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

* 'perf-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip: (57 commits)
  x86, perf events: Check if we have APIC enabled
  perf_event: Fix variable initialization in other codepaths
  perf kmem: Fix unused argument build warning
  perf symbols: perf_header__read_build_ids() offset'n'size should be u64
  perf symbols: dsos__read_build_ids() should read both user and kernel buildids
  perf tools: Align long options which have no short forms
  perf kmem: Show usage if no option is specified
  sched: Mark sched_clock() as notrace
  perf sched: Add max delay time snapshot
  perf tools: Correct size given to memset
  perf_event: Fix perf_swevent_hrtimer() variable initialization
  perf sched: Fix for getting task's execution time
  tracing/kprobes: Fix field creation's bad error handling
  perf_event: Cleanup for cpu_clock_perf_event_update()
  perf_event: Allocate children's perf_event_ctxp at the right time
  perf_event: Clean up __perf_event_init_context()
  hw-breakpoints: Modify breakpoints without unregistering them
  perf probe: Update perf-probe document
  perf probe: Support --del option
  trace-kprobe: Support delete probe syntax
  ...

4 files changed, 159 insertions, 108 deletions

diff --git a/kernel/hw_breakpoint.c b/kernel/hw_breakpoint.c
index cf5ee162841..366eedf949c 100644
--- a/kernel/hw_breakpoint.c
+++ b/kernel/hw_breakpoint.c
@@ -52,7 +52,7 @@
 static DEFINE_PER_CPU(unsigned int, nr_cpu_bp_pinned);
 
 /* Number of pinned task breakpoints in a cpu */
-static DEFINE_PER_CPU(unsigned int, task_bp_pinned[HBP_NUM]);
+static DEFINE_PER_CPU(unsigned int, nr_task_bp_pinned[HBP_NUM]);
 
 /* Number of non-pinned cpu/task breakpoints in a cpu */
 static DEFINE_PER_CPU(unsigned int, nr_bp_flexible);
@@ -73,7 +73,7 @@ static DEFINE_MUTEX(nr_bp_mutex);
 static unsigned int max_task_bp_pinned(int cpu)
 {
 	int i;
-	unsigned int *tsk_pinned = per_cpu(task_bp_pinned, cpu);
+	unsigned int *tsk_pinned = per_cpu(nr_task_bp_pinned, cpu);
 
 	for (i = HBP_NUM -1; i >= 0; i--) {
 		if (tsk_pinned[i] > 0)
@@ -83,15 +83,51 @@ static unsigned int max_task_bp_pinned(int cpu)
 	return 0;
 }
 
+static int task_bp_pinned(struct task_struct *tsk)
+{
+	struct perf_event_context *ctx = tsk->perf_event_ctxp;
+	struct list_head *list;
+	struct perf_event *bp;
+	unsigned long flags;
+	int count = 0;
+
+	if (WARN_ONCE(!ctx, "No perf context for this task"))
+		return 0;
+
+	list = &ctx->event_list;
+
+	spin_lock_irqsave(&ctx->lock, flags);
+
+	/*
+	 * The current breakpoint counter is not included in the list
+	 * at the open() callback time
+	 */
+	list_for_each_entry(bp, list, event_entry) {
+		if (bp->attr.type == PERF_TYPE_BREAKPOINT)
+			count++;
+	}
+
+	spin_unlock_irqrestore(&ctx->lock, flags);
+
+	return count;
+}
+
 /*
  * Report the number of pinned/un-pinned breakpoints we have in
  * a given cpu (cpu > -1) or in all of them (cpu = -1).
  */
-static void fetch_bp_busy_slots(struct bp_busy_slots *slots, int cpu)
+static void
+fetch_bp_busy_slots(struct bp_busy_slots *slots, struct perf_event *bp)
 {
+	int cpu = bp->cpu;
+	struct task_struct *tsk = bp->ctx->task;
+
 	if (cpu >= 0) {
 		slots->pinned = per_cpu(nr_cpu_bp_pinned, cpu);
-		slots->pinned += max_task_bp_pinned(cpu);
+		if (!tsk)
+			slots->pinned += max_task_bp_pinned(cpu);
+		else
+			slots->pinned += task_bp_pinned(tsk);
 		slots->flexible = per_cpu(nr_bp_flexible, cpu);
 
