14 files changed, 1307 insertions, 531 deletions

diff --git a/drivers/oprofile/buffer_sync.c b/drivers/oprofile/buffer_sync.c
index 65e8294a9e2..d93b2b6b1f7 100644
--- a/drivers/oprofile/buffer_sync.c
+++ b/drivers/oprofile/buffer_sync.c
@@ -1,11 +1,12 @@
 /**
  * @file buffer_sync.c
  *
- * @remark Copyright 2002 OProfile authors
+ * @remark Copyright 2002-2009 OProfile authors
  * @remark Read the file COPYING
  *
  * @author John Levon <levon@movementarian.org>
  * @author Barry Kasindorf
+ * @author Robert Richter <robert.richter@amd.com>
  *
  * This is the core of the buffer management. Each
  * CPU buffer is processed and entered into the
@@ -29,6 +30,7 @@
 #include <linux/fs.h>
 #include <linux/oprofile.h>
 #include <linux/sched.h>
+#include <linux/gfp.h>
 
 #include "oprofile_stats.h"
 #include "event_buffer.h"
@@ -37,7 +39,7 @@
 
 static LIST_HEAD(dying_tasks);
 static LIST_HEAD(dead_tasks);
-static cpumask_t marked_cpus = CPU_MASK_NONE;
+static cpumask_var_t marked_cpus;
 static DEFINE_SPINLOCK(task_mortuary);
 static void process_task_mortuary(void);
 
@@ -139,21 +141,19 @@ static struct notifier_block module_load_nb = {
 	.notifier_call = module_load_notify,
 };
 
-
-static void end_sync(void)
+static void free_all_tasks(void)
 {
-	end_cpu_work();
 	/* make sure we don't leak task structs */
 	process_task_mortuary();
 	process_task_mortuary();
 }
 
-
 int sync_start(void)
 {
 	int err;
 
-	start_cpu_work();
+	if (!zalloc_cpumask_var(&marked_cpus, GFP_KERNEL))
+		return -ENOMEM;
 
 	err = task_handoff_register(&task_free_nb);
 	if (err)
@@ -168,6 +168,8 @@ int sync_start(void)
 	if (err)
 		goto out4;
 
+	start_cpu_work();
+
 out:
 	return err;
 out4:
@@ -176,19 +178,26 @@ out3:
 	profile_event_unregister(PROFILE_TASK_EXIT, &task_exit_nb);
 out2:
 	task_handoff_unregister(&task_free_nb);
+	free_all_tasks();
 out1:
-	end_sync();
+	free_cpumask_var(marked_cpus);
 	goto out;
 }
 
 
 void sync_stop(void)
 {
+	end_cpu_work();
 	unregister_module_notifier(&module_load_nb);
 	profile_event_unregister(PROFILE_MUNMAP, &munmap_nb);
 	profile_event_unregister(PROFILE_TASK_EXIT, &task_exit_nb);
 	task_handoff_unregister(&task_free_nb);
-	end_sync();
+	barrier();			/* do all of the above first */
+
+	flush_cpu_work();
+
+	free_all_tasks();
+	free_cpumask_var(marked_cpus);
 }
 
 
@@ -207,7 +216,7 @@ static inline unsigned long fast_get_dcookie(struct path *path)
 }
 
 
-/* Look up the dcookie for the task's first VM_EXECUTABLE mapping,
+/* Look up the dcookie for the task's mm->exe_file,
  * which corresponds loosely to "application name". This is
  * not strictly necessary but allows oprofile to associate
  * shared-library samples with particular applications
@@ -215,21 +224,10 @@ static inline unsigned long fast_get_dcookie(struct path *path)
 static unsigned long get_exec_dcookie(struct mm_struct *mm)
 {
 	unsigned long cookie = NO_COOKIE;
-	struct vm_area_struct *vma;
-
-	if (!mm)
-		goto out;
 
-	for (vma = mm->mmap; vma; vma = vma->vm_next) {
-		if (!vma->vm_file)
-			continue;
-		if (!(vma->vm_flags & VM_EXECUTABLE))
-			continue;
-		cookie = fast_get_dcookie(&vma->vm_file->f_path);
-		break;
-	}
+	if (mm && mm->exe_file)
+		cookie = fast_get_dcookie(&mm->exe_file->f_path);
 
-out:
 	return cookie;
 }
 
@@ -315,88 +313,73 @@ static void add_trace_begin(void)
 	add_event_entry(TRACE_BEGIN_CODE);
 }
 
-#ifdef CONFIG_OPROFILE_IBS
-
-#define IBS_FETCH_CODE_SIZE	2
-#define IBS_OP_CODE_SIZE	5
-
-/*
- * Add IBS fetch and op entries to event buffer
- */
-static void add_ibs_begin(int cpu, int code, struct mm_struct *mm)
+static void add_data(struct op_entry *entry, struct mm_struct *mm)
 {
-	unsigned long rip;
-	int i, count;
-	unsigned long ibs_cookie = 0;
+	unsigned long code, pc, val;
+	unsigned long cookie;
 	off_t offset;
-	struct op_sample *sample;
-
-	sample = cpu_buffer_read_entry(cpu);
-	if (!sample)
-		goto Error;
-	rip = sample->eip;
 
-#ifdef __LP64__
-	rip += sample->event << 32;
-#endif
+	if (!op_cpu_buffer_get_data(entry, &code))
+		return;
+	if (!op_cpu_buffer_get_data(entry, &pc))
+		return;
+	if (!op_cpu_buffer_get_size(entry))
+		return;
 
 	if (mm) {
-		ibs_cookie = lookup_dcookie(mm, rip, &offset);
+		cookie = lookup_dcookie(mm, pc, &offset);
 
-		if (ibs_cookie == NO_COOKIE)
-			offset = rip;
-		if (ibs_cookie == INVALID_COOKIE) {
+		if (cookie == NO_COOKIE)
+			offset = pc;
+		if (cookie == INVALID_COOKIE) {
 			atomic_inc(&oprofile_stats.sample_lost_no_mapping);
-			offset = rip;
+			offset = pc;
 		}
-		if (ibs_cookie != last_cookie) {
-			add_cookie_switch(ibs_cookie);
-			last_cookie = ibs_cookie;
+		if (cookie != last_cookie) {
+			add_cookie_switch(cookie);
+			last_cookie = cookie;
 		}
 	} else
-		offset = rip;
+		offset = pc;
 
 	add_event_entry(ESCAPE_CODE);
 	add_event_entry(code);
 	add_event_entry(offset);	/* Offset from Dcookie */
 
-	/* we send the Dcookie offset, but send the raw Linear Add also*/
-	add_event_entry(sample->eip);
-	add_event_entry(sample->event);
-
-	if (code == IBS_FETCH_CODE)
-		count = IBS_FETCH_CODE_SIZE;	/*IBS FETCH is 2 int64s*/
-	else
-		count = IBS_OP_CODE_SIZE;	/*IBS OP is 5 int64s*/
-
-	for (i = 0; i < count; i++) {
-		sample = cpu_buffer_read_entry(cpu);
-		if (!sample)
-			goto Error;
-		add_event_entry(sample->eip);
-		add_event_entry(sample->event);
-	}
-
-	return;
-
-Error:
-	return;
+	while (op_cpu_buffer_get_data(entry, &val))
+		add_event_entry(val);
 }
 
-#endif
-
-static void add_sample_entry(unsigned long offset, unsigned long event)
+static inline void add_sample_entry(unsigned long offset, unsigned long event)
 {
 	add_event_entry(offset);
 	add_event_entry(event);
 }
 
 
-static int add_us_sample(struct mm_struct *mm, struct op_sample *s)
+/*
+ * Add a sample to the global event buffer. If possible the
+ * sample is converted into a persistent dentry/offset pair
+ * for later lookup from userspace. Return 0 on failure.
+ */
+static int
+add_sample(struct mm_struct *mm, struct op_sample *s, int in_kernel)
 {
 	unsigned long cookie;
 	off_t offset;
 
+	if (in_kernel) {
+		add_sample_entry(s->eip, s->event);
+		return 1;
+	}
+
+	/* add userspace sample */
+
+	if (!mm) {
+		atomic_inc(&oprofile_stats.sample_lost_no_mm);
+		return 0;
+	}
+
 	cookie = lookup_dcookie(mm, s->eip, &offset);
 
 	if (cookie == INVALID_COOKIE) {
@@ -415,25 +398,6 @@ static int add_us_sample(struct mm_struct *mm, struct op_sample *s)
 }
 
 
-/* Add a sample to the global event buffer. If possible the
- * sample is converted into a persistent dentry/offset pair
- * for later lookup from userspace.
- */
-static int
-add_sample(struct mm_struct *mm, struct op_sample *s, int in_kernel)
-{
-	if (in_kernel) {
-		add_sample_entry(s->eip, s->event);
-		return 1;
-	} else if (mm) {
-		return add_us_sample(mm, s);
-	} else {
-		atomic_inc(&oprofile_stats.sample_lost_no_mm);
-	}
-	return 0;
-}
-
-
 static void release_mm(struct mm_struct *mm)
 {
 	if (!mm)
@@ -489,10 +453,10 @@ static void mark_done(int cpu)
 {
 	int i;
 
-	cpu_set(cpu, marked_cpus);
+	cpumask_set_cpu(cpu, marked_cpus);
 
 	for_each_online_cpu(i) {
-		if (!cpu_isset(i, marked_cpus))
+		if (!cpumask_test_cpu(i, marked_cpus))
 			return;
 	}
 
@@ -501,7 +465,7 @@ static void mark_done(int cpu)
 	 */
 	process_task_mortuary();
 
-	cpus_clear(marked_cpus);
+	cpumask_clear(marked_cpus);
 }
 
 
@@ -526,66 +490,69 @@ void sync_buffer(int cpu)
 {
 	struct mm_struct *mm = NULL;
 	struct mm_struct *oldmm;
+	unsigned long val;
 	struct task_struct *new;
 	unsigned long cookie = 0;
 	int in_kernel = 1;
 	sync_buffer_state state = sb_buffer_start;
 	unsigned int i;
 	unsigned long available;
+	unsigned long flags;
+	struct op_entry entry;
+	struct op_sample *sample;
 
 	mutex_lock(&buffer_mutex);
 
 	add_cpu_switch(cpu);
 
-	cpu_buffer_reset(cpu);
-	available = cpu_buffer_entries(cpu);
+	op_cpu_buffer_reset(cpu);
+	available = op_cpu_buffer_entries(cpu);
 
 	for (i = 0; i < available; ++i) {
-		struct op_sample *s = cpu_buffer_read_entry(cpu);
-		if (!s)
+		sample = op_cpu_buffer_read_entry(&entry, cpu);
+		if (!sample)
 			break;
 
