Skip to contentSkip to conversation

tracing: Dump either the oops's cpu source or all cpus buffers

PATCH Superseded Frederic Weisbecker 4 participants 5 messages Apr 18, 2010 – Apr 19, 2010

Subsystems (4)
lore
Share & tools
lore
#0 lore raw

PATCH 1/1 tracing: Dump either the oops's cpu source or all cpus buffers

The ftrace_dump_on_oops kernel parameter, sysctl and sysrq let one dump every cpu buffers when an oops or panic happens.

It's nice when you have few cpus but it may take ages if have many, plus you miss the real origin of the problem in all the cpu traces.

Sometimes, all you need is to dump the cpu buffer that triggered the opps, most of the time it is our main interest.

This patch modifies ftrace_dump_on_oops to handle this choice.

The ftrace_dump_on_oops kernel parameter, when it comes alone, has the same behaviour than before. But ftrace_dump_on_oops=orig_cpu will only dump the buffer of the cpu that oops'ed.

Similarly, sysctl kernel.ftrace_dump_on_oops=1 and
echo 1 > /proc/sys/kernel/ftrace_dump_on_oops keep their previous behaviour. But setting 2 jumps into cpu origin dump mode.

Cc Ingo Molnar <[email protected]>
Cc Thomas Gleixner <[email protected]>
Cc Li Zefan <[email protected]>
Cc Lai Jiangshan <[email protected]>
---
 Documentation/kernel-parameters.txt |    6 +++-
 Documentation/trace/ftrace.txt      |    6 ++-
 drivers/char/sysrq.c                |    2 +-
 include/linux/ftrace.h              |    4 ++-
 include/linux/kernel.h              |   11 ++++++-
 kernel/trace/trace.c                |   53 +++++++++++++++++++++++++++--------
 6 files changed, 63 insertions(+), 19 deletions(-)
Documentation/kernel-parameters.txt +5 -1
index e4cbca5..f6c68e8 100644
@@ -789,8 +789,12 @@ and is between 256 and 4096 characters. It is defined in the file
 			as early as possible in order to facilitate early
 			boot debugging.
 
-	ftrace_dump_on_oops
+	ftrace_dump_on_oops[=orig_cpu]
 			[FTRACE] will dump the trace buffers on oops.
+			If no parameter is passed, ftrace will dump
+			every cpu buffers, but if you pass orig_cpu, it will
+			dump only the buffer of the cpu that triggered the
+			oops.
 
 	ftrace_filter=[function-list]
 			[FTRACE] Limit the functions traced by the function
Documentation/trace/ftrace.txt +4 -2
index 03485bf..d242bc0 100644
@@ -1337,12 +1337,14 @@ ftrace_dump_on_oops must be set. To set ftrace_dump_on_oops, one
 can either use the sysctl function or set it via the proc system
 interface.
 
-  sysctl kernel.ftrace_dump_on_oops=1
+  sysctl kernel.ftrace_dump_on_oops=n
 
 or
 
-  echo 1 > /proc/sys/kernel/ftrace_dump_on_oops
+  echo n > /proc/sys/kernel/ftrace_dump_on_oops
 
+If n = 1, ftrace will dump every cpu buffers, if n = 2 ftrace will
+only dump the buffer of the cpu that triggered the oops.
 
 Here's an example of such a dump after a null pointer
 dereference in a kernel module:
drivers/char/sysrq.c +1 -1
index 59de252..d4e8b21 100644
@@ -289,7 +289,7 @@ static struct sysrq_key_op sysrq_showstate_blocked_op = {
 
 static void sysrq_ftrace_dump(int key, struct tty_struct *tty)
 {
-	ftrace_dump();
+	ftrace_dump(DUMP_ALL);
 }
 static struct sysrq_key_op sysrq_ftrace_dump_op = {
 	.handler	= sysrq_ftrace_dump,
include/linux/ftrace.h +3 -1
index 01e6ade..ea5b1aa 100644
@@ -492,7 +492,9 @@ static inline int test_tsk_trace_graph(struct task_struct *tsk)
 	return tsk->trace & TSK_TRACE_FL_GRAPH;
 }
 
