Home » Mailing list archives » linux-kernel » 2010 » June » 29

[PATCH 4/9] tracing: Convert more sched events to DEFINE_EVENT

Previous thread: [PATCH 1/9] tracing: Use a global field list for all syscall exit events by Steven Rostedt on Monday, June 28, 2010 - 8:08 pm. (1 message)

Next thread: linux-next: build failure after merge of the spi tree by Stephen Rothwell on Monday, June 28, 2010 - 8:29 pm. (2 messages)
[view original message]
From: Steven Rostedt
Subject: [PATCH 0/9] [GI PULL][2.6.36] tracing: various updates
Date: Monday, June 28, 2010 - 8:08 pm

Ingo,

I pushed some of these to you before, but you wanted to wait
till after the merge window to pull them.

The last patch is the full fix for the perf kprobe regression. But
since it was too big of a change for the -rc release we did
a quick fix. This reverts that change back to the true fix for
2.6.36.

Please pull the latest tip/perf/core tree, which can be found at:

  git://git.kernel.org/pub/scm/linux/kernel/git/rostedt/linux-2.6-trace.git
tip/perf/core


Chase Douglas (1):
      tracing/function-graph: Use correct string size for snprintf

Li Zefan (7):
      tracing: Use a global field list for all syscall exit events
      tracing: Don't allocate common fields for every trace events
      tracing: Convert some timer events to DEFINE_EVENT
      tracing: Convert more sched events to DEFINE_EVENT
      tracing: Remove test of NULL define_fields callback
      tracing: Remove redundant raw_init callbacks
      tracing: Remove open-coded __trace_add_event_call()

Steven Rostedt (1):
      tracing: Use class->reg() for all registering of events

----
 include/linux/ftrace_event.h         |    3 +
 include/linux/syscalls.h             |    2 -
 include/trace/events/sched.h         |   32 +----
 include/trace/events/timer.h         |   80 ++++------
 include/trace/ftrace.h               |    2 +
 include/trace/syscall.h              |    1 -
 kernel/trace/trace.h                 |    2 +
 kernel/trace/trace_event_perf.c      |   19 +--
 kernel/trace/trace_events.c          |  264 +++++++++++++++++-----------------
 kernel/trace/trace_events_filter.c   |   27 ++--
 kernel/trace/trace_export.c          |    8 +-
 kernel/trace/trace_functions_graph.c |    3 +-
 kernel/trace/trace_kprobe.c          |   10 +--
 kernel/trace/trace_syscalls.c        |    7 +-
 14 files changed, 203 insertions(+), 257 deletions(-)
--


[ message continues ]
[view original message]
From: Steven Rostedt
Subject: [PATCH 4/9] tracing: Convert more sched events to DEFINE_EVENT
Date: Monday, June 28, 2010 - 8:08 pm

From: Li Zefan <lizf@cn.fujitsu.com>

Convert sched_wait_task to DEFINE_EVENT, and save ~1K:

   text    data     bss     dec     hex filename
 104595    9424    4992  119011   1d0e3 kernel/sched.o.orig
 103619    9344    4992  117955   1ccc3 kernel/sched.o

No change in functionality.

Signed-off-by: Li Zefan <lizf@cn.fujitsu.com>
LKML-Reference: <4BFA3787.2040800@cn.fujitsu.com>
Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
---
 include/trace/events/sched.h |   32 +++++++-------------------------
 1 files changed, 7 insertions(+), 25 deletions(-)

diff --git a/include/trace/events/sched.h b/include/trace/events/sched.h
index b9e1dd6..9208c92 100644
--- a/include/trace/events/sched.h
+++ b/include/trace/events/sched.h
@@ -50,31 +50,6 @@ TRACE_EVENT(sched_kthread_stop_ret,
 );
 