-		if (is_code(s->eip)) {
-			switch (s->event) {
-			case 0:
-			case CPU_IS_KERNEL:
+		if (is_code(sample->eip)) {
+			flags = sample->event;
+			if (flags & TRACE_BEGIN) {
+				state = sb_bt_start;
+				add_trace_begin();
+			}
+			if (flags & KERNEL_CTX_SWITCH) {
 				/* kernel/userspace switch */
-				in_kernel = s->event;
+				in_kernel = flags & IS_KERNEL;
 				if (state == sb_buffer_start)
 					state = sb_sample_start;
-				add_kernel_ctx_switch(s->event);
-				break;
-			case CPU_TRACE_BEGIN:
-				state = sb_bt_start;
-				add_trace_begin();
-				break;
-#ifdef CONFIG_OPROFILE_IBS
-			case IBS_FETCH_BEGIN:
-				state = sb_bt_start;
-				add_ibs_begin(cpu, IBS_FETCH_CODE, mm);
-				break;
-			case IBS_OP_BEGIN:
-				state = sb_bt_start;
-				add_ibs_begin(cpu, IBS_OP_CODE, mm);
-				break;
-#endif
-			default:
+				add_kernel_ctx_switch(flags & IS_KERNEL);
+			}
+			if (flags & USER_CTX_SWITCH
+			    && op_cpu_buffer_get_data(&entry, &val)) {
 				/* userspace context switch */
+				new = (struct task_struct *)val;
 				oldmm = mm;
-				new = (struct task_struct *)s->event;
 				release_mm(oldmm);
 				mm = take_tasks_mm(new);
 				if (mm != oldmm)
 					cookie = get_exec_dcookie(mm);
 				add_user_ctx_switch(new, cookie);
-				break;
-			}
-		} else if (state >= sb_bt_start &&
-			   !add_sample(mm, s, in_kernel)) {
-			if (state == sb_bt_start) {
-				state = sb_bt_ignore;
-				atomic_inc(&oprofile_stats.bt_lost_no_mapping);
 			}
+			if (op_cpu_buffer_get_size(&entry))
+				add_data(&entry, mm);
+			continue;
+		}
+
+		if (state < sb_bt_start)
+			/* ignore sample */
+			continue;
+
+		if (add_sample(mm, sample, in_kernel))
+			continue;
+
+		/* ignore backtraces if failed to add a sample */
+		if (state == sb_bt_start) {
+			state = sb_bt_ignore;
+			atomic_inc(&oprofile_stats.bt_lost_no_mapping);
 		}
 	}
 	release_mm(mm);
diff --git a/drivers/oprofile/cpu_buffer.c b/drivers/oprofile/cpu_buffer.c
index 61090969158..8aa73fac6ad 100644
--- a/drivers/oprofile/cpu_buffer.c
+++ b/drivers/oprofile/cpu_buffer.c
@@ -1,11 +1,12 @@
 /**
  * @file cpu_buffer.c
  *
- * @remark Copyright 2002 OProfile authors
+ * @remark Copyright 2002-2009 OProfile authors
  * @remark Read the file COPYING
  *
  * @author John Levon <levon@movementarian.org>
  * @author Barry Kasindorf <barry.kasindorf@amd.com>
+ * @author Robert Richter <robert.richter@amd.com>
  *
  * Each CPU has a local buffer that stores PC value/event
  * pairs. We also log context switches when we notice them.
@@ -20,7 +21,6 @@
 
 #include <linux/sched.h>
 #include <linux/oprofile.h>
-#include <linux/vmalloc.h>
 #include <linux/errno.h>
 
 #include "event_buffer.h"
@@ -30,75 +30,54 @@
 
 #define OP_BUFFER_FLAGS	0
 
-/*
- * Read and write access is using spin locking. Thus, writing to the
- * buffer by NMI handler (x86) could occur also during critical
- * sections when reading the buffer. To avoid this, there are 2
- * buffers for independent read and write access. Read access is in
- * process context only, write access only in the NMI handler. If the
- * read buffer runs empty, both buffers are swapped atomically. There
- * is potentially a small window during swapping where the buffers are
- * disabled and samples could be lost.
- *
- * Using 2 buffers is a little bit overhead, but the solution is clear
- * and does not require changes in the ring buffer implementation. It
- * can be changed to a single buffer solution when the ring buffer
- * access is implemented as non-locking atomic code.
- */
-struct ring_buffer *op_ring_buffer_read;
-struct ring_buffer *op_ring_buffer_write;
-DEFINE_PER_CPU(struct oprofile_cpu_buffer, cpu_buffer);
+static struct ring_buffer *op_ring_buffer;
+DEFINE_PER_CPU(struct oprofile_cpu_buffer, op_cpu_buffer);
 
 static void wq_sync_buffer(struct work_struct *work);
 
 #define DEFAULT_TIMER_EXPIRE (HZ / 10)
 static int work_enabled;
 
-void free_cpu_buffers(void)
-{
-	if (op_ring_buffer_read)
-		ring_buffer_free(op_ring_buffer_read);
-	op_ring_buffer_read = NULL;
-	if (op_ring_buffer_write)
-		ring_buffer_free(op_ring_buffer_write);
-	op_ring_buffer_write = NULL;
-}
-
 unsigned long oprofile_get_cpu_buffer_size(void)
 {
-	return fs_cpu_buffer_size;
+	return oprofile_cpu_buffer_size;
 }
 
 void oprofile_cpu_buffer_inc_smpl_lost(void)
 {
-	struct oprofile_cpu_buffer *cpu_buf
-		= &__get_cpu_var(cpu_buffer);
+	struct oprofile_cpu_buffer *cpu_buf = &__get_cpu_var(op_cpu_buffer);
 
 	cpu_buf->sample_lost_overflow++;
 }
 
+void free_cpu_buffers(void)
+{
+	if (op_ring_buffer)
+		ring_buffer_free(op_ring_buffer);
+	op_ring_buffer = NULL;
+}
+
+#define RB_EVENT_HDR_SIZE 4
+
 int alloc_cpu_buffers(void)
 {
 	int i;
 
-	unsigned long buffer_size = fs_cpu_buffer_size;
+	unsigned long buffer_size = oprofile_cpu_buffer_size;
+	unsigned long byte_size = buffer_size * (sizeof(struct op_sample) +
+						 RB_EVENT_HDR_SIZE);
 
-	op_ring_buffer_read = ring_buffer_alloc(buffer_size, OP_BUFFER_FLAGS);
-	if (!op_ring_buffer_read)
-		goto fail;
-	op_ring_buffer_write = ring_buffer_alloc(buffer_size, OP_BUFFER_FLAGS);
-	if (!op_ring_buffer_write)
+	op_ring_buffer = ring_buffer_alloc(byte_size, OP_BUFFER_FLAGS);
+	if (!op_ring_buffer)
 		goto fail;
 
 	for_each_possible_cpu(i) {
-		struct oprofile_cpu_buffer *b = &per_cpu(cpu_buffer, i);
+		struct oprofile_cpu_buffer *b = &per_cpu(op_cpu_buffer, i);
 
 		b->last_task = NULL;
 		b->last_is_kernel = -1;
 		b->tracing = 0;
 		b->buffer_size = buffer_size;
-		b->tail_pos = 0;
-		b->head_pos = 0;
 		b->sample_received = 0;
 		b->sample_lost_overflow = 0;
 		b->backtrace_aborted = 0;
@@ -120,7 +99,7 @@ void start_cpu_work(void)
 	work_enabled = 1;
 
 	for_each_online_cpu(i) {
-		struct oprofile_cpu_buffer *b = &per_cpu(cpu_buffer, i);
+		struct oprofile_cpu_buffer *b = &per_cpu(op_cpu_buffer, i);
 
 		/*
 		 * Spread the work by 1 jiffy per cpu so they dont all
@@ -132,60 +111,157 @@ void start_cpu_work(void)
 
 void end_cpu_work(void)
 {
-	int i;
-
 	work_enabled = 0;
+}
+
+void flush_cpu_work(void)
+{
+	int i;
 
 	for_each_online_cpu(i) {
-		struct oprofile_cpu_buffer *b = &per_cpu(cpu_buffer, i);
+		struct oprofile_cpu_buffer *b = &per_cpu(op_cpu_buffer, i);
 
-		cancel_delayed_work(&b->work);
+		/* these works are per-cpu, no need for flush_sync */
+		flush_delayed_work(&b->work);
 	}
+}
 
-	flush_scheduled_work();
+/*
+ * This function prepares the cpu buffer to write a sample.
+ *
+ * Struct op_entry is used during operations on the ring buffer while
+ * struct op_sample contains the data that is stored in the ring
+ * buffer. Struct entry can be uninitialized. The function reserves a
+ * data array that is specified by size. Use
+ * op_cpu_buffer_write_commit() after preparing the sample. In case of
+ * errors a null pointer is returned, otherwise the pointer to the
+ * sample.
+ *
+ */
+struct op_sample
+*op_cpu_buffer_write_reserve(struct op_entry *entry, unsigned long size)
+{
+	entry->event = ring_buffer_lock_reserve
+		(op_ring_buffer, sizeof(struct op_sample) +
+		 size * sizeof(entry->sample->data[0]));
+	if (!entry->event)
+		return NULL;
+	entry->sample = ring_buffer_event_data(entry->event);
+	entry->size = size;
+	entry->data = entry->sample->data;
+
+	return entry->sample;
 }
 
-static inline int
-add_sample(struct oprofile_cpu_buffer *cpu_buf,
-	   unsigned long pc, unsigned long event)
+int op_cpu_buffer_write_commit(struct op_entry *entry)
+{
+	return ring_buffer_unlock_commit(op_ring_buffer, entry->event);
+}
+
+struct op_sample *op_cpu_buffer_read_entry(struct op_entry *entry, int cpu)
+{
+	struct ring_buffer_event *e;
+	e = ring_buffer_consume(op_ring_buffer, cpu, NULL, NULL);
+	if (!e)
+		return NULL;
+
+	entry->event = e;
+	entry->sample = ring_buffer_event_data(e);
+	entry->size = (ring_buffer_event_length(e) - sizeof(struct op_sample))
+		/ sizeof(entry->sample->data[0]);
+	entry->data = entry->sample->data;
+	return entry->sample;
+}
+
+unsigned long op_cpu_buffer_entries(int cpu)
+{
+	return ring_buffer_entries_cpu(op_ring_buffer, cpu);
+}
+
+static int
+op_add_code(struct oprofile_cpu_buffer *cpu_buf, unsigned long backtrace,
+	    int is_kernel, struct task_struct *task)
 {
 	struct op_entry entry;
-	int ret;
+	struct op_sample *sample;
+	unsigned long flags;
+	int size;
+
+	flags = 0;
 
-	ret = cpu_buffer_write_entry(&entry);
-	if (ret)
-		return ret;
+	if (backtrace)
+		flags |= TRACE_BEGIN;
 
-	entry.sample->eip = pc;
-	entry.sample->event = event;
+	/* notice a switch from user->kernel or vice versa */
+	is_kernel = !!is_kernel;
+	if (cpu_buf->last_is_kernel != is_kernel) {
+		cpu_buf->last_is_kernel = is_kernel;
+		flags |= KERNEL_CTX_SWITCH;
+		if (is_kernel)
+			flags |= IS_KERNEL;
+	}
 
-	ret = cpu_buffer_write_commit(&entry);
-	if (ret)
-		return ret;
+	/* notice a task switch */
+	if (cpu_buf->last_task != task) {
+		cpu_buf->last_task = task;
+		flags |= USER_CTX_SWITCH;
+	}
+
+	if (!flags)
+		/* nothing to do */
+		return 0;
+
+	if (flags & USER_CTX_SWITCH)
+		size = 1;
+	else
+		size = 0;
+
+	sample = op_cpu_buffer_write_reserve(&entry, size);
+	if (!sample)
+		return -ENOMEM;
+
+	sample->eip = ESCAPE_CODE;
+	sample->event = flags;
+
+	if (size)
+		op_cpu_buffer_add_data(&entry, (unsigned long)task);
+
+	op_cpu_buffer_write_commit(&entry);
 
 	return 0;
 }
 
 static inline int
-add_code(struct oprofile_cpu_buffer *buffer, unsigned long value)
+op_add_sample(struct oprofile_cpu_buffer *cpu_buf,
+	      unsigned long pc, unsigned long event)
 {
-	return add_sample(buffer, ESCAPE_CODE, value);
+	struct op_entry entry;
+	struct op_sample *sample;
+
+	sample = op_cpu_buffer_write_reserve(&entry, 0);
+	if (!sample)
+		return -ENOMEM;
+
+	sample->eip = pc;
+	sample->event = event;
+
+	return op_cpu_buffer_write_commit(&entry);
 }
 