-extern int ftrace_dump_on_oops;
+enum ftrace_dump_mode;
+
+extern enum ftrace_dump_mode ftrace_dump_on_oops;
 
 #ifdef CONFIG_PREEMPT
 #define INIT_TRACE_RECURSION		.trace_recursion = 0,
include/linux/kernel.h +9 -2
index 9365227..9fb1c12 100644
@@ -490,6 +490,13 @@ static inline void tracing_off(void) { }
 static inline void tracing_off_permanent(void) { }
 static inline int tracing_is_on(void) { return 0; }
 #endif
+
+enum ftrace_dump_mode {
+	DUMP_NONE,
+	DUMP_ALL,
+	DUMP_ORIG,
+};
+
 #ifdef CONFIG_TRACING
 extern void tracing_start(void);
 extern void tracing_stop(void);
@@ -571,7 +578,7 @@ __ftrace_vbprintk(unsigned long ip, const char *fmt, va_list ap);
 extern int
 __ftrace_vprintk(unsigned long ip, const char *fmt, va_list ap);
 
-extern void ftrace_dump(void);
+extern void ftrace_dump(enum ftrace_dump_mode oops_dump_mode);
 #else
 static inline void
 ftrace_special(unsigned long arg1, unsigned long arg2, unsigned long arg3) { }
@@ -592,7 +599,7 @@ ftrace_vprintk(const char *fmt, va_list ap)
 {
 	return 0;
 }
-static inline void ftrace_dump(void) { }
+static inline void ftrace_dump(enum ftrace_dump_mode oops_dump_mode) { }
 #endif /* CONFIG_TRACING */
 
 /*
kernel/trace/trace.c +41 -12

108 lines not on this page · raw message

Signature · 1 line
1.6.2.3
#1 lore raw

PATCH v2 1/1 tracing: Dump either the oops's cpu source or all cpus buffers

The ftrace_dump_on_oops kernel parameter, sysctl and sysrq let one dump every cpu buffers when an oops or panic happens.

It's nice when you have few cpus but it may take ages if you have many, plus you miss the real origin of the problem in all the cpu traces.

Sometimes, all you need is to dump the cpu buffer that triggered the oops, most of the time it is our main interest.

This patch modifies ftrace_dump_on_oops to handle this choice.

The ftrace_dump_on_oops kernel parameter, when it comes alone, has the same behaviour than before. But ftrace_dump_on_oops=orig_cpu will only dump the buffer of the cpu that oops'ed.

Similarly, sysctl kernel.ftrace_dump_on_oops=1 and
echo 1 > /proc/sys/kernel/ftrace_dump_on_oops keep their previous behaviour. But setting 2 jumps into cpu origin dump mode.

v2: Fix double setup

Cc Ingo Molnar <[email protected]>
Cc Thomas Gleixner <[email protected]>
Cc Li Zefan <[email protected]>
Cc Lai Jiangshan <[email protected]>
---
 Documentation/kernel-parameters.txt |    6 +++-
 Documentation/trace/ftrace.txt      |    6 +++-
 drivers/char/sysrq.c                |    2 +-
 include/linux/ftrace.h              |    4 ++-
 include/linux/kernel.h              |   11 ++++++-
 kernel/trace/trace.c                |   51 ++++++++++++++++++++++++++--------
 6 files changed, 61 insertions(+), 19 deletions(-)
Documentation/kernel-parameters.txt +5 -1
index e4cbca5..f6c68e8 100644
@@ -789,8 +789,12 @@ and is between 256 and 4096 characters. It is defined in the file
 			as early as possible in order to facilitate early
 			boot debugging.
 
