---

yaml
---
r: 121272
b: refs/heads/master
c: f3134de
h: refs/heads/master
v: v3

[refs]

@@ -1,2 +1,2 @@
 ---
-refs/heads/master: 361b73d5c34f59c3fd107bb9dbe7a1fbff2c2517
+refs/heads/master: f3134de60624829a57741c1f3796847d4de165f6

trunk/arch/x86/kernel/Makefile

@@ -14,12 +14,6 @@ CFLAGS_REMOVE_paravirt-spinlocks.o = -pg
 CFLAGS_REMOVE_ftrace.o = -pg
 endif
 
-ifdef CONFIG_FUNCTION_GRAPH_TRACER
-# Don't trace __switch_to() but let it for function tracer
-CFLAGS_REMOVE_process_32.o = -pg
-CFLAGS_REMOVE_process_64.o = -pg
-endif
-
 #
 # vsyscalls (which work on the user stack) should have
 # no stack-protector checks:

trunk/arch/x86/kernel/ftrace.c

@@ -476,7 +476,10 @@ void prepare_ftrace_return(unsigned long *parent, unsigned long self_addr)
 				&return_to_handler;
 
 	/* Nmi's are currently unsupported */
-	if (atomic_read(&in_nmi))
+	if (unlikely(atomic_read(&in_nmi)))
+		return;
+
+	if (unlikely(atomic_read(&current->tracing_graph_pause)))
 		return;
 
 	/*

trunk/arch/x86/kernel/process_32.c

@@ -38,6 +38,7 @@
 #include <linux/percpu.h>
 #include <linux/prctl.h>
 #include <linux/dmi.h>
+#include <linux/ftrace.h>
 
 #include <asm/uaccess.h>
 #include <asm/pgtable.h>
@@ -548,7 +549,8 @@ __switch_to_xtra(struct task_struct *prev_p, struct task_struct *next_p,
  * the task-switch, and shows up in ret_from_fork in entry.S,
  * for example.
  */
-struct task_struct * __switch_to(struct task_struct *prev_p, struct task_struct *next_p)
+__notrace_funcgraph struct task_struct *
+__switch_to(struct task_struct *prev_p, struct task_struct *next_p)
 {
 	struct thread_struct *prev = &prev_p->thread,
 				 *next = &next_p->thread;

trunk/arch/x86/kernel/process_64.c

@@ -39,6 +39,7 @@
 #include <linux/prctl.h>
 #include <linux/uaccess.h>
 #include <linux/io.h>
+#include <linux/ftrace.h>
 
 #include <asm/pgtable.h>
 #include <asm/system.h>
@@ -551,8 +552,9 @@ static inline void __switch_to_xtra(struct task_struct *prev_p,
  * - could test fs/gs bitsliced
  *
  * Kprobes not supported here. Set the probe on schedule instead.
+ * Function graph tracer not supported too.
  */
-struct task_struct *
+__notrace_funcgraph struct task_struct *
 __switch_to(struct task_struct *prev_p, struct task_struct *next_p)
 {
 	struct thread_struct *prev = &prev_p->thread;

trunk/include/linux/ftrace.h

@@ -369,6 +369,14 @@ struct ftrace_graph_ret {
 };
 
 #ifdef CONFIG_FUNCTION_GRAPH_TRACER
+
+/*
+ * Sometimes we don't want to trace a function with the function
+ * graph tracer but we want them to keep traced by the usual function
+ * tracer if the function graph tracer is not configured.
+ */
+#define __notrace_funcgraph		notrace
+
 #define FTRACE_RETFUNC_DEPTH 50
 #define FTRACE_RETSTACK_ALLOC_SIZE 32
 /* Type of the callback handlers for tracing function graph*/
@@ -393,14 +401,30 @@ static inline int task_curr_ret_stack(struct task_struct *t)
 {
 	return t->curr_ret_stack;
 }
+
+static inline void pause_graph_tracing(void)
+{
+	atomic_inc(&current->tracing_graph_pause);
+}
+
+static inline void unpause_graph_tracing(void)
+{
+	atomic_dec(&current->tracing_graph_pause);
+}
 #else
+
+#define __notrace_funcgraph
+
 static inline void ftrace_graph_init_task(struct task_struct *t) { }
 static inline void ftrace_graph_exit_task(struct task_struct *t) { }
 
 static inline int task_curr_ret_stack(struct task_struct *tsk)
 {
 	return -1;
 }
+
+static inline void pause_graph_tracing(void) { }
+static inline void unpause_graph_tracing(void) { }
 #endif
 