-/* This must be safe from any context. It's safe writing here
- * because of the head/tail separation of the writer and reader
- * of the CPU buffer.
+/*
+ * This must be safe from any context.
  *
  * is_kernel is needed because on some architectures you cannot
  * tell if you are in kernel or user space simply by looking at
  * pc. We tag this in the buffer by generating kernel enter/exit
  * events whenever is_kernel changes
  */
-static int log_sample(struct oprofile_cpu_buffer *cpu_buf, unsigned long pc,
-		      int is_kernel, unsigned long event)
+static int
+log_sample(struct oprofile_cpu_buffer *cpu_buf, unsigned long pc,
+	   unsigned long backtrace, int is_kernel, unsigned long event,
+	   struct task_struct *task)
 {
-	struct task_struct *task;
-
+	struct task_struct *tsk = task ? task : current;
 	cpu_buf->sample_received++;
 
 	if (pc == ESCAPE_CODE) {
@@ -193,25 +269,10 @@ static int log_sample(struct oprofile_cpu_buffer *cpu_buf, unsigned long pc,
 		return 0;
 	}
 
-	is_kernel = !!is_kernel;
-
-	task = current;
-
-	/* notice a switch from user->kernel or vice versa */
-	if (cpu_buf->last_is_kernel != is_kernel) {
-		cpu_buf->last_is_kernel = is_kernel;
-		if (add_code(cpu_buf, is_kernel))
-			goto fail;
-	}
-
-	/* notice a task switch */
-	if (cpu_buf->last_task != task) {
-		cpu_buf->last_task = task;
-		if (add_code(cpu_buf, (unsigned long)task))
-			goto fail;
-	}
+	if (op_add_code(cpu_buf, backtrace, is_kernel, tsk))
+		goto fail;
 
-	if (add_sample(cpu_buf, pc, event))
+	if (op_add_sample(cpu_buf, pc, event))
 		goto fail;
 
 	return 1;
@@ -221,114 +282,143 @@ fail:
 	return 0;
 }
 
-static int oprofile_begin_trace(struct oprofile_cpu_buffer *cpu_buf)
+static inline void oprofile_begin_trace(struct oprofile_cpu_buffer *cpu_buf)
 {
-	add_code(cpu_buf, CPU_TRACE_BEGIN);
 	cpu_buf->tracing = 1;
-	return 1;
 }
 
-static void oprofile_end_trace(struct oprofile_cpu_buffer *cpu_buf)
+static inline void oprofile_end_trace(struct oprofile_cpu_buffer *cpu_buf)
 {
 	cpu_buf->tracing = 0;
 }
 
-void oprofile_add_ext_sample(unsigned long pc, struct pt_regs * const regs,
-				unsigned long event, int is_kernel)
+static inline void
+__oprofile_add_ext_sample(unsigned long pc, struct pt_regs * const regs,
+			  unsigned long event, int is_kernel,
+			  struct task_struct *task)
 {
-	struct oprofile_cpu_buffer *cpu_buf = &__get_cpu_var(cpu_buffer);
-
-	if (!backtrace_depth) {
-		log_sample(cpu_buf, pc, is_kernel, event);
-		return;
-	}
-
-	if (!oprofile_begin_trace(cpu_buf))
-		return;
+	struct oprofile_cpu_buffer *cpu_buf = &__get_cpu_var(op_cpu_buffer);
+	unsigned long backtrace = oprofile_backtrace_depth;
 
 	/*
 	 * if log_sample() fail we can't backtrace since we lost the
 	 * source of this event
 	 */
-	if (log_sample(cpu_buf, pc, is_kernel, event))
-		oprofile_ops.backtrace(regs, backtrace_depth);
+	if (!log_sample(cpu_buf, pc, backtrace, is_kernel, event, task))
+		/* failed */
+		return;
+
+	if (!backtrace)
+		return;
+
+	oprofile_begin_trace(cpu_buf);
+	oprofile_ops.backtrace(regs, backtrace);
 	oprofile_end_trace(cpu_buf);
 }
 
-void oprofile_add_sample(struct pt_regs * const regs, unsigned long event)
+void oprofile_add_ext_hw_sample(unsigned long pc, struct pt_regs * const regs,
+				unsigned long event, int is_kernel,
+				struct task_struct *task)
 {
-	int is_kernel = !user_mode(regs);
-	unsigned long pc = profile_pc(regs);
+	__oprofile_add_ext_sample(pc, regs, event, is_kernel, task);
+}
 
-	oprofile_add_ext_sample(pc, regs, event, is_kernel);
+void oprofile_add_ext_sample(unsigned long pc, struct pt_regs * const regs,
+			     unsigned long event, int is_kernel)
+{
+	__oprofile_add_ext_sample(pc, regs, event, is_kernel, NULL);
 }
 
-#ifdef CONFIG_OPROFILE_IBS
+void oprofile_add_sample(struct pt_regs * const regs, unsigned long event)
+{
+	int is_kernel;
+	unsigned long pc;
+
+	if (likely(regs)) {
+		is_kernel = !user_mode(regs);
+		pc = profile_pc(regs);
+	} else {
+		is_kernel = 0;    /* This value will not be used */
+		pc = ESCAPE_CODE; /* as this causes an early return. */
+	}
 
-#define MAX_IBS_SAMPLE_SIZE 14
+	__oprofile_add_ext_sample(pc, regs, event, is_kernel, NULL);
+}
 
-void oprofile_add_ibs_sample(struct pt_regs * const regs,
-			     unsigned int * const ibs_sample, int ibs_code)
+/*
+ * Add samples with data to the ring buffer.
+ *
+ * Use oprofile_add_data(&entry, val) to add data and
+ * oprofile_write_commit(&entry) to commit the sample.
+ */
+void
+oprofile_write_reserve(struct op_entry *entry, struct pt_regs * const regs,
+		       unsigned long pc, int code, int size)
 {
+	struct op_sample *sample;
 	int is_kernel = !user_mode(regs);
-	struct oprofile_cpu_buffer *cpu_buf = &__get_cpu_var(cpu_buffer);
-	struct task_struct *task;
-	int fail = 0;
+	struct oprofile_cpu_buffer *cpu_buf = &__get_cpu_var(op_cpu_buffer);
 
 	cpu_buf->sample_received++;
 
-	/* notice a switch from user->kernel or vice versa */
-	if (cpu_buf->last_is_kernel != is_kernel) {
-		if (add_code(cpu_buf, is_kernel))
-			goto fail;
-		cpu_buf->last_is_kernel = is_kernel;
-	}
-
-	/* notice a task switch */
-	if (!is_kernel) {
-		task = current;
-		if (cpu_buf->last_task != task) {
-			if (add_code(cpu_buf, (unsigned long)task))
-				goto fail;
-			cpu_buf->last_task = task;
-		}
-	}
-
-	fail = fail || add_code(cpu_buf, ibs_code);
-	fail = fail || add_sample(cpu_buf, ibs_sample[0], ibs_sample[1]);
-	fail = fail || add_sample(cpu_buf, ibs_sample[2], ibs_sample[3]);
-	fail = fail || add_sample(cpu_buf, ibs_sample[4], ibs_sample[5]);
-
-	if (ibs_code == IBS_OP_BEGIN) {
-		fail = fail || add_sample(cpu_buf, ibs_sample[6], ibs_sample[7]);
-		fail = fail || add_sample(cpu_buf, ibs_sample[8], ibs_sample[9]);
-		fail = fail || add_sample(cpu_buf, ibs_sample[10], ibs_sample[11]);
-	}
+	/* no backtraces for samples with data */
+	if (op_add_code(cpu_buf, 0, is_kernel, current))
+		goto fail;
 
-	if (fail)
+	sample = op_cpu_buffer_write_reserve(entry, size + 2);
+	if (!sample)
 		goto fail;
+	sample->eip = ESCAPE_CODE;
+	sample->event = 0;		/* no flags */
 
-	if (backtrace_depth)
-		oprofile_ops.backtrace(regs, backtrace_depth);
+	op_cpu_buffer_add_data(entry, code);
+	op_cpu_buffer_add_data(entry, pc);
 
 	return;
 
 fail:
+	entry->event = NULL;
 	cpu_buf->sample_lost_overflow++;
-	return;
 }
 
-#endif
+int oprofile_add_data(struct op_entry *entry, unsigned long val)
+{
+	if (!entry->event)
+		return 0;
+	return op_cpu_buffer_add_data(entry, val);
+}
+
+int oprofile_add_data64(struct op_entry *entry, u64 val)
+{
+	if (!entry->event)
+		return 0;
+	if (op_cpu_buffer_get_size(entry) < 2)
+		/*
+		 * the function returns 0 to indicate a too small
+		 * buffer, even if there is some space left
+		 */
+		return 0;
+	if (!op_cpu_buffer_add_data(entry, (u32)val))
+		return 0;
+	return op_cpu_buffer_add_data(entry, (u32)(val >> 32));
+}
+
+int oprofile_write_commit(struct op_entry *entry)
+{
+	if (!entry->event)
+		return -EINVAL;
+	return op_cpu_buffer_write_commit(entry);
+}
 
 void oprofile_add_pc(unsigned long pc, int is_kernel, unsigned long event)
 {
-	struct oprofile_cpu_buffer *cpu_buf = &__get_cpu_var(cpu_buffer);
-	log_sample(cpu_buf, pc, is_kernel, event);
+	struct oprofile_cpu_buffer *cpu_buf = &__get_cpu_var(op_cpu_buffer);
+	log_sample(cpu_buf, pc, 0, is_kernel, event, NULL);
 }
 
 void oprofile_add_trace(unsigned long pc)
 {
-	struct oprofile_cpu_buffer *cpu_buf = &__get_cpu_var(cpu_buffer);
+	struct oprofile_cpu_buffer *cpu_buf = &__get_cpu_var(op_cpu_buffer);
 
 	if (!cpu_buf->tracing)
 		return;
@@ -340,7 +430,7 @@ void oprofile_add_trace(unsigned long pc)
 	if (pc == ESCAPE_CODE)
 		goto fail;
 
-	if (add_sample(cpu_buf, pc, 0))
+	if (op_add_sample(cpu_buf, pc, 0))
 		goto fail;
 
 	return;
@@ -361,14 +451,9 @@ static void wq_sync_buffer(struct work_struct *work)
 {
 	struct oprofile_cpu_buffer *b =
 		container_of(work, struct oprofile_cpu_buffer, work.work);
-	if (b->cpu != smp_processor_id()) {
-		printk(KERN_DEBUG "WQ on CPU%d, prefer CPU%d\n",
-		       smp_processor_id(), b->cpu);
-
-		if (!cpu_online(b->cpu)) {
-			cancel_delayed_work(&b->work);
-			return;
-		}
+	if (b->cpu != smp_processor_id() && !cpu_online(b->cpu)) {
+		cancel_delayed_work(&b->work);
+		return;
 	}
 	sync_buffer(b->cpu);
 
diff --git a/drivers/oprofile/cpu_buffer.h b/drivers/oprofile/cpu_buffer.h
index aacb0f0bc56..e1d097e250a 100644
--- a/drivers/oprofile/cpu_buffer.h
+++ b/drivers/oprofile/cpu_buffer.h
@@ -1,10 +1,11 @@
 /**
  * @file cpu_buffer.h
  *
- * @remark Copyright 2002 OProfile authors
+ * @remark Copyright 2002-2009 OProfile authors
  * @remark Read the file COPYING
  *
  * @author John Levon <levon@movementarian.org>
+ * @author Robert Richter <robert.richter@amd.com>
  */
 
 #ifndef OPROFILE_CPU_BUFFER_H
@@ -24,6 +25,7 @@ void free_cpu_buffers(void);
 
 void start_cpu_work(void);
 void end_cpu_work(void);
+void flush_cpu_work(void);
 
 /* CPU buffer is composed of such entries (which are
  * also used for context switch notes)
@@ -31,17 +33,12 @@ void end_cpu_work(void);
 struct op_sample {
 	unsigned long eip;
 	unsigned long event;
+	unsigned long data[0];
 };
 
-struct op_entry {
-	struct ring_buffer_event *event;
-	struct op_sample *sample;
-	unsigned long irq_flags;
-};
+struct op_entry;
 
 struct oprofile_cpu_buffer {
-	volatile unsigned long head_pos;
-	volatile unsigned long tail_pos;
 	unsigned long buffer_size;
 	struct task_struct *last_task;
 	int last_is_kernel;
@@ -54,9 +51,7 @@ struct oprofile_cpu_buffer {
 	struct delayed_work work;
 };
 