-	ftrace_dump_on_oops
+	ftrace_dump_on_oops[=orig_cpu]
 			[FTRACE] will dump the trace buffers on oops.
+			If no parameter is passed, ftrace will dump
+			every cpu buffers, but if you pass orig_cpu, it will
+			dump only the buffer of the cpu that triggered the
+			oops.
 
 	ftrace_filter=[function-list]
 			[FTRACE] Limit the functions traced by the function
Documentation/trace/ftrace.txt +4 -2
index 03485bf..d242bc0 100644
@@ -1337,12 +1337,14 @@ ftrace_dump_on_oops must be set. To set ftrace_dump_on_oops, one
 can either use the sysctl function or set it via the proc system
 interface.
 
-  sysctl kernel.ftrace_dump_on_oops=1
+  sysctl kernel.ftrace_dump_on_oops=n
 
 or
 
-  echo 1 > /proc/sys/kernel/ftrace_dump_on_oops
+  echo n > /proc/sys/kernel/ftrace_dump_on_oops
 
+If n = 1, ftrace will dump every cpu buffers, if n = 2 ftrace will
+only dump the buffer of the cpu that triggered the oops.
 
 Here's an example of such a dump after a null pointer
 dereference in a kernel module:
drivers/char/sysrq.c +1 -1
index 59de252..d4e8b21 100644
@@ -289,7 +289,7 @@ static struct sysrq_key_op sysrq_showstate_blocked_op = {
 
 static void sysrq_ftrace_dump(int key, struct tty_struct *tty)
 {
-	ftrace_dump();
+	ftrace_dump(DUMP_ALL);
 }
 static struct sysrq_key_op sysrq_ftrace_dump_op = {
 	.handler	= sysrq_ftrace_dump,
include/linux/ftrace.h +3 -1
index 01e6ade..ea5b1aa 100644
@@ -492,7 +492,9 @@ static inline int test_tsk_trace_graph(struct task_struct *tsk)
 	return tsk->trace & TSK_TRACE_FL_GRAPH;
 }
 
-extern int ftrace_dump_on_oops;
+enum ftrace_dump_mode;
+
+extern enum ftrace_dump_mode ftrace_dump_on_oops;
 
 #ifdef CONFIG_PREEMPT
 #define INIT_TRACE_RECURSION		.trace_recursion = 0,
include/linux/kernel.h +9 -2
index 9365227..9fb1c12 100644
@@ -490,6 +490,13 @@ static inline void tracing_off(void) { }
 static inline void tracing_off_permanent(void) { }
 static inline int tracing_is_on(void) { return 0; }
 #endif
+
+enum ftrace_dump_mode {
+	DUMP_NONE,
+	DUMP_ALL,
+	DUMP_ORIG,
+};
+
 #ifdef CONFIG_TRACING
 extern void tracing_start(void);
 extern void tracing_stop(void);
@@ -571,7 +578,7 @@ __ftrace_vbprintk(unsigned long ip, const char *fmt, va_list ap);
 extern int
 __ftrace_vprintk(unsigned long ip, const char *fmt, va_list ap);
 
-extern void ftrace_dump(void);
+extern void ftrace_dump(enum ftrace_dump_mode oops_dump_mode);
 #else
 static inline void
 ftrace_special(unsigned long arg1, unsigned long arg2, unsigned long arg3) { }
@@ -592,7 +599,7 @@ ftrace_vprintk(const char *fmt, va_list ap)
 {
 	return 0;
 }
-static inline void ftrace_dump(void) { }
+static inline void ftrace_dump(enum ftrace_dump_mode oops_dump_mode) { }
 #endif /* CONFIG_TRACING */
 