 #ifdef CONFIG_TRACING

trunk/include/linux/sched.h

@@ -1379,6 +1379,8 @@ struct task_struct {
 	 * because of depth overrun.
 	 */
 	atomic_t trace_overrun;
+	/* Pause for the tracing */
+	atomic_t tracing_graph_pause;
 #endif
 #ifdef CONFIG_TRACING
 	/* state flags for use by tracers */

trunk/kernel/Makefile

@@ -21,10 +21,6 @@ CFLAGS_REMOVE_cgroup-debug.o = -pg
 CFLAGS_REMOVE_sched_clock.o = -pg
 CFLAGS_REMOVE_sched.o = -pg
 endif
-ifdef CONFIG_FUNCTION_GRAPH_TRACER
-CFLAGS_REMOVE_extable.o = -pg # For __kernel_text_address()
-CFLAGS_REMOVE_module.o = -pg # For __module_text_address()
-endif
 
 obj-$(CONFIG_FREEZER) += freezer.o
 obj-$(CONFIG_PROFILING) += profile.o

trunk/kernel/extable.c

@@ -17,6 +17,7 @@
 */
 #include <linux/module.h>
 #include <linux/init.h>
+#include <linux/ftrace.h>
 #include <asm/uaccess.h>
 #include <asm/sections.h>
 
@@ -40,7 +41,7 @@ const struct exception_table_entry *search_exception_tables(unsigned long addr)
 	return e;
 }
 
-int core_kernel_text(unsigned long addr)
+__notrace_funcgraph int core_kernel_text(unsigned long addr)
 {
 	if (addr >= (unsigned long)_stext &&
 	    addr <= (unsigned long)_etext)
@@ -53,7 +54,7 @@ int core_kernel_text(unsigned long addr)
 	return 0;
 }
 
-int __kernel_text_address(unsigned long addr)
+__notrace_funcgraph int __kernel_text_address(unsigned long addr)
 {
 	if (core_kernel_text(addr))
 		return 1;

trunk/kernel/module.c

@@ -2704,7 +2704,7 @@ int is_module_address(unsigned long addr)
 
 
 /* Is this a valid kernel address? */
-struct module *__module_text_address(unsigned long addr)
+__notrace_funcgraph struct module *__module_text_address(unsigned long addr)
 {
 	struct module *mod;
 

trunk/kernel/trace/ftrace.c

@@ -1998,6 +1998,7 @@ static int alloc_retstack_tasklist(struct ftrace_ret_stack **ret_stack_list)
 			/* Make sure IRQs see the -1 first: */
 			barrier();
 			t->ret_stack = ret_stack_list[start++];
+			atomic_set(&t->tracing_graph_pause, 0);
 			atomic_set(&t->trace_overrun, 0);
 		}
 	} while_each_thread(g, t);
@@ -2077,6 +2078,7 @@ void ftrace_graph_init_task(struct task_struct *t)
 		if (!t->ret_stack)
 			return;
 		t->curr_ret_stack = -1;
+		atomic_set(&t->tracing_graph_pause, 0);
 		atomic_set(&t->trace_overrun, 0);
 	} else
 		t->ret_stack = NULL;

trunk/kernel/trace/trace.c

@@ -44,13 +44,14 @@
 unsigned long __read_mostly	tracing_max_latency = (cycle_t)ULONG_MAX;
 unsigned long __read_mostly	tracing_thresh;
 
-/* We need to change this state when a selftest is running.
+/*
+ * We need to change this state when a selftest is running.
  * A selftest will lurk into the ring-buffer to count the
  * entries inserted during the selftest although some concurrent
  * insertions into the ring-buffer such as ftrace_printk could occurred
  * at the same time, giving false positive or negative results.
  */
-static atomic_t tracing_selftest_running = ATOMIC_INIT(0);
+static bool __read_mostly tracing_selftest_running;
 
 /* For tracers that don't implement custom flags */
 static struct tracer_opt dummy_tracer_opt[] = {
@@ -574,6 +575,8 @@ int register_tracer(struct tracer *type)
 	unlock_kernel();
 	mutex_lock(&trace_types_lock);
 
+	tracing_selftest_running = true;
+
 	for (t = trace_types; t; t = t->next) {
 		if (strcmp(type->name, t->name) == 0) {
 			/* already found */
@@ -598,7 +601,6 @@ int register_tracer(struct tracer *type)
 		struct trace_array *tr = &global_trace;
 		int i;
 
-		atomic_set(&tracing_selftest_running, 1);
 		/*
 		 * Run a selftest on this tracer.
 		 * Here we reset the trace buffer, and set the current
@@ -613,7 +615,6 @@ int register_tracer(struct tracer *type)
 		/* the test is responsible for initializing and enabling */
 		pr_info("Testing tracer %s: ", type->name);
 		ret = type->selftest(type, tr);
-		atomic_set(&tracing_selftest_running, 0);
 		/* the test is responsible for resetting too */
 		current_trace = saved_tracer;
 		if (ret) {
@@ -635,6 +636,7 @@ int register_tracer(struct tracer *type)
 		max_tracer_type_len = len;
 
  out:
+	tracing_selftest_running = false;
 	mutex_unlock(&trace_types_lock);
 	lock_kernel();
 
