tracing/ftrace: Introduce the big kernel lock tracer

Introduce the Big Kernel Lock tracer.
This new tracer lets one to observe the latencies caused by the lock_kernel() function.
It captures the time when the task request the spinlock, the time when the spinlock is hold and the time it is released. This way we can measure the latency for a task that wait for this spinlock and the latency caused by the time this lock is hold.

I will surely add some improvements like:

_ Adding a stack trace to know which function locked or unlocked the kernel. _ Adding a script to get some statistics about this: which functions hold it often, which hold it for a long time and some other things....

Changes in this V2:

_ Correction of comments
_ Make static inline the insertion functions when the tracer is not configured _ Remove FTRACE and HAVE_FTRACE as dependencies. Bkl Tracer only depends on TRACING and
STACKTRACE for further features.
_ Use trace_assign_type instead of unsure cast
_ Several cleanups...

Signed-off-by Frederic Weisbecker <[email protected]>

Cc Steven Rostedt <[email protected]>

---
 kernel/trace/Kconfig  |   10 ++++++++++
 kernel/trace/Makefile |    1 +
 kernel/trace/trace.h  |   18 ++++++++++++++++++
 lib/kernel_lock.c     |   36 ++++++++++++++++++++++++++++++++++--
 4 files changed, 63 insertions(+), 2 deletions(-)

kernel/trace/Kconfig +10 -0

index 396aea1..a87484d 100644
@@ -144,6 +144,16 @@ config STACK_TRACER
 	  This tracer records the max stack of the kernel, and displays
 	  it in debugfs/tracing/stack_trace
 
+config BKL_TRACER
+	bool "Trace Big Kernel Lock latencies"
+	depends on DEBUG_KERNEL
+	select TRACING
+	select STACKTRACE
+	help
+	  This tracer records the latencies issued by the Big Kernel Lock.
+	  It traces the time when a task requests the bkl, the time when it
+	  acquires it and the time when it releases it.
+
 config DYNAMIC_FTRACE
 	bool "enable/disable ftrace tracepoints dynamically"
 	depends on FTRACE

kernel/trace/Makefile +1 -0

index a85dfba..2ba8904 100644
@@ -24,5 +24,6 @@ obj-$(CONFIG_NOP_TRACER) += trace_nop.o
 obj-$(CONFIG_STACK_TRACER) += trace_stack.o
 obj-$(CONFIG_MMIOTRACE) += trace_mmiotrace.o
 obj-$(CONFIG_BOOT_TRACER) += trace_boot.o
+obj-$(CONFIG_BKL_TRACER) += trace_bkl.o
 
 libftrace-y := ftrace.o

kernel/trace/trace.h +18 -0

index f1f9957..8660336 100644
@@ -8,6 +8,7 @@
 #include <linux/ring_buffer.h>
 #include <linux/mmiotrace.h>
 #include <linux/ftrace.h>
+#include <trace/bkl.h>
 
 enum trace_type {
 	__TRACE_FIRST_TYPE = 0,

@@ -22,6 +23,8 @@ enum trace_type {
 	TRACE_MMIO_RW,
 	TRACE_MMIO_MAP,
 	TRACE_BOOT,
+	TRACE_BKL_ACQ,
+	TRACE_BKL_REL,
 
 	__TRACE_LAST_TYPE
 };

@@ -117,6 +120,17 @@ struct trace_boot {
 	struct boot_trace	initcall;
 };
 
+/* Trace bkl entries */
+struct trace_bkl_acq {
+	struct trace_entry		ent;
+	struct bkl_trace_acquire	bkl_acq;
+};
+
+struct trace_bkl_rel {
+	struct trace_entry		ent;
+	struct bkl_trace_release	bkl_rel;
+};
+
 /*
  * trace_flag_type is an enumeration that holds different
  * states when a trace occurs. These are:

@@ -217,6 +231,10 @@ extern void __ftrace_bad_type(void);
 		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 trace_bkl_acq,               \
+			  TRACE_BKL_ACQ);				\
+		IF_ASSIGN(var, ent, struct trace_bkl_rel,		\
+			  TRACE_BKL_REL);				\
 		__ftrace_bad_type();					\
 	} while (0)
 

lib/kernel_lock.c +34 -2

index 01a3c22..45828b2 100644
@@ -9,6 +9,7 @@
 #include <linux/module.h>
 #include <linux/kallsyms.h>
 #include <linux/semaphore.h>
+#include <trace/bkl.h>
 