 /*
kernel/trace/trace.c +39 -12

104 lines not on this page · raw message

Signature · 1 line
1.6.2.3
#2 lore raw
Quoted 13 lines · On Mon, 19 Apr 2010 03:15:40 +0200 Frederic Weisbecker wrote:
 Documentation/kernel-parameters.txt |    6 +++-
 Documentation/trace/ftrace.txt      |    6 +++-
 drivers/char/sysrq.c                |    2 +-
 include/linux/ftrace.h              |    4 ++-
 include/linux/kernel.h              |   11 ++++++-
 kernel/trace/trace.c                |   51 ++++++++++++++++++++++++++--------
 6 files changed, 61 insertions(+), 19 deletions(-)
Documentation/kernel-parameters.txt +0 -0
index e4cbca5..f6c68e8 100644

Documentation/kernel-parameters.txt and is between 256 and 4096 characters. It is defined in the file · line 795

Quoted 8 lines
Documentation/kernel-parameters.txt +3 -1
@@ -789,8 +789,12 @@ and is between 256 and 4096 characters. It is defined in the file
 			as early as possible in order to facilitate early
 			boot debugging.
 
-	ftrace_dump_on_oops
+	ftrace_dump_on_oops[=orig_cpu]
 			[FTRACE] will dump the trace buffers on oops.
+			If no parameter is passed, ftrace will dump
+			every cpu buffers, but if you pass orig_cpu, it will

buffers of all CPUs,

Documentation/kernel-parameters.txt and is between 256 and 4096 characters. It is defined in the file · line 796

Quoted 1 line
Documentation/kernel-parameters.txt +1 -0
+			dump only the buffer of the cpu that triggered the

CPU

Quoted 11 lines
Documentation/kernel-parameters.txt +1 -0
+			oops.
 
 	ftrace_filter=[function-list]
 			[FTRACE] Limit the functions traced by the function
Documentation/trace/ftrace.txt +0 -0
index 03485bf..d242bc0 100644
@@ -1337,12 +1337,14 @@ ftrace_dump_on_oops must be set. To set ftrace_dump_on_oops, one
 can either use the sysctl function or set it via the proc system
 interface.

Documentation/trace/ftrace.txt ftrace_dump_on_oops must be set. To set ftrace_dump_on_oops, one · line 1346

Quoted 10 lines
Documentation/trace/ftrace.txt +3 -2
 
-  sysctl kernel.ftrace_dump_on_oops=1
+  sysctl kernel.ftrace_dump_on_oops=n
 
 or
 
-  echo 1 > /proc/sys/kernel/ftrace_dump_on_oops
+  echo n > /proc/sys/kernel/ftrace_dump_on_oops
 
+If n = 1, ftrace will dump every cpu buffers, if n = 2 ftrace will

buffers of all CPUs,

Documentation/trace/ftrace.txt ftrace_dump_on_oops must be set. To set ftrace_dump_on_oops, one · line 1347

Quoted 1 line
Documentation/trace/ftrace.txt +1 -0
+only dump the buffer of the cpu that triggered the oops.

CPU

Quoted 8 lines
Documentation/trace/ftrace.txt +0 -0
 Here's an example of such a dump after a null pointer
 dereference in a kernel module:
kernel/trace/trace.c +0 -0
index bed83ca..4252ffe 100644

kernel/trace/trace.c static cpumask_var_t __read_mostly tracing_buffer_mask; · line 121

Quoted 6 lines
kernel/trace/trace.c +2 -1
@@ -117,9 +117,12 @@ static cpumask_var_t __read_mostly	tracing_buffer_mask;
  *
  * It is default off, but you can enable it with either specifying
  * "ftrace_dump_on_oops" in the kernel command line, or setting
- * /proc/sys/kernel/ftrace_dump_on_oops to true.
+ * /proc/sys/kernel/ftrace_dump_on_oops
+ * Set 1 if you want to dump every cpu buffers

buffers of all CPUs

kernel/trace/trace.c static cpumask_var_t __read_mostly tracing_buffer_mask; · line 122

Quoted 1 line
kernel/trace/trace.c +1 -0
+ * Set 2 if you want to dump the buffer of the cpu that triggered oops

