count = 0;
 		memset(ftrace_graph_funcs, 0, sizeof(ftrace_graph_funcs));
 	}
+	mutex_unlock(&graph_lock);
 
-	if (file->f_mode & FMODE_READ) {
+	if (file->f_mode & FMODE_READ)
 		ret = seq_open(file, &ftrace_graph_seq_ops);
-		if (!ret) {
-			struct seq_file *m = file->private_data;
-			m->private = ftrace_graph_funcs;
-		}
-	} else
-		file->private_data = ftrace_graph_funcs;
-	mutex_unlock(&graph_lock);
 
 	return ret;
 }
@@ -2560,7 +2552,6 @@ ftrace_graph_write(struct file *file, const char __user *ubuf,
 		   size_t cnt, loff_t *ppos)
 {
 	struct trace_parser parser;
-	unsigned long *array;
 	size_t read = 0;
 	ssize_t ret;
 
@@ -2574,12 +2565,6 @@ ftrace_graph_write(struct file *file, const char __user *ubuf,
 		goto out;
 	}
 
-	if (file->f_mode & FMODE_READ) {
-		struct seq_file *m = file->private_data;
-		array = m->private;
-	} else
-		array = file->private_data;
-
 	if (trace_parser_get_init(&parser, FTRACE_BUFF_MAX)) {
 		ret = -ENOMEM;
 		goto out;
@@ -2591,7 +2576,7 @@ ftrace_graph_write(struct file *file, const char __user *ubuf,
 		parser.buffer[parser.idx] = 0;
 
 		/* we allow only one expression at a time */
-		ret = ftrace_set_func(array, &ftrace_graph_count,
+		ret = ftrace_set_func(ftrace_graph_funcs, &ftrace_graph_count,
 					parser.buffer);
 		if (ret)
 			goto out;
diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c
index fd52a19dd17..861308072d2 100644
--- a/kernel/trace/trace.c
+++ b/kernel/trace/trace.c
@@ -125,13 +125,13 @@ int ftrace_dump_on_oops;
 
 static int tracing_set_tracer(const char *buf);
 
-#define BOOTUP_TRACER_SIZE		100
-static char bootup_tracer_buf[BOOTUP_TRACER_SIZE] __initdata;
+#define MAX_TRACER_SIZE		100
+static char bootup_tracer_buf[MAX_TRACER_SIZE] __initdata;
 static char *default_bootup_tracer;
 
 static int __init set_ftrace(char *str)
 {
-	strncpy(bootup_tracer_buf, str, BOOTUP_TRACER_SIZE);
+	strncpy(bootup_tracer_buf, str, MAX_TRACER_SIZE);
 	default_bootup_tracer = bootup_tracer_buf;
 	/* We are using ftrace early, expand it */
 	ring_buffer_expanded = 1;
@@ -242,13 +242,6 @@ static struct tracer		*trace_types __read_mostly;
 static struct tracer		*current_trace __read_mostly;
 
 /*
- * max_tracer_type_len is used to simplify the allocating of
- * buffers to read userspace tracer names. We keep track of
- * the longest tracer name registered.
- */
-static int			max_tracer_type_len;
-
-/*
  * trace_types_lock is used to protect the trace_types list.
  * This lock is also used to keep user access serialized.
  * Accesses from userspace will grab this lock while userspace
@@ -619,7 +612,6 @@ __releases(kernel_lock)
 __acquires(kernel_lock)
 {
 	struct tracer *t;
-	int len;
 	int ret = 0;
 
 	if (!type->name) {
@@ -627,6 +619,11 @@ __acquires(kernel_lock)
 		return -1;
 	}
 