 /*
  * The 'big kernel lock'

@@ -107,6 +108,37 @@ static inline void __unlock_kernel(void)
 	preempt_enable();
 }
 
+#ifdef CONFIG_BKL_TRACER
+static void lock_kernel_trace(void)
+{
+	int cpu;
+	struct bkl_trace_acquire trace;
+
+	preempt_disable();
+	cpu = raw_smp_processor_id();
+	preempt_enable();
+
+	trace.acquire_req_time = cpu_clock(cpu);
+	__lock_kernel();
+	trace.acquire_time = cpu_clock(cpu);
+	trace_bkl_acquire(&trace);
+}
+
+static void unlock_kernel_trace(void)
+{
+	struct bkl_trace_release trace;
+	trace.release_time = cpu_clock(raw_smp_processor_id());
+	trace_bkl_release(&trace);
+	__unlock_kernel();
+}
+
+#else
+
+#define lock_kernel_trace() __lock_kernel()
+#define unlock_kernel_trace() __unlock_kernel()
+
+#endif
+
 /*
  * Getting the big kernel lock.
  *

@@ -117,7 +149,7 @@ void __lockfunc lock_kernel(void)
 {
 	int depth = current->lock_depth+1;
 	if (likely(!depth))
-		__lock_kernel();
+		lock_kernel_trace();
 	current->lock_depth = depth;
 }
 

@@ -125,7 +157,7 @@ void __lockfunc unlock_kernel(void)
 {
 	BUG_ON(current->lock_depth < 0);
 	if (likely(--current->lock_depth < 0))
-		__unlock_kernel();
+		unlock_kernel_trace();
 }
 
 EXPORT_SYMBOL(lock_kernel);

Introduce the Big Kernel Lock tracer.
This new tracer lets one to observe the latencies caused by the lock_kernel() function.
It captures the time when the task request the spinlock, the time when the spinlock is hold and the time it is released. This way we can measure the latency for a task that wait for this spinlock and the latency caused by the time this lock is hold.

I will surely add some improvements like:

_ Adding a stack trace to know which function locked or unlocked the kernel. _ Adding a script to get some statistics about this: which functions hold it often, which hold it for a long time and some other things....

Changes in this V2:

_ Correction of comments
_ Make static inline the insertion functions when the tracer is not configured _ Remove FTRACE and HAVE_FTRACE as dependencies. Bkl Tracer only depends on TRACING and
STACKTRACE for further features.
_ Use trace_assign_type instead of unsure cast
_ Several cleanups...

Signed-off-by Frederic Weisbecker <[email protected]>

Cc Steven Rostedt <[email protected]>

---
 kernel/trace/Kconfig  |   10 ++++++++++
 kernel/trace/Makefile |    1 +
 kernel/trace/trace.h  |   18 ++++++++++++++++++
 lib/kernel_lock.c     |   36 ++++++++++++++++++++++++++++++++++--
 4 files changed, 63 insertions(+), 2 deletions(-)

kernel/trace/Kconfig +10 -0

index 396aea1..a87484d 100644
@@ -144,6 +144,16 @@ config STACK_TRACER
 	  This tracer records the max stack of the kernel, and displays
 	  it in debugfs/tracing/stack_trace
 
+config BKL_TRACER
+	bool "Trace Big Kernel Lock latencies"
+	depends on DEBUG_KERNEL
+	select TRACING
+	select STACKTRACE
+	help
+	  This tracer records the latencies issued by the Big Kernel Lock.
+	  It traces the time when a task requests the bkl, the time when it
+	  acquires it and the time when it releases it.
+
 config DYNAMIC_FTRACE
 	bool "enable/disable ftrace tracepoints dynamically"
 	depends on FTRACE

kernel/trace/Makefile +1 -0

index a85dfba..2ba8904 100644
@@ -24,5 +24,6 @@ obj-$(CONFIG_NOP_TRACER) += trace_nop.o
 obj-$(CONFIG_STACK_TRACER) += trace_stack.o
 obj-$(CONFIG_MMIOTRACE) += trace_mmiotrace.o
 obj-$(CONFIG_BOOT_TRACER) += trace_boot.o
+obj-$(CONFIG_BKL_TRACER) += trace_bkl.o
 
