tracing: add a tracer to catch execution time of kernel functions

Impact: add new tracing plugin which can trace full (entry+exit) function calls

This tracer uses the low level function return ftrace plugin to
measure the execution time of the kernel functions.

The first field is the caller of the function, the second is the
measured function, and the last one is the execution time in
nanoseconds.

- v3:

- HAVE_FUNCTION_RET_TRACER have been added. Each arch that support ftrace return
  should enable it.
- ftrace_return_stub becomes ftrace_stub.
- CONFIG_FUNCTION_RET_TRACER depends now on CONFIG_FUNCTION_TRACER
- Return traces printing can be used for other tracers on trace.c
- Adapt to the new tracing API (no more ctrl_update callback)
- Correct the check of "disabled" during insertion.
- Minor changes...

Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Steven Rostedt <rostedt@goodmis.org>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
diff --git a/kernel/trace/Kconfig b/kernel/trace/Kconfig
index fc4febc..d986216 100644
--- a/kernel/trace/Kconfig
+++ b/kernel/trace/Kconfig
@@ -9,6 +9,9 @@
 config HAVE_FUNCTION_TRACER
 	bool
 
+config HAVE_FUNCTION_RET_TRACER
+	bool
+
 config HAVE_FUNCTION_TRACE_MCOUNT_TEST
 	bool
 	help
@@ -54,6 +57,17 @@
 	  (the bootup default), then the overhead of the instructions is very
 	  small and not measurable even in micro-benchmarks.
 
+config FUNCTION_RET_TRACER
+	bool "Kernel Function return Tracer"
+	depends on !DYNAMIC_FTRACE
+	depends on HAVE_FUNCTION_RET_TRACER
+	depends on FUNCTION_TRACER
+	help
+	  Enable the kernel to trace a function at its return.
+	  It's first purpose is to trace the duration of functions.
+	  This is done by setting the current return address on the thread
+	  info structure of the current task.
+
 config IRQSOFF_TRACER
 	bool "Interrupts-off Latency Tracer"
 	default n
diff --git a/kernel/trace/Makefile b/kernel/trace/Makefile
index c8228b1..3e1f361 100644
--- a/kernel/trace/Makefile
+++ b/kernel/trace/Makefile
@@ -24,5 +24,6 @@
 obj-$(CONFIG_STACK_TRACER) += trace_stack.o
 obj-$(CONFIG_MMIOTRACE) += trace_mmiotrace.o
 obj-$(CONFIG_BOOT_TRACER) += trace_boot.o
+obj-$(CONFIG_FUNCTION_RET_TRACER) += trace_functions_return.o
 
 libftrace-y := ftrace.o
diff --git a/kernel/trace/ftrace.c b/kernel/trace/ftrace.c
index 4d2e751..f03fe74 100644
--- a/kernel/trace/ftrace.c
+++ b/kernel/trace/ftrace.c
@@ -1484,3 +1484,19 @@
 	return ret;
 }
 
+#ifdef CONFIG_FUNCTION_RET_TRACER
+trace_function_return_t ftrace_function_return =
+			(trace_function_return_t)ftrace_stub;
+void register_ftrace_return(trace_function_return_t func)
+{
+	ftrace_function_return = func;
+}
+
+void unregister_ftrace_return(void)
+{
+	ftrace_function_return = (trace_function_return_t)ftrace_stub;
+}
+#endif
+
+
+
diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c
index 216bbe7..a3f7ae9 100644
--- a/kernel/trace/trace.c
+++ b/kernel/trace/trace.c
@@ -244,13 +244,6 @@
 	return nsecs / 1000;
 }
 
-/*
- * TRACE_ITER_SYM_MASK masks the options in trace_flags that
- * control the output of kernel symbols.
- */
-#define TRACE_ITER_SYM_MASK \
-	(TRACE_ITER_PRINT_PARENT|TRACE_ITER_SYM_OFFSET|TRACE_ITER_SYM_ADDR)
-
 /* These must match the bit postions in trace_iterator_flags */
 static const char *trace_options[] = {
 	"print-parent",
@@ -810,6 +803,35 @@
 	ring_buffer_unlock_commit(tr->buffer, event, irq_flags);
 }
 