CPU

Quoted 19 lines
kernel/trace/trace.c +4 -4
  */
-int ftrace_dump_on_oops;
+
+enum ftrace_dump_mode ftrace_dump_on_oops;
 
 static int tracing_set_tracer(const char *buf);
 
@@ -4429,12 +4442,25 @@ static void __ftrace_dump(bool disable_tracing)
 	/* don't look at user memory in panic mode */
 	trace_flags &= ~TRACE_ITER_SYM_USEROBJ;
 
-	printk(KERN_TRACE "Dumping ftrace buffer:\n");
-
 	/* Simulate the iterator */
 	iter.tr = &global_trace;
 	iter.trace = current_trace;
-	iter.cpu_file = TRACE_PIPE_ALL_CPU;
+
+	switch (oops_dump_mode) {

kernel/trace/trace.c static void __ftrace_dump(bool disable_tracing) · line 4459

Quoted 10 lines
kernel/trace/trace.c +10 -0
+	case DUMP_ALL:
+		iter.cpu_file = TRACE_PIPE_ALL_CPU;
+		break;
+	case DUMP_ORIG:
+		iter.cpu_file = raw_smp_processor_id();
+		break;
+	case DUMP_NONE:
+		goto out_enable;
+	default:
+		printk(KERN_TRACE "Bad dumping mode, switching to all cpu dump\n");

CPUs

kernel/trace/trace.c static void __ftrace_dump(bool disable_tracing) · line 4466

Quoted 7 lines
kernel/trace/trace.c +4 -0
+		iter.cpu_file = TRACE_PIPE_ALL_CPU;
+	}
+
+	printk(KERN_TRACE "Dumping ftrace buffer:\n");
 
 	/*
 	 * We need to stop all tracing on all CPUS to read the

---
~Randy

maintainer Acked-by
#3 lore raw
Quoted 9 lines · Randy Dunlap
Quoted 7 lines · On Mon, 19 Apr 2010 03:15:40 +0200 Frederic Weisbecker wrote:
 Documentation/kernel-parameters.txt |    6 +++-
 Documentation/trace/ftrace.txt      |    6 +++-
 drivers/char/sysrq.c                |    2 +-
 include/linux/ftrace.h              |    4 ++-
 include/linux/kernel.h              |   11 ++++++-
 kernel/trace/trace.c                |   51 ++++++++++++++++++++++++++--------
 6 files changed, 61 insertions(+), 19 deletions(-)

Frederic,

Send out v3 with Randy's updates and:

-- Steve

Quoted 10 lines
Quoted 10 lines
Documentation/kernel-parameters.txt +0 -0
index e4cbca5..f6c68e8 100644
@@ -789,8 +789,12 @@ and is between 256 and 4096 characters. It is defined in the file
 			as early as possible in order to facilitate early
 			boot debugging.
 
 
Acked-by on v1 1/1
#4 lore raw
In reply to Frederic Weisbecker

From: Frederic Weisbecker <[email protected]>
Date: Sun, 18 Apr 2010 22:50:10 +0200

Quoted 15 lines

The ftrace_dump_on_oops kernel parameter, sysctl and sysrq let one dump every cpu buffers when an oops or panic happens.

It's nice when you have few cpus but it may take ages if have many, plus you miss the real origin of the problem in all the cpu traces.

Sometimes, all you need is to dump the cpu buffer that triggered the opps, most of the time it is our main interest.

This patch modifies ftrace_dump_on_oops to handle this choice.

The ftrace_dump_on_oops kernel parameter, when it comes alone, has the same behaviour than before. But ftrace_dump_on_oops=orig_cpu will only dump the buffer of the cpu that oops'ed.

Quoted 5 lines

Similarly, sysctl kernel.ftrace_dump_on_oops=1 and
echo 1 > /proc/sys/kernel/ftrace_dump_on_oops keep their previous behaviour. But setting 2 jumps into cpu origin dump mode.

Thanks I was trying to figure out a way to make ftrace dump do exactly this over the past few days :-)