+	if (strlen(type->name) > MAX_TRACER_SIZE) {
+		pr_info("Tracer has a name longer than %d\n", MAX_TRACER_SIZE);
+		return -1;
+	}
+
 	/*
 	 * When this gets called we hold the BKL which means that
 	 * preemption is disabled. Various trace selftests however
@@ -641,7 +638,7 @@ __acquires(kernel_lock)
 	for (t = trace_types; t; t = t->next) {
 		if (strcmp(type->name, t->name) == 0) {
 			/* already found */
-			pr_info("Trace %s already registered\n",
+			pr_info("Tracer %s already registered\n",
 				type->name);
 			ret = -1;
 			goto out;
@@ -692,9 +689,6 @@ __acquires(kernel_lock)
 
 	type->next = trace_types;
 	trace_types = type;
-	len = strlen(type->name);
-	if (len > max_tracer_type_len)
-		max_tracer_type_len = len;
 
  out:
 	tracing_selftest_running = false;
@@ -703,7 +697,7 @@ __acquires(kernel_lock)
 	if (ret || !default_bootup_tracer)
 		goto out_unlock;
 
-	if (strncmp(default_bootup_tracer, type->name, BOOTUP_TRACER_SIZE))
+	if (strncmp(default_bootup_tracer, type->name, MAX_TRACER_SIZE))
 		goto out_unlock;
 
 	printk(KERN_INFO "Starting tracer '%s'\n", type->name);
@@ -725,14 +719,13 @@ __acquires(kernel_lock)
 void unregister_tracer(struct tracer *type)
 {
 	struct tracer **t;
-	int len;
 
 	mutex_lock(&trace_types_lock);
 	for (t = &trace_types; *t; t = &(*t)->next) {
 		if (*t == type)
 			goto found;
 	}
-	pr_info("Trace %s not registered\n", type->name);
+	pr_info("Tracer %s not registered\n", type->name);
 	goto out;
 
  found:
@@ -745,17 +738,7 @@ void unregister_tracer(struct tracer *type)
 			current_trace->stop(&global_trace);
 		current_trace = &nop_trace;
 	}
-
-	if (strlen(type->name) != max_tracer_type_len)
-		goto out;
-
-	max_tracer_type_len = 0;
-	for (t = &trace_types; *t; t = &(*t)->next) {
-		len = strlen((*t)->name);
-		if (len > max_tracer_type_len)
-			max_tracer_type_len = len;
-	}
- out:
+out:
 	mutex_unlock(&trace_types_lock);
 }
 
@@ -2604,7 +2587,7 @@ static ssize_t
 tracing_set_trace_read(struct file *filp, char __user *ubuf,
 		       size_t cnt, loff_t *ppos)
 {
-	char buf[max_tracer_type_len+2];
+	char buf[MAX_TRACER_SIZE+2];
 	int r;
 
 	mutex_lock(&trace_types_lock);
@@ -2754,15 +2737,15 @@ static ssize_t
 tracing_set_trace_write(struct file *filp, const char __user *ubuf,
 			size_t cnt, loff_t *ppos)
 {
-	char buf[max_tracer_type_len+1];
+	char buf[MAX_TRACER_SIZE+1];
 	int i;
 	size_t ret;
 	int err;
 
 	ret = cnt;
 
-	if (cnt > max_tracer_type_len)
-		cnt = max_tracer_type_len;
+	if (cnt > MAX_TRACER_SIZE)
+		cnt = MAX_TRACER_SIZE;
 
 	if (copy_from_user(&buf, ubuf, cnt))
 		return -EFAULT;
diff --git a/kernel/trace/trace_event_profile.c b/kernel/trace/trace_event_profile.c
index 55a25c933d1..dd44b876886 100644
--- a/kernel/trace/trace_event_profile.c
+++ b/kernel/trace/trace_event_profile.c
@@ -8,6 +8,57 @@
 #include <linux/module.h>
 #include "trace.h"
 