 		return;
@@ -101,7 +137,10 @@ static void fetch_bp_busy_slots(struct bp_busy_slots *slots, int cpu)
 		unsigned int nr;
 
 		nr = per_cpu(nr_cpu_bp_pinned, cpu);
-		nr += max_task_bp_pinned(cpu);
+		if (!tsk)
+			nr += max_task_bp_pinned(cpu);
+		else
+			nr += task_bp_pinned(tsk);
 
 		if (nr > slots->pinned)
 			slots->pinned = nr;
@@ -118,35 +157,12 @@ static void fetch_bp_busy_slots(struct bp_busy_slots *slots, int cpu)
  */
 static void toggle_bp_task_slot(struct task_struct *tsk, int cpu, bool enable)
 {
-	int count = 0;
-	struct perf_event *bp;
-	struct perf_event_context *ctx = tsk->perf_event_ctxp;
 	unsigned int *tsk_pinned;
-	struct list_head *list;
-	unsigned long flags;
-
-	if (WARN_ONCE(!ctx, "No perf context for this task"))
-		return;
-
-	list = &ctx->event_list;
-
-	spin_lock_irqsave(&ctx->lock, flags);
-
-	/*
-	 * The current breakpoint counter is not included in the list
-	 * at the open() callback time
-	 */
-	list_for_each_entry(bp, list, event_entry) {
-		if (bp->attr.type == PERF_TYPE_BREAKPOINT)
-			count++;
-	}
+	int count = 0;
 
-	spin_unlock_irqrestore(&ctx->lock, flags);
+	count = task_bp_pinned(tsk);
 
-	if (WARN_ONCE(count < 0, "No breakpoint counter found in the counter list"))
-		return;
-
-	tsk_pinned = per_cpu(task_bp_pinned, cpu);
+	tsk_pinned = per_cpu(nr_task_bp_pinned, cpu);
 	if (enable) {
 		tsk_pinned[count]++;
 		if (count > 0)
@@ -193,7 +209,7 @@ static void toggle_bp_slot(struct perf_event *bp, bool enable)
  *   - If attached to a single cpu, check:
  *
  *       (per_cpu(nr_bp_flexible, cpu) || (per_cpu(nr_cpu_bp_pinned, cpu)
- *           + max(per_cpu(task_bp_pinned, cpu)))) < HBP_NUM
+ *           + max(per_cpu(nr_task_bp_pinned, cpu)))) < HBP_NUM
  *
  *       -> If there are already non-pinned counters in this cpu, it means
  *          there is already a free slot for them.
@@ -204,7 +220,7 @@ static void toggle_bp_slot(struct perf_event *bp, bool enable)
  *   - If attached to every cpus, check:
  *
  *       (per_cpu(nr_bp_flexible, *) || (max(per_cpu(nr_cpu_bp_pinned, *))
- *           + max(per_cpu(task_bp_pinned, *)))) < HBP_NUM
+ *           + max(per_cpu(nr_task_bp_pinned, *)))) < HBP_NUM
  *
  *       -> This is roughly the same, except we check the number of per cpu
  *          bp for every cpu and we keep the max one. Same for the per tasks
@@ -216,7 +232,7 @@ static void toggle_bp_slot(struct perf_event *bp, bool enable)
  *   - If attached to a single cpu, check:
  *
  *       ((per_cpu(nr_bp_flexible, cpu) > 1) + per_cpu(nr_cpu_bp_pinned, cpu)
- *            + max(per_cpu(task_bp_pinned, cpu))) < HBP_NUM
+ *            + max(per_cpu(nr_task_bp_pinned, cpu))) < HBP_NUM
  *
  *       -> Same checks as before. But now the nr_bp_flexible, if any, must keep
  *          one register at least (or they will never be fed).
@@ -224,7 +240,7 @@ static void toggle_bp_slot(struct perf_event *bp, bool enable)
  *   - If attached to every cpus, check:
  *
  *       ((per_cpu(nr_bp_flexible, *) > 1) + max(per_cpu(nr_cpu_bp_pinned, *))
- *            + max(per_cpu(task_bp_pinned, *))) < HBP_NUM
+ *            + max(per_cpu(nr_task_bp_pinned, *))) < HBP_NUM
  */
 int reserve_bp_slot(struct perf_event *bp)
 {
@@ -233,7 +249,7 @@ int reserve_bp_slot(struct perf_event *bp)
 
 	mutex_lock(&nr_bp_mutex);
 
