tracing: add function profiler

Impact: new profiling feature

This patch adds a function profiler. In debugfs/tracing/ two new
files are created.

  function_profile_enabled  - to enable or disable profiling

  trace_stat/functions   - the profiled functions.

For example:

  echo 1 > /debugfs/tracing/function_profile_enabled
  ./hackbench 50
  echo 0 > /debugfs/tracing/function_profile_enabled

yields:

  cat /debugfs/tracing/trace_stat/functions

  Function                               Hit
  --------                               ---
  _spin_lock                        10106442
  _spin_unlock                      10097492
  kfree                              6013704
  _spin_unlock_irqrestore            4423941
  _spin_lock_irqsave                 4406825
  __phys_addr                        4181686
  __slab_free                        4038222
  dput                               4030130
  path_put                           4023387
  unroll_tree_refs                   4019532
[...]

The most hit functions are listed first. Functions that are not
hit are not listed.

This feature depends on and uses dynamic function tracing. When the
function profiling is disabled, no overhead occurs. But it still
takes up around 300KB to hold the data, thus it is not recomended
to keep it enabled for systems low on memory.

When a '1' is echoed into the function_profile_enabled file, the
counters for is function is reset back to zero. Thus you can see what
functions are hit most by different programs.

Signed-off-by: Steven Rostedt <srostedt@redhat.com>

include/linux/ftrace.h

@@ -153,6 +153,10 @@ struct dyn_ftrace {
 		unsigned long		flags;
 		struct dyn_ftrace	*newlist;
 	};
+#ifdef CONFIG_FUNCTION_PROFILER
+	unsigned long			counter;
+	struct hlist_node		node;
+#endif
 	struct dyn_arch_ftrace		arch;
 };
 

kernel/trace/Kconfig

@@ -105,6 +105,7 @@ config FUNCTION_GRAPH_TRACER
 	  This is done by setting the current return address on the current
 	  task structure into a stack of calls.
 
+
 config IRQSOFF_TRACER
 	bool "Interrupts-off Latency Tracer"
 	default n
@@ -376,6 +377,24 @@ config DYNAMIC_FTRACE
 	 were made. If so, it runs stop_machine (stops all CPUS)
 	 and modifies the code to jump over the call to ftrace.
 
+config FUNCTION_PROFILER
+	bool "Kernel function profiler"
+	depends on DYNAMIC_FTRACE
+	default n
+	help
+	 This option enables the kernel function profiler. When the dynamic
+	 function tracing is enabled, a counter is added into the function
+	 records used by the dynamic function tracer. A file is created in
+	 debugfs called function_profile_enabled which defaults to zero.
+	 When a 1 is echoed into this file profiling begins, and when a
+	 zero is entered, profiling stops. A file in the trace_stats
+	 directory called functions, that show the list of functions that
+	 have been hit and their counters.
+
+	 This takes up around 320K more memory.
+
+	 If in doubt, say N
+
 config FTRACE_MCOUNT_RECORD
 	def_bool y
 	depends on DYNAMIC_FTRACE