 /*
- * Tracepoint for waiting on task to unschedule:
- */
-TRACE_EVENT(sched_wait_task,
-
-	TP_PROTO(struct task_struct *p),
-
-	TP_ARGS(p),
-
-	TP_STRUCT__entry(
-		__array(	char,	comm,	TASK_COMM_LEN	)
-		__field(	pid_t,	pid			)
-		__field(	int,	prio			)
-	),
-
-	TP_fast_assign(
-		memcpy(__entry->comm, p->comm, TASK_COMM_LEN);
-		__entry->pid	= p->pid;
-		__entry->prio	= p->prio;
-	),
-
-	TP_printk("comm=%s pid=%d prio=%d",
-		  __entry->comm, __entry->pid, __entry->prio)
-);
-
-/*
  * Tracepoint for waking up a task:
  */
 DECLARE_EVENT_CLASS(sched_wakeup_template,
@@ -240,6 +215,13 @@ DEFINE_EVENT(sched_process_template, sched_process_exit,
 	     TP_ARGS(p));
 
 /*
+ * Tracepoint for waiting on task to unschedule:
+ */
+DEFINE_EVENT(sched_process_template, sched_wait_task,
+	TP_PROTO(struct task_struct *p),
+	TP_ARGS(p));
+
+/*
  * Tracepoint for a waiting task:
  */
 TRACE_EVENT(sched_process_wait,
-- 
1.7.1


--


[ message continues ]
[view original message]
From: Steven Rostedt
Subject: [PATCH 9/9] tracing: Use class->reg() for all registering of events
Date: Monday, June 28, 2010 - 8:08 pm

From: Steven Rostedt <srostedt@redhat.com>

Because kprobes and syscalls need special processing to register
events, the class->reg() method was created to handle the differences.

But instead of creating a default ->reg for perf and ftrace events,
the code was scattered with:

	if (class->reg)
		class->reg();
	else
		default_reg();

This is messy and can also lead to bugs.

This patch cleans up this code and creates a default reg() entry for
the events allowing for the code to directly call the class->reg()
without the condition.

Reported-by: Peter Zijlstra <peterz@infradead.org>
Acked-by: Peter Zijlstra <peterz@infradead.org>
Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
---
 include/linux/ftrace_event.h    |    3 ++
 include/trace/ftrace.h          |    2 +
 kernel/trace/trace_event_perf.c |   19 ++-----------
 kernel/trace/trace_events.c     |   55 +++++++++++++++++++++++++-------------
 4 files changed, 44 insertions(+), 35 deletions(-)

diff --git a/include/linux/ftrace_event.h b/include/linux/ftrace_event.h
index 0af31cd..01df7ca 100644
--- a/include/linux/ftrace_event.h
+++ b/include/linux/ftrace_event.h
@@ -146,6 +146,9 @@ struct ftrace_event_class {
 	int			(*raw_init)(struct ftrace_event_call *);
 };
 
+extern int ftrace_event_reg(struct ftrace_event_call *event,
+			    enum trace_reg type);
+
 enum {
 	TRACE_EVENT_FL_ENABLED_BIT,
 	TRACE_EVENT_FL_FILTERED_BIT,
diff --git a/include/trace/ftrace.h b/include/trace/ftrace.h
index fc013a8..55c1fd1 100644
--- a/include/trace/ftrace.h
+++ b/include/trace/ftrace.h
@@ -439,6 +439,7 @@ static inline notrace int ftrace_get_offsets_##call(			\
  *	.fields			= LIST_HEAD_INIT(event_class_##call.fields),
  *	.raw_init		= trace_event_raw_init,
  *	.probe			= ftrace_raw_event_##call,
+ *	.reg			= ftrace_event_reg,
  * };
  *
  * static struct ftrace_event_call __used
@@ -567,6 +568,7 @@ static struct ftrace_event_class __used event_class_##call = {		\
 	.fields			= ...
[ message continues ]
[view original message]
From: Steven Rostedt
Subject: [PATCH 7/9] tracing: Remove open-coded __trace_add_event_call()
Date: Monday, June 28, 2010 - 8:08 pm

From: Li Zefan <lizf@cn.fujitsu.com>

Let trace_module_add_events() and event_trace_init() call
__trace_add_event_call().