-extern struct ring_buffer *op_ring_buffer_read;
-extern struct ring_buffer *op_ring_buffer_write;
-DECLARE_PER_CPU(struct oprofile_cpu_buffer, cpu_buffer);
+DECLARE_PER_CPU(struct oprofile_cpu_buffer, op_cpu_buffer);
 
 /*
  * Resets the cpu buffer to a sane state.
@@ -64,63 +59,63 @@ DECLARE_PER_CPU(struct oprofile_cpu_buffer, cpu_buffer);
  * reset these to invalid values; the next sample collected will
  * populate the buffer with proper values to initialize the buffer
  */
-static inline void cpu_buffer_reset(int cpu)
+static inline void op_cpu_buffer_reset(int cpu)
 {
-	struct oprofile_cpu_buffer *cpu_buf = &per_cpu(cpu_buffer, cpu);
+	struct oprofile_cpu_buffer *cpu_buf = &per_cpu(op_cpu_buffer, cpu);
 
 	cpu_buf->last_is_kernel = -1;
 	cpu_buf->last_task = NULL;
 }
 
-static inline int cpu_buffer_write_entry(struct op_entry *entry)
-{
-	entry->event = ring_buffer_lock_reserve(op_ring_buffer_write,
-						sizeof(struct op_sample),
-						&entry->irq_flags);
-	if (entry->event)
-		entry->sample = ring_buffer_event_data(entry->event);
-	else
-		entry->sample = NULL;
-
-	if (!entry->sample)
-		return -ENOMEM;
-
-	return 0;
-}
+/*
+ * op_cpu_buffer_add_data() and op_cpu_buffer_write_commit() may be
+ * called only if op_cpu_buffer_write_reserve() did not return NULL or
+ * entry->event != NULL, otherwise entry->size or entry->event will be
+ * used uninitialized.
+ */
+
+struct op_sample
+*op_cpu_buffer_write_reserve(struct op_entry *entry, unsigned long size);
+int op_cpu_buffer_write_commit(struct op_entry *entry);
+struct op_sample *op_cpu_buffer_read_entry(struct op_entry *entry, int cpu);
+unsigned long op_cpu_buffer_entries(int cpu);
 
-static inline int cpu_buffer_write_commit(struct op_entry *entry)
+/* returns the remaining free size of data in the entry */
+static inline
+int op_cpu_buffer_add_data(struct op_entry *entry, unsigned long val)
 {
-	return ring_buffer_unlock_commit(op_ring_buffer_write, entry->event,
-					 entry->irq_flags);
+	if (!entry->size)
+		return 0;
+	*entry->data = val;
+	entry->size--;
+	entry->data++;
+	return entry->size;
 }
 
-static inline struct op_sample *cpu_buffer_read_entry(int cpu)
+/* returns the size of data in the entry */
+static inline
+int op_cpu_buffer_get_size(struct op_entry *entry)
 {
-	struct ring_buffer_event *e;
-	e = ring_buffer_consume(op_ring_buffer_read, cpu, NULL);
-	if (e)
-		return ring_buffer_event_data(e);
-	if (ring_buffer_swap_cpu(op_ring_buffer_read,
-				 op_ring_buffer_write,
-				 cpu))
-		return NULL;
-	e = ring_buffer_consume(op_ring_buffer_read, cpu, NULL);
-	if (e)
-		return ring_buffer_event_data(e);
-	return NULL;
+	return entry->size;
 }
 
-/* "acquire" as many cpu buffer slots as we can */
-static inline unsigned long cpu_buffer_entries(int cpu)
+/* returns 0 if empty or the size of data including the current value */
+static inline
+int op_cpu_buffer_get_data(struct op_entry *entry, unsigned long *val)
 {
-	return ring_buffer_entries_cpu(op_ring_buffer_read, cpu)
-		+ ring_buffer_entries_cpu(op_ring_buffer_write, cpu);
+	int size = entry->size;
+	if (!size)
+		return 0;
+	*val = *entry->data;
+	entry->size--;
+	entry->data++;
+	return size;
 }
 
-/* transient events for the CPU buffer -> event buffer */
-#define CPU_IS_KERNEL 1
-#define CPU_TRACE_BEGIN 2
-#define IBS_FETCH_BEGIN 3
-#define IBS_OP_BEGIN    4
+/* extra data flags */
+#define KERNEL_CTX_SWITCH	(1UL << 0)
+#define IS_KERNEL		(1UL << 1)
+#define TRACE_BEGIN		(1UL << 2)
+#define USER_CTX_SWITCH		(1UL << 3)
 
 #endif /* OPROFILE_CPU_BUFFER_H */
diff --git a/drivers/oprofile/event_buffer.c b/drivers/oprofile/event_buffer.c
index 191a3202cec..c0cc4e7ff02 100644
--- a/drivers/oprofile/event_buffer.c
+++ b/drivers/oprofile/event_buffer.c
@@ -35,12 +35,23 @@ static size_t buffer_pos;
 /* atomic_t because wait_event checks it outside of buffer_mutex */
 static atomic_t buffer_ready = ATOMIC_INIT(0);
 
-/* Add an entry to the event buffer. When we
- * get near to the end we wake up the process
- * sleeping on the read() of the file.
+/*
+ * Add an entry to the event buffer. When we get near to the end we
+ * wake up the process sleeping on the read() of the file. To protect
+ * the event_buffer this function may only be called when buffer_mutex
+ * is set.
  */
 void add_event_entry(unsigned long value)
 {
+	/*
+	 * This shouldn't happen since all workqueues or handlers are
+	 * canceled or flushed before the event buffer is freed.
+	 */
+	if (!event_buffer) {
+		WARN_ON_ONCE(1);
+		return;
+	}
+
 	if (buffer_pos == buffer_size) {
 		atomic_inc(&oprofile_stats.event_lost_overflow);
 		return;
@@ -69,32 +80,32 @@ void wake_up_buffer_waiter(void)
 
 int alloc_event_buffer(void)
 {
-	int err = -ENOMEM;
 	unsigned long flags;
 
-	spin_lock_irqsave(&oprofilefs_lock, flags);
-	buffer_size = fs_buffer_size;
-	buffer_watershed = fs_buffer_watershed;
-	spin_unlock_irqrestore(&oprofilefs_lock, flags);
+	raw_spin_lock_irqsave(&oprofilefs_lock, flags);
+	buffer_size = oprofile_buffer_size;
+	buffer_watershed = oprofile_buffer_watershed;
+	raw_spin_unlock_irqrestore(&oprofilefs_lock, flags);
 
 	if (buffer_watershed >= buffer_size)
 		return -EINVAL;
 
+	buffer_pos = 0;
 	event_buffer = vmalloc(sizeof(unsigned long) * buffer_size);
 	if (!event_buffer)
-		goto out;
+		return -ENOMEM;
 
-	err = 0;
-out:
-	return err;
+	return 0;
 }
 
 
 void free_event_buffer(void)
 {
+	mutex_lock(&buffer_mutex);
 	vfree(event_buffer);
-
+	buffer_pos = 0;
 	event_buffer = NULL;
+	mutex_unlock(&buffer_mutex);
 }
 
 
@@ -124,7 +135,7 @@ static int event_buffer_open(struct inode *inode, struct file *file)
 	 * echo 1 >/dev/oprofile/enable
 	 */
 
-	return 0;
+	return nonseekable_open(inode, file);
 
 fail:
 	dcookie_unregister(file->private_data);
@@ -167,6 +178,12 @@ static ssize_t event_buffer_read(struct file *file, char __user *buf,
 
 	mutex_lock(&buffer_mutex);
 
+	/* May happen if the buffer is freed during pending reads. */
+	if (!event_buffer) {
+		retval = -EINTR;
+		goto out;
+	}
+
 	atomic_set(&buffer_ready, 0);
 
 	retval = -EFAULT;
@@ -188,4 +205,5 @@ const struct file_operations event_buffer_fops = {
 	.open		= event_buffer_open,
 	.release	= event_buffer_release,
 	.read		= event_buffer_read,
+	.llseek		= no_llseek,
 };
diff --git a/drivers/oprofile/event_buffer.h b/drivers/oprofile/event_buffer.h
index 4e70749f8d1..a8d5bb3cba8 100644
--- a/drivers/oprofile/event_buffer.h
+++ b/drivers/oprofile/event_buffer.h
@@ -11,7 +11,7 @@
 #define EVENT_BUFFER_H
 
 #include <linux/types.h>
-#include <asm/mutex.h>
+#include <linux/mutex.h>
 
 int alloc_event_buffer(void);
 
diff --git a/drivers/oprofile/nmi_timer_int.c b/drivers/oprofile/nmi_timer_int.c
new file mode 100644
index 00000000000..9559829fb23
--- /dev/null
+++ b/drivers/oprofile/nmi_timer_int.c
@@ -0,0 +1,176 @@
+/**
+ * @file nmi_timer_int.c
+ *
+ * @remark Copyright 2011 Advanced Micro Devices, Inc.
+ *
+ * @author Robert Richter <robert.richter@amd.com>
+ */
+
+#include <linux/init.h>
+#include <linux/smp.h>
+#include <linux/errno.h>
+#include <linux/oprofile.h>
+#include <linux/perf_event.h>
+
+#ifdef CONFIG_OPROFILE_NMI_TIMER
+
+static DEFINE_PER_CPU(struct perf_event *, nmi_timer_events);
+static int ctr_running;
+
+static struct perf_event_attr nmi_timer_attr = {
+	.type           = PERF_TYPE_HARDWARE,
+	.config         = PERF_COUNT_HW_CPU_CYCLES,
+	.size           = sizeof(struct perf_event_attr),
+	.pinned         = 1,
+	.disabled       = 1,
+};
+
+static void nmi_timer_callback(struct perf_event *event,
+			       struct perf_sample_data *data,
+			       struct pt_regs *regs)
+{
+	event->hw.interrupts = 0;       /* don't throttle interrupts */
+	oprofile_add_sample(regs, 0);
+}
+
+static int nmi_timer_start_cpu(int cpu)
+{
+	struct perf_event *event = per_cpu(nmi_timer_events, cpu);
+
+	if (!event) {
+		event = perf_event_create_kernel_counter(&nmi_timer_attr, cpu, NULL,
+							 nmi_timer_callback, NULL);
+		if (IS_ERR(event))
+			return PTR_ERR(event);
+		per_cpu(nmi_timer_events, cpu) = event;
+	}
+
+	if (event && ctr_running)
+		perf_event_enable(event);
+
+	return 0;
+}
+
+static void nmi_timer_stop_cpu(int cpu)
+{
+	struct perf_event *event = per_cpu(nmi_timer_events, cpu);
+
+	if (event && ctr_running)
+		perf_event_disable(event);
+}
+
+static int nmi_timer_cpu_notifier(struct notifier_block *b, unsigned long action,
+				  void *data)
+{
+	int cpu = (unsigned long)data;
+	switch (action) {
+	case CPU_DOWN_FAILED:
+	case CPU_ONLINE:
+		nmi_timer_start_cpu(cpu);
+		break;
+	case CPU_DOWN_PREPARE:
+		nmi_timer_stop_cpu(cpu);
+		break;
+	}
+	return NOTIFY_DONE;
+}
+
+static struct notifier_block nmi_timer_cpu_nb = {
+	.notifier_call = nmi_timer_cpu_notifier
+};
+
+static int nmi_timer_start(void)
+{
+	int cpu;
+
+	get_online_cpus();
+	ctr_running = 1;
+	for_each_online_cpu(cpu)
+		nmi_timer_start_cpu(cpu);
+	put_online_cpus();
+
+	return 0;
+}
+
+static void nmi_timer_stop(void)
+{
+	int cpu;
+
+	get_online_cpus();
+	for_each_online_cpu(cpu)
+		nmi_timer_stop_cpu(cpu);
+	ctr_running = 0;
+	put_online_cpus();
+}
+
+static void nmi_timer_shutdown(void)
+{
+	struct perf_event *event;
+	int cpu;
+
+	cpu_notifier_register_begin();
+	__unregister_cpu_notifier(&nmi_timer_cpu_nb);
+	for_each_possible_cpu(cpu) {
+		event = per_cpu(nmi_timer_events, cpu);
+		if (!event)
+			continue;
+		perf_event_disable(event);
+		per_cpu(nmi_timer_events, cpu) = NULL;
+		perf_event_release_kernel(event);
+	}
+
+	cpu_notifier_register_done();
+}
+
+static int nmi_timer_setup(void)
+{
+	int cpu, err;
+	u64 period;
+
+	/* clock cycles per tick: */
+	period = (u64)cpu_khz * 1000;
+	do_div(period, HZ);
+	nmi_timer_attr.sample_period = period;
+
+	cpu_notifier_register_begin();
+	err = __register_cpu_notifier(&nmi_timer_cpu_nb);
+	if (err)
+		goto out;
+
+	/* can't attach events to offline cpus: */
+	for_each_online_cpu(cpu) {
+		err = nmi_timer_start_cpu(cpu);
+		if (err) {
+			cpu_notifier_register_done();
+			nmi_timer_shutdown();
+			return err;
+		}
+	}
+
+out:
+	cpu_notifier_register_done();
+	return err;
+}
+
+int __init op_nmi_timer_init(struct oprofile_operations *ops)
+{
+	int err = 0;
+
+	err = nmi_timer_setup();
+	if (err)
+		return err;
+	nmi_timer_shutdown();		/* only check, don't alloc */
+
+	ops->create_files	= NULL;
+	ops->setup		= nmi_timer_setup;
+	ops->shutdown		= nmi_timer_shutdown;
+	ops->start		= nmi_timer_start;
+	ops->stop		= nmi_timer_stop;
+	ops->cpu_type		= "timer";
+
+	printk(KERN_INFO "oprofile: using NMI timer interrupt.\n");
+
+	return 0;
+}
+
+#endif
diff --git a/drivers/oprofile/oprof.c b/drivers/oprofile/oprof.c
index cd375907f26..ed2c3ec0702 100644
--- a/drivers/oprofile/oprof.c
+++ b/drivers/oprofile/oprof.c
@@ -12,7 +12,9 @@
 #include <linux/init.h>
 #include <linux/oprofile.h>
 #include <linux/moduleparam.h>
-#include <asm/mutex.h>
+#include <linux/workqueue.h>
+#include <linux/time.h>
+#include <linux/mutex.h>
 