+#ifdef CONFIG_FUNCTION_RET_TRACER
+static void __trace_function_return(struct trace_array *tr,
+				struct trace_array_cpu *data,
+				struct ftrace_retfunc *trace,
+				unsigned long flags,
+				int pc)
+{
+	struct ring_buffer_event *event;
+	struct ftrace_ret_entry *entry;
+	unsigned long irq_flags;
+
+	if (unlikely(local_read(&__get_cpu_var(ftrace_cpu_disabled))))
+		return;
+
+	event = ring_buffer_lock_reserve(global_trace.buffer, sizeof(*entry),
+					 &irq_flags);
+	if (!event)
+		return;
+	entry	= ring_buffer_event_data(event);
+	tracing_generic_entry_update(&entry->ent, flags, pc);
+	entry->ent.type			= TRACE_FN_RET;
+	entry->ip			= trace->func;
+	entry->parent_ip	= trace->ret;
+	entry->rettime		= trace->rettime;
+	entry->calltime		= trace->calltime;
+	ring_buffer_unlock_commit(global_trace.buffer, event, irq_flags);
+}
+#endif
+
 void
 ftrace(struct trace_array *tr, struct trace_array_cpu *data,
        unsigned long ip, unsigned long parent_ip, unsigned long flags,
@@ -1038,6 +1060,29 @@
 	raw_local_irq_restore(flags);
 }
 
+#ifdef CONFIG_FUNCTION_RET_TRACER
+void trace_function_return(struct ftrace_retfunc *trace)
+{
+	struct trace_array *tr = &global_trace;
+	struct trace_array_cpu *data;
+	unsigned long flags;
+	long disabled;
+	int cpu;
+	int pc;
+
+	raw_local_irq_save(flags);
+	cpu = raw_smp_processor_id();
+	data = tr->data[cpu];
+	disabled = atomic_inc_return(&data->disabled);
+	if (likely(disabled == 1)) {
+		pc = preempt_count();
+		__trace_function_return(tr, data, trace, flags, pc);
+	}
+	atomic_dec(&data->disabled);
+	raw_local_irq_restore(flags);
+}
+#endif /* CONFIG_FUNCTION_RET_TRACER */
+
 static struct ftrace_ops trace_ops __read_mostly =
 {
 	.func = function_trace_call,
@@ -1285,7 +1330,7 @@
 # define IP_FMT "%016lx"
 #endif
 
-static int
+int
 seq_print_ip_sym(struct trace_seq *s, unsigned long ip, unsigned long sym_flags)
 {
 	int ret;
@@ -1738,6 +1783,10 @@
 			trace_seq_print_cont(s, iter);
 		break;
 	}
+	case TRACE_FN_RET: {
+		return print_return_function(iter);
+		break;
+	}
 	}
 	return TRACE_TYPE_HANDLED;
 }
diff --git a/kernel/trace/trace.h b/kernel/trace/trace.h
index 9781450..e40ce0c 100644
--- a/kernel/trace/trace.h
+++ b/kernel/trace/trace.h
@@ -22,6 +22,7 @@
 	TRACE_MMIO_RW,
 	TRACE_MMIO_MAP,
 	TRACE_BOOT,
+	TRACE_FN_RET,
 
 	__TRACE_LAST_TYPE
 };
@@ -48,6 +49,15 @@
 	unsigned long		ip;
 	unsigned long		parent_ip;
 };
+
+/* Function return entry */
+struct ftrace_ret_entry {
+	struct trace_entry	ent;
+	unsigned long		ip;
+	unsigned long		parent_ip;
+	unsigned long long	calltime;
+	unsigned long long	rettime;
+};
 extern struct tracer boot_tracer;
 
 /*
@@ -218,6 +228,7 @@
 		IF_ASSIGN(var, ent, struct trace_mmiotrace_map,		\
 			  TRACE_MMIO_MAP);				\
 		IF_ASSIGN(var, ent, struct trace_boot, TRACE_BOOT);	\
+		IF_ASSIGN(var, ent, struct ftrace_ret_entry, TRACE_FN_RET); \
 		__ftrace_bad_type();					\
 	} while (0)
 