+/*
+ * We can't use a size but a type in alloc_percpu()
+ * So let's create a dummy type that matches the desired size
+ */
+typedef struct {char buf[FTRACE_MAX_PROFILE_SIZE];} profile_buf_t;
+
+char		*trace_profile_buf;
+EXPORT_SYMBOL_GPL(trace_profile_buf);
+
+char		*trace_profile_buf_nmi;
+EXPORT_SYMBOL_GPL(trace_profile_buf_nmi);
+
+/* Count the events in use (per event id, not per instance) */
+static int	total_profile_count;
+
+static int ftrace_profile_enable_event(struct ftrace_event_call *event)
+{
+	char *buf;
+	int ret = -ENOMEM;
+
+	if (atomic_inc_return(&event->profile_count))
+		return 0;
+
+	if (!total_profile_count++) {
+		buf = (char *)alloc_percpu(profile_buf_t);
+		if (!buf)
+			goto fail_buf;
+
+		rcu_assign_pointer(trace_profile_buf, buf);
+
+		buf = (char *)alloc_percpu(profile_buf_t);
+		if (!buf)
+			goto fail_buf_nmi;
+
+		rcu_assign_pointer(trace_profile_buf_nmi, buf);
+	}
+
+	ret = event->profile_enable();
+	if (!ret)
+		return 0;
+
+	kfree(trace_profile_buf_nmi);
+fail_buf_nmi:
+	kfree(trace_profile_buf);
+fail_buf:
+	total_profile_count--;
+	atomic_dec(&event->profile_count);
+
+	return ret;
+}
+
 int ftrace_profile_enable(int event_id)
 {
 	struct ftrace_event_call *event;
@@ -17,7 +68,7 @@ int ftrace_profile_enable(int event_id)
 	list_for_each_entry(event, &ftrace_events, list) {
 		if (event->id == event_id && event->profile_enable &&
 		    try_module_get(event->mod)) {
-			ret = event->profile_enable(event);
+			ret = ftrace_profile_enable_event(event);
 			break;
 		}
 	}
@@ -26,6 +77,33 @@ int ftrace_profile_enable(int event_id)
 	return ret;
 }
 
+static void ftrace_profile_disable_event(struct ftrace_event_call *event)
+{
+	char *buf, *nmi_buf;
+
+	if (!atomic_add_negative(-1, &event->profile_count))
+		return;
+
+	event->profile_disable();
+
+	if (!--total_profile_count) {
+		buf = trace_profile_buf;
+		rcu_assign_pointer(trace_profile_buf, NULL);
+
+		nmi_buf = trace_profile_buf_nmi;
+		rcu_assign_pointer(trace_profile_buf_nmi, NULL);
+
+		/*
+		 * Ensure every events in profiling have finished before
+		 * releasing the buffers
+		 */
+		synchronize_sched();
+
+		free_percpu(buf);
+		free_percpu(nmi_buf);
+	}
+}
+
 void ftrace_profile_disable(int event_id)
 {
 	struct ftrace_event_call *event;
@@ -33,7 +111,7 @@ void ftrace_profile_disable(int event_id)
 	mutex_lock(&event_mutex);
 	list_for_each_entry(event, &ftrace_events, list) {
 		if (event->id == event_id) {
-			event->profile_disable(event);
+			ftrace_profile_disable_event(event);
 			module_put(event->mod);
 			break;
 		}
diff --git a/kernel/trace/trace_events.c b/kernel/trace/trace_events.c
index 56c260b83a9..6f03c8a1105 100644
--- a/kernel/trace/trace_events.c
+++ b/kernel/trace/trace_events.c
@@ -271,42 +271,32 @@ ftrace_event_write(struct file *file, const char __user *ubuf,
 static void *
 t_next(struct seq_file *m, void *v, loff_t *pos)
 {
-	struct list_head *list = m->private;
-	struct ftrace_event_call *call;
+	struct ftrace_event_call *call = v;
 
 	(*pos)++;
 
-	for (;;) {
-		if (list == &ftrace_events)
-			return NULL;
-
-		call = list_entry(list, struct ftrace_event_call, list);
-
+	list_for_each_entry_continue(call, &ftrace_events, list) {
 		/*
 		 * The ftrace subsystem is for showing formats only.
 		 * They can not be enabled or disabled via the event files.
 		 */
 		if (call->regfunc)
-			break;
-
-		list = list->next;
+			return call;
 	}
 