 #include "oprof.h"
 #include "event_buffer.h"
@@ -23,7 +25,7 @@
 struct oprofile_operations oprofile_ops;
 
 unsigned long oprofile_started;
-unsigned long backtrace_depth;
+unsigned long oprofile_backtrace_depth;
 static unsigned long is_setup;
 static DEFINE_MUTEX(start_mutex);
 
@@ -87,6 +89,69 @@ out:
 	return err;
 }
 
+#ifdef CONFIG_OPROFILE_EVENT_MULTIPLEX
+
+static void switch_worker(struct work_struct *work);
+static DECLARE_DELAYED_WORK(switch_work, switch_worker);
+
+static void start_switch_worker(void)
+{
+	if (oprofile_ops.switch_events)
+		schedule_delayed_work(&switch_work, oprofile_time_slice);
+}
+
+static void stop_switch_worker(void)
+{
+	cancel_delayed_work_sync(&switch_work);
+}
+
+static void switch_worker(struct work_struct *work)
+{
+	if (oprofile_ops.switch_events())
+		return;
+
+	atomic_inc(&oprofile_stats.multiplex_counter);
+	start_switch_worker();
+}
+
+/* User inputs in ms, converts to jiffies */
+int oprofile_set_timeout(unsigned long val_msec)
+{
+	int err = 0;
+	unsigned long time_slice;
+
+	mutex_lock(&start_mutex);
+
+	if (oprofile_started) {
+		err = -EBUSY;
+		goto out;
+	}
+
+	if (!oprofile_ops.switch_events) {
+		err = -EINVAL;
+		goto out;
+	}
+
+	time_slice = msecs_to_jiffies(val_msec);
+	if (time_slice == MAX_JIFFY_OFFSET) {
+		err = -EINVAL;
+		goto out;
+	}
+
+	oprofile_time_slice = time_slice;
+
+out:
+	mutex_unlock(&start_mutex);
+	return err;
+
+}
+
+#else
+
+static inline void start_switch_worker(void) { }
+static inline void stop_switch_worker(void) { }
+
+#endif
 
 /* Actually start profiling (echo 1>/dev/oprofile/enable) */
 int oprofile_start(void)
@@ -108,6 +173,8 @@ int oprofile_start(void)
 	if ((err = oprofile_ops.start()))
 		goto out;
 
+	start_switch_worker();
+
 	oprofile_started = 1;
 out:
 	mutex_unlock(&start_mutex);
@@ -123,6 +190,9 @@ void oprofile_stop(void)
 		goto out;
 	oprofile_ops.stop();
 	oprofile_started = 0;
+
+	stop_switch_worker();
+
 	/* wake up the daemon to read what remains */
 	wake_up_buffer_waiter();
 out:
@@ -155,53 +225,53 @@ post_sync:
 	mutex_unlock(&start_mutex);
 }
 
-
-int oprofile_set_backtrace(unsigned long val)
+int oprofile_set_ulong(unsigned long *addr, unsigned long val)
 {
-	int err = 0;
+	int err = -EBUSY;
 
 	mutex_lock(&start_mutex);
-
-	if (oprofile_started) {
-		err = -EBUSY;
-		goto out;
+	if (!oprofile_started) {
+		*addr = val;
+		err = 0;
 	}
-
-	if (!oprofile_ops.backtrace) {
-		err = -EINVAL;
-		goto out;
-	}
-
-	backtrace_depth = val;
-
-out:
 	mutex_unlock(&start_mutex);
+
 	return err;
 }
 
+static int timer_mode;
+
 static int __init oprofile_init(void)
 {
 	int err;
 
+	/* always init architecture to setup backtrace support */
+	timer_mode = 0;
 	err = oprofile_arch_init(&oprofile_ops);
-
-	if (err < 0 || timer) {
-		printk(KERN_INFO "oprofile: using timer interrupt.\n");
-		oprofile_timer_init(&oprofile_ops);
+	if (!err) {
+		if (!timer && !oprofilefs_register())
+			return 0;
+		oprofile_arch_exit();
 	}
 
-	err = oprofilefs_register();
-	if (err)
-		oprofile_arch_exit();
+	/* setup timer mode: */
+	timer_mode = 1;
+	/* no nmi timer mode if oprofile.timer is set */
+	if (timer || op_nmi_timer_init(&oprofile_ops)) {
+		err = oprofile_timer_init(&oprofile_ops);
+		if (err)
+			return err;
+	}
 
-	return err;
+	return oprofilefs_register();
 }
 
 
 static void __exit oprofile_exit(void)
 {
 	oprofilefs_unregister();
-	oprofile_arch_exit();
+	if (!timer_mode)
+		oprofile_arch_exit();
 }
 
 
diff --git a/drivers/oprofile/oprof.h b/drivers/oprofile/oprof.h
index 5df0c21a608..d5412060ab0 100644
--- a/drivers/oprofile/oprof.h
+++ b/drivers/oprofile/oprof.h
@@ -21,19 +21,30 @@ void oprofile_stop(void);
 
 struct oprofile_operations;
 
-extern unsigned long fs_buffer_size;
-extern unsigned long fs_cpu_buffer_size;
-extern unsigned long fs_buffer_watershed;
+extern unsigned long oprofile_buffer_size;
+extern unsigned long oprofile_cpu_buffer_size;
+extern unsigned long oprofile_buffer_watershed;
+extern unsigned long oprofile_time_slice;
+
 extern struct oprofile_operations oprofile_ops;
 extern unsigned long oprofile_started;
-extern unsigned long backtrace_depth;
+extern unsigned long oprofile_backtrace_depth;
 
-struct super_block;
 struct dentry;
 
-void oprofile_create_files(struct super_block *sb, struct dentry *root);
-void oprofile_timer_init(struct oprofile_operations *ops);
+void oprofile_create_files(struct dentry *root);
+int oprofile_timer_init(struct oprofile_operations *ops);
+#ifdef CONFIG_OPROFILE_NMI_TIMER
+int op_nmi_timer_init(struct oprofile_operations *ops);
+#else
+static inline int op_nmi_timer_init(struct oprofile_operations *ops)
+{
+	return -ENODEV;
+}
+#endif
+
 
-int oprofile_set_backtrace(unsigned long depth);
+int oprofile_set_ulong(unsigned long *addr, unsigned long val);
+int oprofile_set_timeout(unsigned long time);
 
 #endif /* OPROF_H */
diff --git a/drivers/oprofile/oprofile_files.c b/drivers/oprofile/oprofile_files.c
index d8201998b0b..ee2cfce358b 100644
--- a/drivers/oprofile/oprofile_files.c
+++ b/drivers/oprofile/oprofile_files.c
@@ -9,26 +9,34 @@
 
 #include <linux/fs.h>
 #include <linux/oprofile.h>
+#include <linux/jiffies.h>
 
 #include "event_buffer.h"
 #include "oprofile_stats.h"
 #include "oprof.h"
 
-#define FS_BUFFER_SIZE_DEFAULT		131072
-#define FS_CPU_BUFFER_SIZE_DEFAULT	8192
-#define FS_BUFFER_WATERSHED_DEFAULT	32768	/* FIXME: tune */
+#define BUFFER_SIZE_DEFAULT		131072
+#define CPU_BUFFER_SIZE_DEFAULT		8192
+#define BUFFER_WATERSHED_DEFAULT	32768	/* FIXME: tune */
+#define TIME_SLICE_DEFAULT		1
 
-unsigned long fs_buffer_size;
-unsigned long fs_cpu_buffer_size;
-unsigned long fs_buffer_watershed;
+unsigned long oprofile_buffer_size;
+unsigned long oprofile_cpu_buffer_size;
+unsigned long oprofile_buffer_watershed;
+unsigned long oprofile_time_slice;
 
-static ssize_t depth_read(struct file *file, char __user *buf, size_t count, loff_t *offset)
+#ifdef CONFIG_OPROFILE_EVENT_MULTIPLEX
+
+static ssize_t timeout_read(struct file *file, char __user *buf,
+		size_t count, loff_t *offset)
 {
-	return oprofilefs_ulong_to_user(backtrace_depth, buf, count, offset);
+	return oprofilefs_ulong_to_user(jiffies_to_msecs(oprofile_time_slice),
+					buf, count, offset);
 }
 
 
-static ssize_t depth_write(struct file *file, char const __user *buf, size_t count, loff_t *offset)
+static ssize_t timeout_write(struct file *file, char const __user *buf,
+		size_t count, loff_t *offset)
 {
 	unsigned long val;
 	int retval;
@@ -37,20 +45,60 @@ static ssize_t depth_write(struct file *file, char const __user *buf, size_t cou
 		return -EINVAL;
 
 	retval = oprofilefs_ulong_from_user(&val, buf, count);
+	if (retval <= 0)
+		return retval;
+
+	retval = oprofile_set_timeout(val);
+
 	if (retval)
 		return retval;
+	return count;
+}
+
+
+static const struct file_operations timeout_fops = {
+	.read		= timeout_read,
+	.write		= timeout_write,
+	.llseek		= default_llseek,
+};
+
+#endif
+
+
+static ssize_t depth_read(struct file *file, char __user *buf, size_t count, loff_t *offset)
+{
+	return oprofilefs_ulong_to_user(oprofile_backtrace_depth, buf, count,
+					offset);
+}
+
+
+static ssize_t depth_write(struct file *file, char const __user *buf, size_t count, loff_t *offset)
+{
+	unsigned long val;
+	int retval;
+
+	if (*offset)
+		return -EINVAL;
+
+	if (!oprofile_ops.backtrace)
+		return -EINVAL;
 