-	fetch_bp_busy_slots(&slots, bp->cpu);
+	fetch_bp_busy_slots(&slots, bp);
 
 	/* Flexible counters need to keep at least one slot */
 	if (slots.pinned + (!!slots.flexible) == HBP_NUM) {
@@ -259,7 +275,7 @@ void release_bp_slot(struct perf_event *bp)
 }
 
 
-int __register_perf_hw_breakpoint(struct perf_event *bp)
+int register_perf_hw_breakpoint(struct perf_event *bp)
 {
 	int ret;
 
@@ -276,19 +292,12 @@ int __register_perf_hw_breakpoint(struct perf_event *bp)
 	 * This is a quick hack that will be removed soon, once we remove
 	 * the tmp breakpoints from ptrace
 	 */
-	if (!bp->attr.disabled || bp->callback == perf_bp_event)
+	if (!bp->attr.disabled || !bp->overflow_handler)
 		ret = arch_validate_hwbkpt_settings(bp, bp->ctx->task);
 
 	return ret;
 }
 
-int register_perf_hw_breakpoint(struct perf_event *bp)
-{
-	bp->callback = perf_bp_event;
-
-	return __register_perf_hw_breakpoint(bp);
-}
-
 /**
  * register_user_hw_breakpoint - register a hardware breakpoint for user space
  * @attr: breakpoint attributes
@@ -297,7 +306,7 @@ int register_perf_hw_breakpoint(struct perf_event *bp)
  */
 struct perf_event *
 register_user_hw_breakpoint(struct perf_event_attr *attr,
-			    perf_callback_t triggered,
+			    perf_overflow_handler_t triggered,
 			    struct task_struct *tsk)
 {
 	return perf_event_create_kernel_counter(attr, -1, tsk->pid, triggered);
@@ -311,19 +320,40 @@ EXPORT_SYMBOL_GPL(register_user_hw_breakpoint);
  * @triggered: callback to trigger when we hit the breakpoint
  * @tsk: pointer to 'task_struct' of the process to which the address belongs
  */
-struct perf_event *
-modify_user_hw_breakpoint(struct perf_event *bp, struct perf_event_attr *attr,
-			  perf_callback_t triggered,
-			  struct task_struct *tsk)
+int modify_user_hw_breakpoint(struct perf_event *bp, struct perf_event_attr *attr)
 {
-	/*
-	 * FIXME: do it without unregistering
-	 * - We don't want to lose our slot
-	 * - If the new bp is incorrect, don't lose the older one
-	 */
-	unregister_hw_breakpoint(bp);
+	u64 old_addr = bp->attr.bp_addr;
+	int old_type = bp->attr.bp_type;
+	int old_len = bp->attr.bp_len;
+	int err = 0;
 
-	return perf_event_create_kernel_counter(attr, -1, tsk->pid, triggered);
+	perf_event_disable(bp);
+
+	bp->attr.bp_addr = attr->bp_addr;
+	bp->attr.bp_type = attr->bp_type;
+	bp->attr.bp_len = attr->bp_len;
+
+	if (attr->disabled)
+		goto end;
+
+	err = arch_validate_hwbkpt_settings(bp, bp->ctx->task);
+	if (!err)
+		perf_event_enable(bp);
+
+	if (err) {
+		bp->attr.bp_addr = old_addr;
+		bp->attr.bp_type = old_type;
+		bp->attr.bp_len = old_len;
+		if (!bp->attr.disabled)
+			perf_event_enable(bp);
+
+		return err;
+	}
+
+end:
+	bp->attr.disabled = attr->disabled;
+
+	return 0;
 }
 EXPORT_SYMBOL_GPL(modify_user_hw_breakpoint);
 