-	m->private = list->next;
-
-	return call;
+	return NULL;
 }
 
 static void *t_start(struct seq_file *m, loff_t *pos)
 {
-	struct ftrace_event_call *call = NULL;
+	struct ftrace_event_call *call;
 	loff_t l;
 
 	mutex_lock(&event_mutex);
 
-	m->private = ftrace_events.next;
+	call = list_entry(&ftrace_events, struct ftrace_event_call, list);
 	for (l = 0; l <= *pos; ) {
-		call = t_next(m, NULL, &l);
+		call = t_next(m, call, &l);
 		if (!call)
 			break;
 	}
@@ -316,37 +306,28 @@ static void *t_start(struct seq_file *m, loff_t *pos)
 static void *
 s_next(struct seq_file *m, void *v, loff_t *pos)
 {
-	struct list_head *list = m->private;
-	struct ftrace_event_call *call;
+	struct ftrace_event_call *call = v;
 
 	(*pos)++;
 
- retry:
-	if (list == &ftrace_events)
-		return NULL;
-
-	call = list_entry(list, struct ftrace_event_call, list);
-
-	if (!call->enabled) {
-		list = list->next;
-		goto retry;
+	list_for_each_entry_continue(call, &ftrace_events, list) {
+		if (call->enabled)
+			return call;
 	}
 
-	m->private = list->next;
-
-	return call;
+	return NULL;
 }
 
 static void *s_start(struct seq_file *m, loff_t *pos)
 {
-	struct ftrace_event_call *call = NULL;
+	struct ftrace_event_call *call;
 	loff_t l;
 
 	mutex_lock(&event_mutex);
 
-	m->private = ftrace_events.next;
+	call = list_entry(&ftrace_events, struct ftrace_event_call, list);
 	for (l = 0; l <= *pos; ) {
-		call = s_next(m, NULL, &l);
+		call = s_next(m, call, &l);
 		if (!call)
 			break;
 	}
diff --git a/kernel/trace/trace_printk.c b/kernel/trace/trace_printk.c
index 687699d365a..2547d8813cf 100644
--- a/kernel/trace/trace_printk.c
+++ b/kernel/trace/trace_printk.c
@@ -11,7 +11,6 @@
 #include <linux/ftrace.h>
 #include <linux/string.h>
 #include <linux/module.h>
-#include <linux/marker.h>
 #include <linux/mutex.h>
 #include <linux/ctype.h>
 #include <linux/list.h>
diff --git a/kernel/trace/trace_syscalls.c b/kernel/trace/trace_syscalls.c
index 8712ce3c6a0..7a3550cf259 100644
--- a/kernel/trace/trace_syscalls.c
+++ b/kernel/trace/trace_syscalls.c
@@ -384,10 +384,13 @@ static int sys_prof_refcount_exit;
 
 static void prof_syscall_enter(struct pt_regs *regs, long id)
 {
-	struct syscall_trace_enter *rec;
 	struct syscall_metadata *sys_data;
+	struct syscall_trace_enter *rec;
+	unsigned long flags;
+	char *raw_data;
 	int syscall_nr;
 	int size;
+	int cpu;
 
 	syscall_nr = syscall_get_nr(current, regs);
 	if (!test_bit(syscall_nr, enabled_prof_enter_syscalls))
@@ -402,20 +405,38 @@ static void prof_syscall_enter(struct pt_regs *regs, long id)
 	size = ALIGN(size + sizeof(u32), sizeof(u64));
 	size -= sizeof(u32);
 
-	do {
-		char raw_data[size];
+	if (WARN_ONCE(size > FTRACE_MAX_PROFILE_SIZE,
+		      "profile buffer not large enough"))
+		return;
+
+	/* Protect the per cpu buffer, begin the rcu read side */
+	local_irq_save(flags);
 
-		/* zero the dead bytes from align to not leak stack to user */
-		*(u64 *)(&raw_data[size - sizeof(u64)]) = 0ULL;
+	cpu = smp_processor_id();
+
+	if (in_nmi())
+		raw_data = rcu_dereference(trace_profile_buf_nmi);
+	else
+		raw_data = rcu_dereference(trace_profile_buf);
+
+	if (!raw_data)
+		goto end;
 