-	retval = oprofile_set_backtrace(val);
+	retval = oprofilefs_ulong_from_user(&val, buf, count);
+	if (retval <= 0)
+		return retval;
 
+	retval = oprofile_set_ulong(&oprofile_backtrace_depth, val);
 	if (retval)
 		return retval;
+
 	return count;
 }
 
 
 static const struct file_operations depth_fops = {
 	.read		= depth_read,
-	.write		= depth_write
+	.write		= depth_write,
+	.llseek		= default_llseek,
 };
 
 
@@ -62,6 +110,7 @@ static ssize_t pointer_size_read(struct file *file, char __user *buf, size_t cou
 
 static const struct file_operations pointer_size_fops = {
 	.read		= pointer_size_read,
+	.llseek		= default_llseek,
 };
 
 
@@ -73,6 +122,7 @@ static ssize_t cpu_type_read(struct file *file, char __user *buf, size_t count,
 
 static const struct file_operations cpu_type_fops = {
 	.read		= cpu_type_read,
+	.llseek		= default_llseek,
 };
 
 
@@ -91,9 +141,10 @@ static ssize_t enable_write(struct file *file, char const __user *buf, size_t co
 		return -EINVAL;
 
 	retval = oprofilefs_ulong_from_user(&val, buf, count);
-	if (retval)
+	if (retval <= 0)
 		return retval;
 
+	retval = 0;
 	if (val)
 		retval = oprofile_start();
 	else
@@ -108,6 +159,7 @@ static ssize_t enable_write(struct file *file, char const __user *buf, size_t co
 static const struct file_operations enable_fops = {
 	.read		= enable_read,
 	.write		= enable_write,
+	.llseek		= default_llseek,
 };
 
 
@@ -120,25 +172,30 @@ static ssize_t dump_write(struct file *file, char const __user *buf, size_t coun
 
 static const struct file_operations dump_fops = {
 	.write		= dump_write,
+	.llseek		= noop_llseek,
 };
 
-void oprofile_create_files(struct super_block *sb, struct dentry *root)
+void oprofile_create_files(struct dentry *root)
 {
 	/* reinitialize default values */
-	fs_buffer_size =	FS_BUFFER_SIZE_DEFAULT;
-	fs_cpu_buffer_size =	FS_CPU_BUFFER_SIZE_DEFAULT;
-	fs_buffer_watershed =	FS_BUFFER_WATERSHED_DEFAULT;
-
-	oprofilefs_create_file(sb, root, "enable", &enable_fops);
-	oprofilefs_create_file_perm(sb, root, "dump", &dump_fops, 0666);
-	oprofilefs_create_file(sb, root, "buffer", &event_buffer_fops);
-	oprofilefs_create_ulong(sb, root, "buffer_size", &fs_buffer_size);
-	oprofilefs_create_ulong(sb, root, "buffer_watershed", &fs_buffer_watershed);
-	oprofilefs_create_ulong(sb, root, "cpu_buffer_size", &fs_cpu_buffer_size);
-	oprofilefs_create_file(sb, root, "cpu_type", &cpu_type_fops);
-	oprofilefs_create_file(sb, root, "backtrace_depth", &depth_fops);
-	oprofilefs_create_file(sb, root, "pointer_size", &pointer_size_fops);
-	oprofile_create_stats_files(sb, root);
+	oprofile_buffer_size =		BUFFER_SIZE_DEFAULT;
+	oprofile_cpu_buffer_size =	CPU_BUFFER_SIZE_DEFAULT;
+	oprofile_buffer_watershed =	BUFFER_WATERSHED_DEFAULT;
+	oprofile_time_slice =		msecs_to_jiffies(TIME_SLICE_DEFAULT);
+
+	oprofilefs_create_file(root, "enable", &enable_fops);
+	oprofilefs_create_file_perm(root, "dump", &dump_fops, 0666);
+	oprofilefs_create_file(root, "buffer", &event_buffer_fops);
+	oprofilefs_create_ulong(root, "buffer_size", &oprofile_buffer_size);
+	oprofilefs_create_ulong(root, "buffer_watershed", &oprofile_buffer_watershed);
+	oprofilefs_create_ulong(root, "cpu_buffer_size", &oprofile_cpu_buffer_size);
+	oprofilefs_create_file(root, "cpu_type", &cpu_type_fops);
+	oprofilefs_create_file(root, "backtrace_depth", &depth_fops);
+	oprofilefs_create_file(root, "pointer_size", &pointer_size_fops);
+#ifdef CONFIG_OPROFILE_EVENT_MULTIPLEX
+	oprofilefs_create_file(root, "time_slice", &timeout_fops);
+#endif
+	oprofile_create_stats_files(root);
 	if (oprofile_ops.create_files)
-		oprofile_ops.create_files(sb, root);
+		oprofile_ops.create_files(root);
 }
diff --git a/drivers/oprofile/oprofile_perf.c b/drivers/oprofile/oprofile_perf.c
new file mode 100644
index 00000000000..d5b2732b1b8
--- /dev/null
+++ b/drivers/oprofile/oprofile_perf.c
@@ -0,0 +1,327 @@
+/*
+ * Copyright 2010 ARM Ltd.
+ * Copyright 2012 Advanced Micro Devices, Inc., Robert Richter
+ *
+ * Perf-events backend for OProfile.
+ */
+#include <linux/perf_event.h>
+#include <linux/platform_device.h>
+#include <linux/oprofile.h>
+#include <linux/slab.h>
+
+/*
+ * Per performance monitor configuration as set via oprofilefs.
+ */
+struct op_counter_config {
+	unsigned long count;
+	unsigned long enabled;
+	unsigned long event;
+	unsigned long unit_mask;
+	unsigned long kernel;
+	unsigned long user;
+	struct perf_event_attr attr;
+};
+
+static int oprofile_perf_enabled;
+static DEFINE_MUTEX(oprofile_perf_mutex);
+
+static struct op_counter_config *counter_config;
+static DEFINE_PER_CPU(struct perf_event **, perf_events);
+static int num_counters;
+
+/*
+ * Overflow callback for oprofile.
+ */
+static void op_overflow_handler(struct perf_event *event,
+			struct perf_sample_data *data, struct pt_regs *regs)
+{
+	int id;
+	u32 cpu = smp_processor_id();
+
+	for (id = 0; id < num_counters; ++id)
+		if (per_cpu(perf_events, cpu)[id] == event)
+			break;
+
+	if (id != num_counters)
+		oprofile_add_sample(regs, id);
+	else
+		pr_warning("oprofile: ignoring spurious overflow "
+				"on cpu %u\n", cpu);
+}
+
+/*
+ * Called by oprofile_perf_setup to create perf attributes to mirror the oprofile
+ * settings in counter_config. Attributes are created as `pinned' events and
+ * so are permanently scheduled on the PMU.
+ */
+static void op_perf_setup(void)
+{
+	int i;
+	u32 size = sizeof(struct perf_event_attr);
+	struct perf_event_attr *attr;
+
+	for (i = 0; i < num_counters; ++i) {
+		attr = &counter_config[i].attr;
+		memset(attr, 0, size);
+		attr->type		= PERF_TYPE_RAW;
+		attr->size		= size;
+		attr->config		= counter_config[i].event;
+		attr->sample_period	= counter_config[i].count;
+		attr->pinned		= 1;
+	}
+}
+
+static int op_create_counter(int cpu, int event)
+{
+	struct perf_event *pevent;
+
+	if (!counter_config[event].enabled || per_cpu(perf_events, cpu)[event])
+		return 0;
+
+	pevent = perf_event_create_kernel_counter(&counter_config[event].attr,
+						  cpu, NULL,
+						  op_overflow_handler, NULL);
+
+	if (IS_ERR(pevent))
+		return PTR_ERR(pevent);
+
+	if (pevent->state != PERF_EVENT_STATE_ACTIVE) {
+		perf_event_release_kernel(pevent);
+		pr_warning("oprofile: failed to enable event %d "
+				"on CPU %d\n", event, cpu);
+		return -EBUSY;
+	}
+
+	per_cpu(perf_events, cpu)[event] = pevent;
+
+	return 0;
+}
+
+static void op_destroy_counter(int cpu, int event)
+{
+	struct perf_event *pevent = per_cpu(perf_events, cpu)[event];
+
+	if (pevent) {
+		perf_event_release_kernel(pevent);
+		per_cpu(perf_events, cpu)[event] = NULL;
+	}
+}
+
+/*
+ * Called by oprofile_perf_start to create active perf events based on the
+ * perviously configured attributes.
+ */
+static int op_perf_start(void)
+{
+	int cpu, event, ret = 0;
+
+	for_each_online_cpu(cpu) {
+		for (event = 0; event < num_counters; ++event) {
+			ret = op_create_counter(cpu, event);
+			if (ret)
+				return ret;
+		}
+	}
+
+	return ret;
+}
+
+/*
+ * Called by oprofile_perf_stop at the end of a profiling run.
+ */
+static void op_perf_stop(void)
+{
+	int cpu, event;
+
+	for_each_online_cpu(cpu)
+		for (event = 0; event < num_counters; ++event)
+			op_destroy_counter(cpu, event);
+}
+
+static int oprofile_perf_create_files(struct dentry *root)
+{
+	unsigned int i;
+
+	for (i = 0; i < num_counters; i++) {
+		struct dentry *dir;
+		char buf[4];
+
+		snprintf(buf, sizeof buf, "%d", i);
+		dir = oprofilefs_mkdir(root, buf);
+		oprofilefs_create_ulong(dir, "enabled", &counter_config[i].enabled);
+		oprofilefs_create_ulong(dir, "event", &counter_config[i].event);
+		oprofilefs_create_ulong(dir, "count", &counter_config[i].count);
+		oprofilefs_create_ulong(dir, "unit_mask", &counter_config[i].unit_mask);
+		oprofilefs_create_ulong(dir, "kernel", &counter_config[i].kernel);
+		oprofilefs_create_ulong(dir, "user", &counter_config[i].user);
+	}
+
+	return 0;
+}
+
+static int oprofile_perf_setup(void)
+{
+	raw_spin_lock(&oprofilefs_lock);
+	op_perf_setup();
+	raw_spin_unlock(&oprofilefs_lock);
+	return 0;
+}
+
+static int oprofile_perf_start(void)
+{
+	int ret = -EBUSY;
+
+	mutex_lock(&oprofile_perf_mutex);
+	if (!oprofile_perf_enabled) {
+		ret = 0;
+		op_perf_start();
+		oprofile_perf_enabled = 1;
+	}
+	mutex_unlock(&oprofile_perf_mutex);
+	return ret;
+}
+
+static void oprofile_perf_stop(void)
+{
+	mutex_lock(&oprofile_perf_mutex);
+	if (oprofile_perf_enabled)
+		op_perf_stop();
+	oprofile_perf_enabled = 0;
+	mutex_unlock(&oprofile_perf_mutex);
+}
+
+#ifdef CONFIG_PM
+
+static int oprofile_perf_suspend(struct platform_device *dev, pm_message_t state)
+{
+	mutex_lock(&oprofile_perf_mutex);
+	if (oprofile_perf_enabled)
+		op_perf_stop();
+	mutex_unlock(&oprofile_perf_mutex);
+	return 0;
+}
+
+static int oprofile_perf_resume(struct platform_device *dev)
+{
+	mutex_lock(&oprofile_perf_mutex);
+	if (oprofile_perf_enabled && op_perf_start())
+		oprofile_perf_enabled = 0;
+	mutex_unlock(&oprofile_perf_mutex);
+	return 0;
+}
+
+static struct platform_driver oprofile_driver = {
+	.driver		= {
+		.name		= "oprofile-perf",
+	},
+	.resume		= oprofile_perf_resume,
+	.suspend	= oprofile_perf_suspend,
+};
+
+static struct platform_device *oprofile_pdev;
+
+static int __init init_driverfs(void)
+{
+	int ret;
+
+	ret = platform_driver_register(&oprofile_driver);
+	if (ret)
+		return ret;
+
+	oprofile_pdev =	platform_device_register_simple(
+				oprofile_driver.driver.name, 0, NULL, 0);
+	if (IS_ERR(oprofile_pdev)) {
+		ret = PTR_ERR(oprofile_pdev);
+		platform_driver_unregister(&oprofile_driver);
+	}
+
+	return ret;
+}
+
+static void exit_driverfs(void)
+{
+	platform_device_unregister(oprofile_pdev);
+	platform_driver_unregister(&oprofile_driver);
+}
+
+#else
+
+static inline int  init_driverfs(void) { return 0; }
+static inline void exit_driverfs(void) { }
+
+#endif /* CONFIG_PM */
+
+void oprofile_perf_exit(void)
+{
+	int cpu, id;
+	struct perf_event *event;
+
+	for_each_possible_cpu(cpu) {
+		for (id = 0; id < num_counters; ++id) {
+			event = per_cpu(perf_events, cpu)[id];
+			if (event)
+				perf_event_release_kernel(event);
+		}
+
+		kfree(per_cpu(perf_events, cpu));
+	}
+
+	kfree(counter_config);
+	exit_driverfs();
+}
+
+int __init oprofile_perf_init(struct oprofile_operations *ops)
+{
+	int cpu, ret = 0;
+
+	ret = init_driverfs();
+	if (ret)
+		return ret;
+
+	num_counters = perf_num_counters();
+	if (num_counters <= 0) {
+		pr_info("oprofile: no performance counters\n");
+		ret = -ENODEV;
+		goto out;
+	}
+
+	counter_config = kcalloc(num_counters,
+			sizeof(struct op_counter_config), GFP_KERNEL);
+
+	if (!counter_config) {
+		pr_info("oprofile: failed to allocate %d "
+				"counters\n", num_counters);
+		ret = -ENOMEM;
+		num_counters = 0;
+		goto out;
+	}
+
+	for_each_possible_cpu(cpu) {
+		per_cpu(perf_events, cpu) = kcalloc(num_counters,
+				sizeof(struct perf_event *), GFP_KERNEL);
+		if (!per_cpu(perf_events, cpu)) {
+			pr_info("oprofile: failed to allocate %d perf events "
+					"for cpu %d\n", num_counters, cpu);
+			ret = -ENOMEM;
+			goto out;
+		}
+	}
+
+	ops->create_files	= oprofile_perf_create_files;
+	ops->setup		= oprofile_perf_setup;
+	ops->start		= oprofile_perf_start;
+	ops->stop		= oprofile_perf_stop;
+	ops->shutdown		= oprofile_perf_stop;
+	ops->cpu_type		= op_name_from_perf_id();
+
+	if (!ops->cpu_type)
+		ret = -ENODEV;
+	else
+		pr_info("oprofile: using %s\n", ops->cpu_type);
+
+out:
+	if (ret)
+		oprofile_perf_exit();
+
+	return ret;
+}
diff --git a/drivers/oprofile/oprofile_stats.c b/drivers/oprofile/oprofile_stats.c
index e1f6ce03705..59659cea458 100644
--- a/drivers/oprofile/oprofile_stats.c
+++ b/drivers/oprofile/oprofile_stats.c
@@ -23,7 +23,7 @@ void oprofile_reset_stats(void)
 	int i;
 