@@ -348,7 +378,7 @@ EXPORT_SYMBOL_GPL(unregister_hw_breakpoint);
  */
 struct perf_event **
 register_wide_hw_breakpoint(struct perf_event_attr *attr,
-			    perf_callback_t triggered)
+			    perf_overflow_handler_t triggered)
 {
 	struct perf_event **cpu_events, **pevent, *bp;
 	long err;
diff --git a/kernel/perf_event.c b/kernel/perf_event.c
index 40a996ec39f..e73e53c7582 100644
--- a/kernel/perf_event.c
+++ b/kernel/perf_event.c
@@ -36,7 +36,7 @@
 /*
  * Each CPU has a list of per CPU events:
  */
-DEFINE_PER_CPU(struct perf_cpu_context, perf_cpu_context);
+static DEFINE_PER_CPU(struct perf_cpu_context, perf_cpu_context);
 
 int perf_max_events __read_mostly = 1;
 static int perf_reserved_percpu __read_mostly;
@@ -567,7 +567,7 @@ static void __perf_event_disable(void *info)
  * is the current context on this CPU and preemption is disabled,
  * hence we can't get into perf_event_task_sched_out for this context.
  */
-static void perf_event_disable(struct perf_event *event)
+void perf_event_disable(struct perf_event *event)
 {
 	struct perf_event_context *ctx = event->ctx;
 	struct task_struct *task = ctx->task;
@@ -971,7 +971,7 @@ static void __perf_event_enable(void *info)
  * perf_event_for_each_child or perf_event_for_each as described
  * for perf_event_disable.
  */
-static void perf_event_enable(struct perf_event *event)
+void perf_event_enable(struct perf_event *event)
 {
 	struct perf_event_context *ctx = event->ctx;
 	struct task_struct *task = ctx->task;
@@ -1579,7 +1579,6 @@ static void
 __perf_event_init_context(struct perf_event_context *ctx,
 			    struct task_struct *task)
 {
-	memset(ctx, 0, sizeof(*ctx));
 	spin_lock_init(&ctx->lock);
 	mutex_init(&ctx->mutex);
 	INIT_LIST_HEAD(&ctx->group_list);
@@ -1654,7 +1653,7 @@ static struct perf_event_context *find_get_context(pid_t pid, int cpu)
 	}
 
 	if (!ctx) {
-		ctx = kmalloc(sizeof(struct perf_event_context), GFP_KERNEL);
+		ctx = kzalloc(sizeof(struct perf_event_context), GFP_KERNEL);
 		err = -ENOMEM;
 		if (!ctx)
 			goto errout;
@@ -4011,6 +4010,7 @@ static enum hrtimer_restart perf_swevent_hrtimer(struct hrtimer *hrtimer)
 	event->pmu->read(event);
 
 	data.addr = 0;
+	data.raw = NULL;
 	data.period = event->hw.last_period;
 	regs = get_irq_regs();
 	/*
@@ -4080,8 +4080,7 @@ static void cpu_clock_perf_event_update(struct perf_event *event)
 	u64 now;
 
 	now = cpu_clock(cpu);
-	prev = atomic64_read(&event->hw.prev_count);
-	atomic64_set(&event->hw.prev_count, now);
+	prev = atomic64_xchg(&event->hw.prev_count, now);
 	atomic64_add(now - prev, &event->count);
 }
 
@@ -4286,15 +4285,8 @@ static void bp_perf_event_destroy(struct perf_event *event)
 static const struct pmu *bp_perf_event_init(struct perf_event *bp)
 {
 	int err;
-	/*
-	 * The breakpoint is already filled if we haven't created the counter
-	 * through perf syscall
-	 * FIXME: manage to get trigerred to NULL if it comes from syscalls
-	 */
-	if (!bp->callback)
-		err = register_perf_hw_breakpoint(bp);
-	else
-		err = __register_perf_hw_breakpoint(bp);
+
+	err = register_perf_hw_breakpoint(bp);
 	if (err)
 		return ERR_PTR(err);
 
@@ -4308,6 +4300,7 @@ void perf_bp_event(struct perf_event *bp, void *data)
 	struct perf_sample_data sample;
 	struct pt_regs *regs = data;
 