@@ -321,6 +332,8 @@
 		    unsigned long ip,
 		    unsigned long parent_ip,
 		    unsigned long flags, int pc);
+void
+trace_function_return(struct ftrace_retfunc *trace);
 
 void tracing_start_cmdline_record(void);
 void tracing_stop_cmdline_record(void);
@@ -393,6 +406,10 @@
 extern int trace_seq_printf(struct trace_seq *s, const char *fmt, ...);
 extern void trace_seq_print_cont(struct trace_seq *s,
 				 struct trace_iterator *iter);
+
+extern int
+seq_print_ip_sym(struct trace_seq *s, unsigned long ip,
+		unsigned long sym_flags);
 extern ssize_t trace_seq_to_user(struct trace_seq *s, char __user *ubuf,
 				 size_t cnt);
 extern long ns2usecs(cycle_t nsec);
@@ -400,6 +417,17 @@
 
 extern unsigned long trace_flags;
 
+/* Standard output formatting function used for function return traces */
+#ifdef CONFIG_FUNCTION_RET_TRACER
+extern enum print_line_t print_return_function(struct trace_iterator *iter);
+#else
+static inline enum print_line_t
+print_return_function(struct trace_iterator *iter)
+{
+	return TRACE_TYPE_UNHANDLED;
+}
+#endif
+
 /*
  * trace_iterator_flags is an enumeration that defines bit
  * positions into trace_flags that controls the output.
@@ -422,6 +450,13 @@
 	TRACE_ITER_PREEMPTONLY		= 0x800,
 };
 
+/*
+ * TRACE_ITER_SYM_MASK masks the options in trace_flags that
+ * control the output of kernel symbols.
+ */
+#define TRACE_ITER_SYM_MASK \
+	(TRACE_ITER_PRINT_PARENT|TRACE_ITER_SYM_OFFSET|TRACE_ITER_SYM_ADDR)
+
 extern struct tracer nop_trace;
 
 /**
diff --git a/kernel/trace/trace_functions_return.c b/kernel/trace/trace_functions_return.c
new file mode 100644
index 0000000..7680b21
--- /dev/null
+++ b/kernel/trace/trace_functions_return.c
@@ -0,0 +1,82 @@
+/*
+ *
+ * Function return tracer.
+ * Copyright (c) 2008 Frederic Weisbecker <fweisbec@gmail.com>
+ * Mostly borrowed from function tracer which
+ * is Copyright (c) Steven Rostedt <srostedt@redhat.com>
+ *
+ */
+#include <linux/debugfs.h>
+#include <linux/uaccess.h>
+#include <linux/ftrace.h>
+#include <linux/fs.h>
+
+#include "trace.h"
+
+
+static void start_return_trace(struct trace_array *tr)
+{
+	register_ftrace_return(&trace_function_return);
+}
+
+static void stop_return_trace(struct trace_array *tr)
+{
+	unregister_ftrace_return();
+}
+
+static void return_trace_init(struct trace_array *tr)
+{
+	int cpu;
+	for_each_online_cpu(cpu)
+		tracing_reset(tr, cpu);
+
+	start_return_trace(tr);
+}
+
+static void return_trace_reset(struct trace_array *tr)
+{
+		stop_return_trace(tr);
+}
+
+
+enum print_line_t
+print_return_function(struct trace_iterator *iter)
+{
+	struct trace_seq *s = &iter->seq;
+	struct trace_entry *entry = iter->ent;
+	struct ftrace_ret_entry *field;
+	int ret;
+
+	if (entry->type == TRACE_FN_RET) {
+		trace_assign_type(field, entry);
+		ret = trace_seq_printf(s, "%pF -> ", (void *)field->parent_ip);
+		if (!ret)
+			return TRACE_TYPE_PARTIAL_LINE;
+		ret = seq_print_ip_sym(s, field->ip,
+					trace_flags & TRACE_ITER_SYM_MASK);
+		if (!ret)
+			return TRACE_TYPE_PARTIAL_LINE;
+		ret = trace_seq_printf(s, " (%llu ns)\n",
+					field->rettime - field->calltime);
+		if (!ret)
+			return TRACE_TYPE_PARTIAL_LINE;
+		else
+			return TRACE_TYPE_HANDLED;
+	}
+	return TRACE_TYPE_UNHANDLED;
+}
+
+static struct tracer return_trace __read_mostly =
+{
+	.name	     = "return",
+	.init	     = return_trace_init,
+	.reset	     = return_trace_reset,
+	.print_line = print_return_function
+};
+
+static __init int init_return_trace(void)
+{
+	return register_tracer(&return_trace);
+}
+
+device_initcall(init_return_trace);