 libftrace-y := ftrace.o

kernel/trace/trace.h +18 -0

index f1f9957..8660336 100644
@@ -8,6 +8,7 @@
 #include <linux/ring_buffer.h>
 #include <linux/mmiotrace.h>
 #include <linux/ftrace.h>
+#include <trace/bkl.h>
 
 enum trace_type {
 	__TRACE_FIRST_TYPE = 0,

@@ -22,6 +23,8 @@ enum trace_type {
 	TRACE_MMIO_RW,
 	TRACE_MMIO_MAP,
 	TRACE_BOOT,
+	TRACE_BKL_ACQ,
+	TRACE_BKL_REL,
 
 	__TRACE_LAST_TYPE
 };

@@ -117,6 +120,17 @@ struct trace_boot {
 	struct boot_trace	initcall;
 };
 
+/* Trace bkl entries */
+struct trace_bkl_acq {
+	struct trace_entry		ent;
+	struct bkl_trace_acquire	bkl_acq;
+};
+
+struct trace_bkl_rel {
+	struct trace_entry		ent;
+	struct bkl_trace_release	bkl_rel;
+};
+
 /*
  * trace_flag_type is an enumeration that holds different
  * states when a trace occurs. These are:

@@ -217,6 +231,10 @@ extern void __ftrace_bad_type(void);
 		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 trace_bkl_acq,               \
+			  TRACE_BKL_ACQ);				\
+		IF_ASSIGN(var, ent, struct trace_bkl_rel,		\
+			  TRACE_BKL_REL);				\
 		__ftrace_bad_type();					\
 	} while (0)
 

lib/kernel_lock.c +33 -1

index 01a3c22..45828b2 100644
@@ -9,6 +9,7 @@
 #include <linux/module.h>
 #include <linux/kallsyms.h>
 #include <linux/semaphore.h>
+#include <trace/bkl.h>
 
 /*
  * The 'big kernel lock'

@@ -107,6 +108,37 @@ static inline void __unlock_kernel(void)
 	preempt_enable();
 }
 
+#ifdef CONFIG_BKL_TRACER
+static void lock_kernel_trace(void)
+{
+	int cpu;
+	struct bkl_trace_acquire trace;
+
+	preempt_disable();
+	cpu = raw_smp_processor_id();
+	preempt_enable();
+
+	trace.acquire_req_time = cpu_clock(cpu);
+	__lock_kernel();
+	trace.acquire_time = cpu_clock(cpu);
+	trace_bkl_acquire(&trace);
+}
+
+static void unlock_kernel_trace(void)
+{
+	struct bkl_trace_release trace;
+	trace.release_time = cpu_clock(raw_smp_processor_id());
+	trace_bkl_release(&trace);
+	__unlock_kernel();
+}
+
+#else
+
+#define lock_kernel_trace() __lock_kernel()
+#define unlock_kernel_trace() __unlock_kernel()
+
+#endif
+
 /*
  * Getting the big kernel lock.
  *

@@ -117,7 +149,7 @@ void __lockfunc lock_kernel(void)
 {
 	int depth = current->lock_depth+1;
 	if (likely(!depth))
-		__lock_kernel();
+		lock_kernel_trace();
 	current->lock_depth = depth;
 }
 

lib/kernel_lock.c +21 -0

index 01a3c22..45828b2 100644
@@ -9,6 +9,7 @@
 #include <linux/module.h>
 #include <linux/kallsyms.h>
 #include <linux/semaphore.h>
+#include <trace/bkl.h>
 
 /*
  * The 'big kernel lock'

@@ -107,6 +108,37 @@ static inline void __unlock_kernel(void)
 	preempt_enable();
 }
 
+#ifdef CONFIG_BKL_TRACER
+static void lock_kernel_trace(void)
+{
+	int cpu;
+	struct bkl_trace_acquire trace;
+
+	preempt_disable();
+	cpu = raw_smp_processor_id();
+	preempt_enable();
+
+	trace.acquire_req_time = cpu_clock(cpu);
+	__lock_kernel();
+	trace.acquire_time = cpu_clock(cpu);
+	trace_bkl_acquire(&trace);
+}
+
+static void unlock_kernel_trace(void)
+{
+	struct bkl_trace_release trace;
+	trace.release_time = cpu_clock(raw_smp_processor_id());