+	sample.raw = NULL;
 	sample.addr = bp->attr.bp_addr;
 
 	if (!perf_exclude_event(bp, regs))
@@ -4390,7 +4383,7 @@ perf_event_alloc(struct perf_event_attr *attr,
 		   struct perf_event_context *ctx,
 		   struct perf_event *group_leader,
 		   struct perf_event *parent_event,
-		   perf_callback_t callback,
+		   perf_overflow_handler_t overflow_handler,
 		   gfp_t gfpflags)
 {
 	const struct pmu *pmu;
@@ -4433,10 +4426,10 @@ perf_event_alloc(struct perf_event_attr *attr,
 
 	event->state		= PERF_EVENT_STATE_INACTIVE;
 
-	if (!callback && parent_event)
-		callback = parent_event->callback;
+	if (!overflow_handler && parent_event)
+		overflow_handler = parent_event->overflow_handler;
 	
-	event->callback	= callback;
+	event->overflow_handler	= overflow_handler;
 
 	if (attr->disabled)
 		event->state = PERF_EVENT_STATE_OFF;
@@ -4776,7 +4769,8 @@ err_put_context:
  */
 struct perf_event *
 perf_event_create_kernel_counter(struct perf_event_attr *attr, int cpu,
-				 pid_t pid, perf_callback_t callback)
+				 pid_t pid,
+				 perf_overflow_handler_t overflow_handler)
 {
 	struct perf_event *event;
 	struct perf_event_context *ctx;
@@ -4793,7 +4787,7 @@ perf_event_create_kernel_counter(struct perf_event_attr *attr, int cpu,
 	}
 
 	event = perf_event_alloc(attr, cpu, ctx, NULL,
-				     NULL, callback, GFP_KERNEL);
+				 NULL, overflow_handler, GFP_KERNEL);
 	if (IS_ERR(event)) {
 		err = PTR_ERR(event);
 		goto err_put_context;
@@ -5090,7 +5084,7 @@ again:
  */
 int perf_event_init_task(struct task_struct *child)
 {
-	struct perf_event_context *child_ctx, *parent_ctx;
+	struct perf_event_context *child_ctx = NULL, *parent_ctx;
 	struct perf_event_context *cloned_ctx;
 	struct perf_event *event;
 	struct task_struct *parent = current;
@@ -5106,20 +5100,6 @@ int perf_event_init_task(struct task_struct *child)
 		return 0;
 
 	/*
-	 * This is executed from the parent task context, so inherit
-	 * events that have been marked for cloning.
-	 * First allocate and initialize a context for the child.
-	 */
-
-	child_ctx = kmalloc(sizeof(struct perf_event_context), GFP_KERNEL);
-	if (!child_ctx)
-		return -ENOMEM;
-
-	__perf_event_init_context(child_ctx, child);
-	child->perf_event_ctxp = child_ctx;
-	get_task_struct(child);
-
-	/*
 	 * If the parent's context is a clone, pin it so it won't get
 	 * swapped under us.
 	 */
@@ -5149,6 +5129,26 @@ int perf_event_init_task(struct task_struct *child)
 			continue;
 		}
 
+		if (!child->perf_event_ctxp) {
+			/*
+			 * This is executed from the parent task context, so
+			 * inherit events that have been marked for cloning.
+			 * First allocate and initialize a context for the
+			 * child.
+			 */
+
+			child_ctx = kzalloc(sizeof(struct perf_event_context),
+					    GFP_KERNEL);
+			if (!child_ctx) {
+				ret = -ENOMEM;
+				goto exit;
+			}
+
+			__perf_event_init_context(child_ctx, child);
+			child->perf_event_ctxp = child_ctx;
+			get_task_struct(child);
+		}
+
 		ret = inherit_group(event, parent, parent_ctx,
 					     child, child_ctx);
 		if (ret) {
@@ -5177,6 +5177,7 @@ int perf_event_init_task(struct task_struct *child)
 		get_ctx(child_ctx->parent_ctx);
 	}
 