 	for_each_possible_cpu(i) {
-		cpu_buf = &per_cpu(cpu_buffer, i);
+		cpu_buf = &per_cpu(op_cpu_buffer, i);
 		cpu_buf->sample_received = 0;
 		cpu_buf->sample_lost_overflow = 0;
 		cpu_buf->backtrace_aborted = 0;
@@ -33,10 +33,12 @@ void oprofile_reset_stats(void)
 	atomic_set(&oprofile_stats.sample_lost_no_mm, 0);
 	atomic_set(&oprofile_stats.sample_lost_no_mapping, 0);
 	atomic_set(&oprofile_stats.event_lost_overflow, 0);
+	atomic_set(&oprofile_stats.bt_lost_no_mapping, 0);
+	atomic_set(&oprofile_stats.multiplex_counter, 0);
 }
 
 
-void oprofile_create_stats_files(struct super_block *sb, struct dentry *root)
+void oprofile_create_stats_files(struct dentry *root)
 {
 	struct oprofile_cpu_buffer *cpu_buf;
 	struct dentry *cpudir;
@@ -44,35 +46,39 @@ void oprofile_create_stats_files(struct super_block *sb, struct dentry *root)
 	char buf[10];
 	int i;
 
-	dir = oprofilefs_mkdir(sb, root, "stats");
+	dir = oprofilefs_mkdir(root, "stats");
 	if (!dir)
 		return;
 
 	for_each_possible_cpu(i) {
-		cpu_buf = &per_cpu(cpu_buffer, i);
+		cpu_buf = &per_cpu(op_cpu_buffer, i);
 		snprintf(buf, 10, "cpu%d", i);
-		cpudir = oprofilefs_mkdir(sb, dir, buf);
+		cpudir = oprofilefs_mkdir(dir, buf);
 
 		/* Strictly speaking access to these ulongs is racy,
 		 * but we can't simply lock them, and they are
 		 * informational only.
 		 */
-		oprofilefs_create_ro_ulong(sb, cpudir, "sample_received",
+		oprofilefs_create_ro_ulong(cpudir, "sample_received",
 			&cpu_buf->sample_received);
-		oprofilefs_create_ro_ulong(sb, cpudir, "sample_lost_overflow",
+		oprofilefs_create_ro_ulong(cpudir, "sample_lost_overflow",
 			&cpu_buf->sample_lost_overflow);
-		oprofilefs_create_ro_ulong(sb, cpudir, "backtrace_aborted",
+		oprofilefs_create_ro_ulong(cpudir, "backtrace_aborted",
 			&cpu_buf->backtrace_aborted);
-		oprofilefs_create_ro_ulong(sb, cpudir, "sample_invalid_eip",
+		oprofilefs_create_ro_ulong(cpudir, "sample_invalid_eip",
 			&cpu_buf->sample_invalid_eip);
 	}
 
-	oprofilefs_create_ro_atomic(sb, dir, "sample_lost_no_mm",
+	oprofilefs_create_ro_atomic(dir, "sample_lost_no_mm",
 		&oprofile_stats.sample_lost_no_mm);
-	oprofilefs_create_ro_atomic(sb, dir, "sample_lost_no_mapping",
+	oprofilefs_create_ro_atomic(dir, "sample_lost_no_mapping",
 		&oprofile_stats.sample_lost_no_mapping);
-	oprofilefs_create_ro_atomic(sb, dir, "event_lost_overflow",
+	oprofilefs_create_ro_atomic(dir, "event_lost_overflow",
 		&oprofile_stats.event_lost_overflow);
-	oprofilefs_create_ro_atomic(sb, dir, "bt_lost_no_mapping",
+	oprofilefs_create_ro_atomic(dir, "bt_lost_no_mapping",
 		&oprofile_stats.bt_lost_no_mapping);
+#ifdef CONFIG_OPROFILE_EVENT_MULTIPLEX
+	oprofilefs_create_ro_atomic(dir, "multiplex_counter",
+		&oprofile_stats.multiplex_counter);
+#endif
 }
diff --git a/drivers/oprofile/oprofile_stats.h b/drivers/oprofile/oprofile_stats.h
index 3da0d08dc1f..1fc622bd183 100644
--- a/drivers/oprofile/oprofile_stats.h
+++ b/drivers/oprofile/oprofile_stats.h
@@ -10,13 +10,14 @@
 #ifndef OPROFILE_STATS_H
 #define OPROFILE_STATS_H
 
-#include <asm/atomic.h>
+#include <linux/atomic.h>
 
 struct oprofile_stat_struct {
 	atomic_t sample_lost_no_mm;
 	atomic_t sample_lost_no_mapping;
 	atomic_t bt_lost_no_mapping;
 	atomic_t event_lost_overflow;
+	atomic_t multiplex_counter;
 };
 
 extern struct oprofile_stat_struct oprofile_stats;
@@ -24,10 +25,9 @@ extern struct oprofile_stat_struct oprofile_stats;
 /* reset all stats to zero */
 void oprofile_reset_stats(void);
 
-struct super_block;
 struct dentry;
 
 /* create the stats/ dir */
-void oprofile_create_stats_files(struct super_block *sb, struct dentry *root);
+void oprofile_create_stats_files(struct dentry *root);
 
 #endif /* OPROFILE_STATS_H */
diff --git a/drivers/oprofile/oprofilefs.c b/drivers/oprofile/oprofilefs.c
index ddc4c59f02d..3f493459378 100644
--- a/drivers/oprofile/oprofilefs.c
+++ b/drivers/oprofile/oprofilefs.c
@@ -21,24 +21,22 @@
 
 #define OPROFILEFS_MAGIC 0x6f70726f
 
-DEFINE_SPINLOCK(oprofilefs_lock);
+DEFINE_RAW_SPINLOCK(oprofilefs_lock);
 
 static struct inode *oprofilefs_get_inode(struct super_block *sb, int mode)
 {
 	struct inode *inode = new_inode(sb);
 
 	if (inode) {
+		inode->i_ino = get_next_ino();
 		inode->i_mode = mode;
-		inode->i_uid = 0;
-		inode->i_gid = 0;
-		inode->i_blocks = 0;
 		inode->i_atime = inode->i_mtime = inode->i_ctime = CURRENT_TIME;
 	}
 	return inode;
 }
 
 
-static struct super_operations s_ops = {
+static const struct super_operations s_ops = {
 	.statfs		= simple_statfs,
 	.drop_inode 	= generic_delete_inode,
 };
@@ -62,6 +60,13 @@ ssize_t oprofilefs_ulong_to_user(unsigned long val, char __user *buf, size_t cou
 }
 
 
+/*
+ * Note: If oprofilefs_ulong_from_user() returns 0, then *val remains
+ * unchanged and might be uninitialized. This follows write syscall
+ * implementation when count is zero: "If count is zero ... [and if]
+ * no errors are detected, 0 will be returned without causing any
+ * other effect." (man 2 write)
+ */
 int oprofilefs_ulong_from_user(unsigned long *val, char const __user *buf, size_t count)
 {
 	char tmpbuf[TMPBUFSIZE];
@@ -78,10 +83,10 @@ int oprofilefs_ulong_from_user(unsigned long *val, char const __user *buf, size_
 	if (copy_from_user(tmpbuf, buf, count))
 		return -EFAULT;
 
-	spin_lock_irqsave(&oprofilefs_lock, flags);
+	raw_spin_lock_irqsave(&oprofilefs_lock, flags);
 	*val = simple_strtoul(tmpbuf, NULL, 0);
-	spin_unlock_irqrestore(&oprofilefs_lock, flags);
-	return 0;
+	raw_spin_unlock_irqrestore(&oprofilefs_lock, flags);
+	return count;
 }
 
 
@@ -94,85 +99,78 @@ static ssize_t ulong_read_file(struct file *file, char __user *buf, size_t count
 
 static ssize_t ulong_write_file(struct file *file, char const __user *buf, size_t count, loff_t *offset)
 {
-	unsigned long *value = file->private_data;
+	unsigned long value;
 	int retval;
 
 	if (*offset)
 		return -EINVAL;
 
-	retval = oprofilefs_ulong_from_user(value, buf, count);
+	retval = oprofilefs_ulong_from_user(&value, buf, count);
+	if (retval <= 0)
+		return retval;
 
+	retval = oprofile_set_ulong(file->private_data, value);
 	if (retval)
 		return retval;
-	return count;
-}
-
 