Quoted 37 lines

lib/kernel_lock.c +21 -0

index 01a3c22..45828b2 100644
@@ -9,6 +9,7 @@
 #include <linux/module.h>
 #include <linux/kallsyms.h>
 #include <linux/semaphore.h>
+#include <trace/bkl.h>
 
 /*
  * The 'big kernel lock'

@@ -107,6 +108,37 @@ static inline void __unlock_kernel(void)
      preempt_enable();
 }
 
+#ifdef CONFIG_BKL_TRACER
+static void lock_kernel_trace(void)
+{
+     int cpu;
+     struct bkl_trace_acquire trace;
+
+     preempt_disable();
+     cpu = raw_smp_processor_id();
+     preempt_enable();
+
+     trace.acquire_req_time = cpu_clock(cpu);
+     __lock_kernel();
+     trace.acquire_time = cpu_clock(cpu);
+     trace_bkl_acquire(&trace);
+}
+
+static void unlock_kernel_trace(void)
+{
+     struct bkl_trace_release trace;
+     trace.release_time = cpu_clock(raw_smp_processor_id());

lib/kernel_lock.c static inline void __unlock_kernel(void) · line 140

Quoted 10 lines

lib/kernel_lock.c +10 -0

+     trace_bkl_release(&trace);
+     __unlock_kernel();
+}
+
+#else
+
+#define lock_kernel_trace() __lock_kernel()
+#define unlock_kernel_trace() __unlock_kernel()
+
+#endif

hm, this looks a bit ugly.

are you aware of the tip/kill-the-BKL branch? It's an old-ish but otherwise sane branch that needs some refreshing (hence it's not part of tip/master).

Once we have that "kill the BKL by turning it into a mutex" feature alive, and have fixed the places that rely on odd properties of the BKL, the BKL becomes just an ordinary mutex and we could trace its latencies via the existing lockdep/lockstat callbacks.

and we could trace all the other mutexes as well.

2008/10/27 Ingo Molnar <[email protected]>:

Quoted 56 lines · Frederic Weisbecker

Quoted 37 lines

lib/kernel_lock.c +21 -0

index 01a3c22..45828b2 100644
@@ -9,6 +9,7 @@
 #include <linux/module.h>
 #include <linux/kallsyms.h>
 #include <linux/semaphore.h>
+#include <trace/bkl.h>
 
 /*
  * The 'big kernel lock'

@@ -107,6 +108,37 @@ static inline void __unlock_kernel(void)
      preempt_enable();
 }
 
+#ifdef CONFIG_BKL_TRACER
+static void lock_kernel_trace(void)
+{
+     int cpu;
+     struct bkl_trace_acquire trace;
+
+     preempt_disable();
+     cpu = raw_smp_processor_id();
+     preempt_enable();
+
+     trace.acquire_req_time = cpu_clock(cpu);
+     __lock_kernel();
+     trace.acquire_time = cpu_clock(cpu);
+     trace_bkl_acquire(&trace);
+}
+
+static void unlock_kernel_trace(void)
+{
+     struct bkl_trace_release trace;
+     trace.release_time = cpu_clock(raw_smp_processor_id());

lib/kernel_lock.c static inline void __unlock_kernel(void) · line 140

Quoted 10 lines

lib/kernel_lock.c +10 -0

+     trace_bkl_release(&trace);
+     __unlock_kernel();
+}
+
+#else
+
+#define lock_kernel_trace() __lock_kernel()
+#define unlock_kernel_trace() __unlock_kernel()
+
+#endif

hm, this looks a bit ugly.

are you aware of the tip/kill-the-BKL branch? It's an old-ish but otherwise sane branch that needs some refreshing (hence it's not part of tip/master).

Quoted 6 lines

Once we have that "kill the BKL by turning it into a mutex" feature alive, and have fixed the places that rely on odd properties of the BKL, the BKL becomes just an ordinary mutex and we could trace its latencies via the existing lockdep/lockstat callbacks.

and we could trace all the other mutexes as well.

No problem, we can forget about it. My goal was to produce some statistics to locate the points that most often hold the bkl. That would help to define some priorities on which bkl holding is to remove first.

But if that would be better to rather invest the time on the kill-the-bkl tree (which I thought was dead), so I would be pleased to help.