-		rec = (struct syscall_trace_enter *) raw_data;
-		tracing_generic_entry_update(&rec->ent, 0, 0);
-		rec->ent.type = sys_data->enter_id;
-		rec->nr = syscall_nr;
-		syscall_get_arguments(current, regs, 0, sys_data->nb_args,
-				       (unsigned long *)&rec->args);
-		perf_tpcounter_event(sys_data->enter_id, 0, 1, rec, size);
-	} while(0);
+	raw_data = per_cpu_ptr(raw_data, cpu);
+
+	/* zero the dead bytes from align to not leak stack to user */
+	*(u64 *)(&raw_data[size - sizeof(u64)]) = 0ULL;
+
+	rec = (struct syscall_trace_enter *) raw_data;
+	tracing_generic_entry_update(&rec->ent, 0, 0);
+	rec->ent.type = sys_data->enter_id;
+	rec->nr = syscall_nr;
+	syscall_get_arguments(current, regs, 0, sys_data->nb_args,
+			       (unsigned long *)&rec->args);
+	perf_tpcounter_event(sys_data->enter_id, 0, 1, rec, size);
+
+end:
+	local_irq_restore(flags);
 }
 
 int reg_prof_syscall_enter(char *name)
@@ -460,8 +481,12 @@ void unreg_prof_syscall_enter(char *name)
 static void prof_syscall_exit(struct pt_regs *regs, long ret)
 {
 	struct syscall_metadata *sys_data;
-	struct syscall_trace_exit rec;
+	struct syscall_trace_exit *rec;
+	unsigned long flags;
 	int syscall_nr;
+	char *raw_data;
+	int size;
+	int cpu;
 
 	syscall_nr = syscall_get_nr(current, regs);
 	if (!test_bit(syscall_nr, enabled_prof_exit_syscalls))
@@ -471,12 +496,46 @@ static void prof_syscall_exit(struct pt_regs *regs, long ret)
 	if (!sys_data)
 		return;
 
-	tracing_generic_entry_update(&rec.ent, 0, 0);
-	rec.ent.type = sys_data->exit_id;
-	rec.nr = syscall_nr;
-	rec.ret = syscall_get_return_value(current, regs);
+	/* We can probably do that at build time */
+	size = ALIGN(sizeof(*rec) + sizeof(u32), sizeof(u64));
+	size -= sizeof(u32);
 
-	perf_tpcounter_event(sys_data->exit_id, 0, 1, &rec, sizeof(rec));
+	/*
+	 * Impossible, but be paranoid with the future
+	 * How to put this check outside runtime?
+	 */
+	if (WARN_ONCE(size > FTRACE_MAX_PROFILE_SIZE,
+		"exit event has grown above profile buffer size"))
+		return;
+
+	/* Protect the per cpu buffer, begin the rcu read side */
+	local_irq_save(flags);
+	cpu = smp_processor_id();
+
+	if (in_nmi())
+		raw_data = rcu_dereference(trace_profile_buf_nmi);
+	else
+		raw_data = rcu_dereference(trace_profile_buf);
+
+	if (!raw_data)
+		goto end;
+
+	raw_data = per_cpu_ptr(raw_data, cpu);
+
+	/* zero the dead bytes from align to not leak stack to user */
+	*(u64 *)(&raw_data[size - sizeof(u64)]) = 0ULL;
+
+	rec = (struct syscall_trace_exit *)raw_data;
+
+	tracing_generic_entry_update(&rec->ent, 0, 0);
+	rec->ent.type = sys_data->exit_id;
+	rec->nr = syscall_nr;
+	rec->ret = syscall_get_return_value(current, regs);
+
+	perf_tpcounter_event(sys_data->exit_id, 0, 1, rec, size);
+
+end:
+	local_irq_restore(flags);
 }
 
 int reg_prof_syscall_exit(char *name)