-static int default_open(struct inode *inode, struct file *filp)
-{
-	if (inode->i_private)
-		filp->private_data = inode->i_private;
-	return 0;
+	return count;
 }
 
 
 static const struct file_operations ulong_fops = {
 	.read		= ulong_read_file,
 	.write		= ulong_write_file,
-	.open		= default_open,
+	.open		= simple_open,
+	.llseek		= default_llseek,
 };
 
 
 static const struct file_operations ulong_ro_fops = {
 	.read		= ulong_read_file,
-	.open		= default_open,
+	.open		= simple_open,
+	.llseek		= default_llseek,
 };
 
 
-static struct dentry *__oprofilefs_create_file(struct super_block *sb,
-	struct dentry *root, char const *name, const struct file_operations *fops,
-	int perm)
+static int __oprofilefs_create_file(struct dentry *root, char const *name,
+	const struct file_operations *fops, int perm, void *priv)
 {
 	struct dentry *dentry;
 	struct inode *inode;
 
+	mutex_lock(&root->d_inode->i_mutex);
 	dentry = d_alloc_name(root, name);
-	if (!dentry)
-		return NULL;
-	inode = oprofilefs_get_inode(sb, S_IFREG | perm);
+	if (!dentry) {
+		mutex_unlock(&root->d_inode->i_mutex);
+		return -ENOMEM;
+	}
+	inode = oprofilefs_get_inode(root->d_sb, S_IFREG | perm);
 	if (!inode) {
 		dput(dentry);
-		return NULL;
+		mutex_unlock(&root->d_inode->i_mutex);
+		return -ENOMEM;
 	}
 	inode->i_fop = fops;
+	inode->i_private = priv;
 	d_add(dentry, inode);
-	return dentry;
+	mutex_unlock(&root->d_inode->i_mutex);
+	return 0;
 }
 
 
-int oprofilefs_create_ulong(struct super_block *sb, struct dentry *root,
+int oprofilefs_create_ulong(struct dentry *root,
 	char const *name, unsigned long *val)
 {
-	struct dentry *d = __oprofilefs_create_file(sb, root, name,
-						     &ulong_fops, 0644);
-	if (!d)
-		return -EFAULT;
-
-	d->d_inode->i_private = val;
-	return 0;
+	return __oprofilefs_create_file(root, name,
+					&ulong_fops, 0644, val);
 }
 
 
-int oprofilefs_create_ro_ulong(struct super_block *sb, struct dentry *root,
+int oprofilefs_create_ro_ulong(struct dentry *root,
 	char const *name, unsigned long *val)
 {
-	struct dentry *d = __oprofilefs_create_file(sb, root, name,
-						     &ulong_ro_fops, 0444);
-	if (!d)
-		return -EFAULT;
-
-	d->d_inode->i_private = val;
-	return 0;
+	return __oprofilefs_create_file(root, name,
+					&ulong_ro_fops, 0444, val);
 }
 
 
@@ -185,58 +183,54 @@ static ssize_t atomic_read_file(struct file *file, char __user *buf, size_t coun
 
 static const struct file_operations atomic_ro_fops = {
 	.read		= atomic_read_file,
-	.open		= default_open,
+	.open		= simple_open,
+	.llseek		= default_llseek,
 };
 
 
-int oprofilefs_create_ro_atomic(struct super_block *sb, struct dentry *root,
+int oprofilefs_create_ro_atomic(struct dentry *root,
 	char const *name, atomic_t *val)
 {
-	struct dentry *d = __oprofilefs_create_file(sb, root, name,
-						     &atomic_ro_fops, 0444);
-	if (!d)
-		return -EFAULT;
-
-	d->d_inode->i_private = val;
-	return 0;
+	return __oprofilefs_create_file(root, name,
+					&atomic_ro_fops, 0444, val);
 }
 
 
-int oprofilefs_create_file(struct super_block *sb, struct dentry *root,
+int oprofilefs_create_file(struct dentry *root,
 	char const *name, const struct file_operations *fops)
 {
-	if (!__oprofilefs_create_file(sb, root, name, fops, 0644))
-		return -EFAULT;
-	return 0;
+	return __oprofilefs_create_file(root, name, fops, 0644, NULL);
 }
 
 
-int oprofilefs_create_file_perm(struct super_block *sb, struct dentry *root,
+int oprofilefs_create_file_perm(struct dentry *root,
 	char const *name, const struct file_operations *fops, int perm)
 {
-	if (!__oprofilefs_create_file(sb, root, name, fops, perm))
-		return -EFAULT;
-	return 0;
+	return __oprofilefs_create_file(root, name, fops, perm, NULL);
 }
 
 
-struct dentry *oprofilefs_mkdir(struct super_block *sb,
-	struct dentry *root, char const *name)
+struct dentry *oprofilefs_mkdir(struct dentry *parent, char const *name)
 {
 	struct dentry *dentry;
 	struct inode *inode;
 
-	dentry = d_alloc_name(root, name);
-	if (!dentry)
+	mutex_lock(&parent->d_inode->i_mutex);
+	dentry = d_alloc_name(parent, name);
+	if (!dentry) {
+		mutex_unlock(&parent->d_inode->i_mutex);
 		return NULL;
-	inode = oprofilefs_get_inode(sb, S_IFDIR | 0755);
+	}
+	inode = oprofilefs_get_inode(parent->d_sb, S_IFDIR | 0755);
 	if (!inode) {
 		dput(dentry);
+		mutex_unlock(&parent->d_inode->i_mutex);
 		return NULL;
 	}
 	inode->i_op = &simple_dir_inode_operations;
 	inode->i_fop = &simple_dir_operations;
 	d_add(dentry, inode);
+	mutex_unlock(&parent->d_inode->i_mutex);
 	return dentry;
 }
 
@@ -244,7 +238,6 @@ struct dentry *oprofilefs_mkdir(struct super_block *sb,
 static int oprofilefs_fill_super(struct super_block *sb, void *data, int silent)
 {
 	struct inode *root_inode;
-	struct dentry *root_dentry;
 
 	sb->s_blocksize = PAGE_CACHE_SIZE;
 	sb->s_blocksize_bits = PAGE_CACHE_SHIFT;
@@ -257,34 +250,31 @@ static int oprofilefs_fill_super(struct super_block *sb, void *data, int silent)
 		return -ENOMEM;
 	root_inode->i_op = &simple_dir_inode_operations;
 	root_inode->i_fop = &simple_dir_operations;
-	root_dentry = d_alloc_root(root_inode);
-	if (!root_dentry) {
-		iput(root_inode);
+	sb->s_root = d_make_root(root_inode);
+	if (!sb->s_root)
 		return -ENOMEM;
-	}
-
-	sb->s_root = root_dentry;
 
-	oprofile_create_files(sb, root_dentry);
+	oprofile_create_files(sb->s_root);
 
 	// FIXME: verify kill_litter_super removes our dentries
 	return 0;
 }
 
 
-static int oprofilefs_get_sb(struct file_system_type *fs_type,
-	int flags, const char *dev_name, void *data, struct vfsmount *mnt)
+static struct dentry *oprofilefs_mount(struct file_system_type *fs_type,
+	int flags, const char *dev_name, void *data)
 {
-	return get_sb_single(fs_type, flags, data, oprofilefs_fill_super, mnt);
+	return mount_single(fs_type, flags, data, oprofilefs_fill_super);
 }
 
 
 static struct file_system_type oprofilefs_type = {
 	.owner		= THIS_MODULE,
 	.name		= "oprofilefs",
-	.get_sb		= oprofilefs_get_sb,
+	.mount		= oprofilefs_mount,
 	.kill_sb	= kill_litter_super,
 };
+MODULE_ALIAS_FS("oprofilefs");
 
 
 int __init oprofilefs_register(void)
diff --git a/drivers/oprofile/timer_int.c b/drivers/oprofile/timer_int.c
index 333f915568c..61be1d9c16c 100644
--- a/drivers/oprofile/timer_int.c
+++ b/drivers/oprofile/timer_int.c
@@ -13,34 +13,108 @@
 #include <linux/oprofile.h>
 #include <linux/profile.h>
 #include <linux/init.h>
+#include <linux/cpu.h>
+#include <linux/hrtimer.h>
+#include <asm/irq_regs.h>
 #include <asm/ptrace.h>
 
 #include "oprof.h"
 
-static int timer_notify(struct pt_regs *regs)
+static DEFINE_PER_CPU(struct hrtimer, oprofile_hrtimer);
+static int ctr_running;
+
+static enum hrtimer_restart oprofile_hrtimer_notify(struct hrtimer *hrtimer)
+{
+	oprofile_add_sample(get_irq_regs(), 0);
+	hrtimer_forward_now(hrtimer, ns_to_ktime(TICK_NSEC));
+	return HRTIMER_RESTART;
+}
+
+static void __oprofile_hrtimer_start(void *unused)
+{
+	struct hrtimer *hrtimer = &__get_cpu_var(oprofile_hrtimer);
+
+	if (!ctr_running)
+		return;
+
+	hrtimer_init(hrtimer, CLOCK_MONOTONIC, HRTIMER_MODE_REL);
+	hrtimer->function = oprofile_hrtimer_notify;
+
+	hrtimer_start(hrtimer, ns_to_ktime(TICK_NSEC),
+		      HRTIMER_MODE_REL_PINNED);
+}
+
+static int oprofile_hrtimer_start(void)
 {
-	oprofile_add_sample(regs, 0);
+	get_online_cpus();
+	ctr_running = 1;
+	on_each_cpu(__oprofile_hrtimer_start, NULL, 1);
+	put_online_cpus();
 	return 0;
 }
 
-static int timer_start(void)
+static void __oprofile_hrtimer_stop(int cpu)
 {
-	return register_timer_hook(timer_notify);
+	struct hrtimer *hrtimer = &per_cpu(oprofile_hrtimer, cpu);
+
+	if (!ctr_running)
+		return;
+
+	hrtimer_cancel(hrtimer);
 }
 
+static void oprofile_hrtimer_stop(void)
+{
+	int cpu;
+
+	get_online_cpus();
+	for_each_online_cpu(cpu)
+		__oprofile_hrtimer_stop(cpu);
+	ctr_running = 0;
+	put_online_cpus();
+}
 
-static void timer_stop(void)
+static int oprofile_cpu_notify(struct notifier_block *self,
+			       unsigned long action, void *hcpu)
 {
-	unregister_timer_hook(timer_notify);
+	long cpu = (long) hcpu;
+
+	switch (action) {
+	case CPU_ONLINE:
+	case CPU_ONLINE_FROZEN:
+		smp_call_function_single(cpu, __oprofile_hrtimer_start,
+					 NULL, 1);
+		break;
+	case CPU_DEAD:
+	case CPU_DEAD_FROZEN:
+		__oprofile_hrtimer_stop(cpu);
+		break;
+	}
+	return NOTIFY_OK;
 }
 
+static struct notifier_block __refdata oprofile_cpu_notifier = {
+	.notifier_call = oprofile_cpu_notify,
+};
 
-void __init oprofile_timer_init(struct oprofile_operations *ops)
+static int oprofile_hrtimer_setup(void)
 {
-	ops->create_files = NULL;
-	ops->setup = NULL;
-	ops->shutdown = NULL;
-	ops->start = timer_start;
-	ops->stop = timer_stop;
-	ops->cpu_type = "timer";
+	return register_hotcpu_notifier(&oprofile_cpu_notifier);
+}
+
+static void oprofile_hrtimer_shutdown(void)
+{
+	unregister_hotcpu_notifier(&oprofile_cpu_notifier);
+}
+
+int oprofile_timer_init(struct oprofile_operations *ops)
+{
+	ops->create_files	= NULL;
+	ops->setup		= oprofile_hrtimer_setup;
+	ops->shutdown		= oprofile_hrtimer_shutdown;
+	ops->start		= oprofile_hrtimer_start;
+	ops->stop		= oprofile_hrtimer_stop;
+	ops->cpu_type		= "timer";
+	printk(KERN_INFO "oprofile: using timer interrupt.\n");
+	return 0;
 }