@@ -3588,24 +3590,17 @@ static __init int tracer_init_debugfs(void)
 
 int trace_vprintk(unsigned long ip, int depth, const char *fmt, va_list args)
 {
-	/*
-	 * Raw Spinlock because a normal spinlock would be traced here
-	 * and append an irrelevant couple spin_lock_irqsave/
-	 * spin_unlock_irqrestore traced by ftrace around this
-	 * TRACE_PRINTK trace.
-	 */
-	static raw_spinlock_t trace_buf_lock =
-				(raw_spinlock_t)__RAW_SPIN_LOCK_UNLOCKED;
+	static DEFINE_SPINLOCK(trace_buf_lock);
 	static char trace_buf[TRACE_BUF_SIZE];
 
 	struct ring_buffer_event *event;
 	struct trace_array *tr = &global_trace;
 	struct trace_array_cpu *data;
-	struct print_entry *entry;
-	unsigned long flags, irq_flags;
 	int cpu, len = 0, size, pc;
+	struct print_entry *entry;
+	unsigned long irq_flags;
 
-	if (tracing_disabled || atomic_read(&tracing_selftest_running))
+	if (tracing_disabled || tracing_selftest_running)
 		return 0;
 
 	pc = preempt_count();
@@ -3616,8 +3611,8 @@ int trace_vprintk(unsigned long ip, int depth, const char *fmt, va_list args)
 	if (unlikely(atomic_read(&data->disabled)))
 		goto out;
 
-	local_irq_save(flags);
-	__raw_spin_lock(&trace_buf_lock);
+	pause_graph_tracing();
+	spin_lock_irqsave(&trace_buf_lock, irq_flags);
 	len = vsnprintf(trace_buf, TRACE_BUF_SIZE, fmt, args);
 
 	len = min(len, TRACE_BUF_SIZE-1);
@@ -3628,7 +3623,7 @@ int trace_vprintk(unsigned long ip, int depth, const char *fmt, va_list args)
 	if (!event)
 		goto out_unlock;
 	entry = ring_buffer_event_data(event);
-	tracing_generic_entry_update(&entry->ent, flags, pc);
+	tracing_generic_entry_update(&entry->ent, irq_flags, pc);
 	entry->ent.type			= TRACE_PRINT;
 	entry->ip			= ip;
 	entry->depth			= depth;
@@ -3638,9 +3633,8 @@ int trace_vprintk(unsigned long ip, int depth, const char *fmt, va_list args)
 	ring_buffer_unlock_commit(tr->buffer, event, irq_flags);
 
  out_unlock:
-	__raw_spin_unlock(&trace_buf_lock);
-	local_irq_restore(flags);
-
+	spin_unlock_irqrestore(&trace_buf_lock, irq_flags);
+	unpause_graph_tracing();
  out:
 	preempt_enable_notrace();
 

trunk/kernel/trace/trace_functions_graph.c

@@ -570,11 +570,36 @@ print_graph_function(struct trace_iterator *iter)
 	}
 }
 
+static void print_graph_headers(struct seq_file *s)
+{
+	/* 1st line */
+	seq_printf(s, "# ");
+	if (tracer_flags.val & TRACE_GRAPH_PRINT_CPU)
+		seq_printf(s, "CPU ");
+	if (tracer_flags.val & TRACE_GRAPH_PRINT_PROC)
+		seq_printf(s, "TASK/PID     ");
+	if (tracer_flags.val & TRACE_GRAPH_PRINT_OVERHEAD)
+		seq_printf(s, "OVERHEAD/");
+	seq_printf(s, "DURATION            FUNCTION CALLS\n");
+
+	/* 2nd line */
+	seq_printf(s, "# ");
+	if (tracer_flags.val & TRACE_GRAPH_PRINT_CPU)
+		seq_printf(s, "|   ");
+	if (tracer_flags.val & TRACE_GRAPH_PRINT_PROC)
+		seq_printf(s, "|      |     ");
+	if (tracer_flags.val & TRACE_GRAPH_PRINT_OVERHEAD) {
+		seq_printf(s, "|        ");
+		seq_printf(s, "|                   |   |   |   |\n");
+	} else
+		seq_printf(s, "    |               |   |   |   |\n");
+}
 static struct tracer graph_trace __read_mostly = {
-	.name	     = "function_graph",
-	.init	     = graph_trace_init,
-	.reset	     = graph_trace_reset,
-	.print_line = print_graph_function,
+	.name	     	= "function_graph",
+	.init	     	= graph_trace_init,
+	.reset	     	= graph_trace_reset,
+	.print_line	= print_graph_function,
+	.print_header	= print_graph_headers,
 	.flags		= &tracer_flags,
 };