+exit:
 	mutex_unlock(&parent_ctx->mutex);
 
 	perf_unpin_context(parent_ctx);
diff --git a/kernel/trace/trace_kprobe.c b/kernel/trace/trace_kprobe.c
index aff5f80b59b..b52d397e57e 100644
--- a/kernel/trace/trace_kprobe.c
+++ b/kernel/trace/trace_kprobe.c
@@ -606,23 +606,22 @@ static int create_trace_probe(int argc, char **argv)
 	 */
 	struct trace_probe *tp;
 	int i, ret = 0;
-	int is_return = 0;
+	int is_return = 0, is_delete = 0;
 	char *symbol = NULL, *event = NULL, *arg = NULL, *group = NULL;
 	unsigned long offset = 0;
 	void *addr = NULL;
 	char buf[MAX_EVENT_NAME_LEN];
 
-	if (argc < 2) {
-		pr_info("Probe point is not specified.\n");
-		return -EINVAL;
-	}
-
+	/* argc must be >= 1 */
 	if (argv[0][0] == 'p')
 		is_return = 0;
 	else if (argv[0][0] == 'r')
 		is_return = 1;
+	else if (argv[0][0] == '-')
+		is_delete = 1;
 	else {
-		pr_info("Probe definition must be started with 'p' or 'r'.\n");
+		pr_info("Probe definition must be started with 'p', 'r' or"
+			" '-'.\n");
 		return -EINVAL;
 	}
 
@@ -642,7 +641,29 @@ static int create_trace_probe(int argc, char **argv)
 			return -EINVAL;
 		}
 	}
+	if (!group)
+		group = KPROBE_EVENT_SYSTEM;
 
+	if (is_delete) {
+		if (!event) {
+			pr_info("Delete command needs an event name.\n");
+			return -EINVAL;
+		}
+		tp = find_probe_event(event, group);
+		if (!tp) {
+			pr_info("Event %s/%s doesn't exist.\n", group, event);
+			return -ENOENT;
+		}
+		/* delete an event */
+		unregister_trace_probe(tp);
+		free_trace_probe(tp);
+		return 0;
+	}
+
+	if (argc < 2) {
+		pr_info("Probe point is not specified.\n");
+		return -EINVAL;
+	}
 	if (isdigit(argv[1][0])) {
 		if (is_return) {
 			pr_info("Return probe point must be a symbol.\n");
@@ -671,8 +692,6 @@ static int create_trace_probe(int argc, char **argv)
 	argc -= 2; argv += 2;
 
 	/* setup a probe */
-	if (!group)
-		group = KPROBE_EVENT_SYSTEM;
 	if (!event) {
 		/* Make a new event name */
 		if (symbol)
@@ -1114,7 +1133,7 @@ static int kprobe_event_define_fields(struct ftrace_event_call *event_call)
 	struct trace_probe *tp = (struct trace_probe *)event_call->data;
 
 	ret = trace_define_common_fields(event_call);
-	if (!ret)
+	if (ret)
 		return ret;
 
 	DEFINE_FIELD(unsigned long, ip, FIELD_STRING_IP, 0);
@@ -1132,7 +1151,7 @@ static int kretprobe_event_define_fields(struct ftrace_event_call *event_call)
 	struct trace_probe *tp = (struct trace_probe *)event_call->data;
 
 	ret = trace_define_common_fields(event_call);
-	if (!ret)
+	if (ret)
 		return ret;
 
 	DEFINE_FIELD(unsigned long, func, FIELD_STRING_FUNC, 0);
diff --git a/kernel/trace/trace_ksym.c b/kernel/trace/trace_ksym.c
index ddfa0fd43bc..acb87d4a4ac 100644
--- a/kernel/trace/trace_ksym.c
+++ b/kernel/trace/trace_ksym.c
@@ -79,11 +79,12 @@ void ksym_collect_stats(unsigned long hbp_hit_addr)
 }
 #endif /* CONFIG_PROFILE_KSYM_TRACER */
 
-void ksym_hbp_handler(struct perf_event *hbp, void *data)
+void ksym_hbp_handler(struct perf_event *hbp, int nmi,
+		      struct perf_sample_data *data,
+		      struct pt_regs *regs)
 {
 	struct ring_buffer_event *event;
 	struct ksym_trace_entry *entry;
-	struct pt_regs *regs = data;
 	struct ring_buffer *buffer;
 	int pc;