Signed-off-by: Li Zefan <lizf@cn.fujitsu.com>
LKML-Reference: <4BFA37E9.1020106@cn.fujitsu.com>
Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
---
 kernel/trace/trace_events.c |   70 +++++++++++-------------------------------
 1 files changed, 19 insertions(+), 51 deletions(-)

diff --git a/kernel/trace/trace_events.c b/kernel/trace/trace_events.c
index 5bad9cb..69bee4c 100644
--- a/kernel/trace/trace_events.c
+++ b/kernel/trace/trace_events.c
@@ -1009,11 +1009,17 @@ event_create_dir(struct ftrace_event_call *call, struct dentry *d_events,
 	return 0;
 }
 
-static int __trace_add_event_call(struct ftrace_event_call *call)
+static int
+__trace_add_event_call(struct ftrace_event_call *call, struct module *mod,
+		       const struct file_operations *id,
+		       const struct file_operations *enable,
+		       const struct file_operations *filter,
+		       const struct file_operations *format)
 {
 	struct dentry *d_events;
 	int ret;
 
+	/* The linker may leave blanks */
 	if (!call->name)
 		return -EINVAL;
 
@@ -1021,8 +1027,8 @@ static int __trace_add_event_call(struct ftrace_event_call *call)
 		ret = call->class->raw_init(call);
 		if (ret < 0) {
 			if (ret != -ENOSYS)
-				pr_warning("Could not initialize trace "
-				"events/%s\n", call->name);
+				pr_warning("Could not initialize trace events/%s\n",
+					   call->name);
 			return ret;
 		}
 	}
@@ -1031,11 +1037,10 @@ static int __trace_add_event_call(struct ftrace_event_call *call)
 	if (!d_events)
 		return -ENOENT;
 
-	ret = event_create_dir(call, d_events, &ftrace_event_id_fops,
-				&ftrace_enable_fops, &ftrace_event_filter_fops,
-				&ftrace_event_format_fops);
+	ret = event_create_dir(call, d_events, id, enable, filter, format);
 	if (!ret)
 		list_add(&call->list, &ftrace_events);
+	call->mod = mod;
 
 	return ret;
 }
@@ -1045,7 +1050,10 @@ int ...
[ message continues ]
[view original message]
From: Steven Rostedt
Subject: [PATCH 6/9] tracing: Remove redundant raw_init callbacks
Date: Monday, June 28, 2010 - 8:08 pm

From: Li Zefan <lizf@cn.fujitsu.com>

raw_init callback is optional.

Signed-off-by: Li Zefan <lizf@cn.fujitsu.com>
LKML-Reference: <4BFA37D4.7070500@cn.fujitsu.com>
Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
---
 kernel/trace/trace_export.c |    8 +-------
 kernel/trace/trace_kprobe.c |   10 +---------
 2 files changed, 2 insertions(+), 16 deletions(-)

diff --git a/kernel/trace/trace_export.c b/kernel/trace/trace_export.c
index 8536e2a..4ba44de 100644
--- a/kernel/trace/trace_export.c
+++ b/kernel/trace/trace_export.c
@@ -125,12 +125,6 @@ ftrace_define_fields_##name(struct ftrace_event_call *event_call)	\
 
 #include "trace_entries.h"
 
-static int ftrace_raw_init_event(struct ftrace_event_call *call)
-{
-	INIT_LIST_HEAD(&call->class->fields);
-	return 0;
-}
-
 #undef __entry
 #define __entry REC
 
@@ -158,7 +152,7 @@ static int ftrace_raw_init_event(struct ftrace_event_call *call)
 struct ftrace_event_class event_class_ftrace_##call = {			\
 	.system			= __stringify(TRACE_SYSTEM),		\
 	.define_fields		= ftrace_define_fields_##call,		\
-	.raw_init		= ftrace_raw_init_event,		\
+	.fields			= LIST_HEAD_INIT(event_class_ftrace_##call.fields),\
 };									\
 									\
 struct ftrace_event_call __used						\
diff --git a/kernel/trace/trace_kprobe.c b/kernel/trace/trace_kprobe.c
index f52b5f5..3b831d8 100644
--- a/kernel/trace/trace_kprobe.c
+++ b/kernel/trace/trace_kprobe.c
@@ -1214,11 +1214,6 @@ static void probe_event_disable(struct ftrace_event_call *call)
 	}
 }
 
-static int probe_event_raw_init(struct ftrace_event_call *event_call)
-{
-	return 0;
-}
-
 #undef DEFINE_FIELD
 #define DEFINE_FIELD(type, item, name, is_signed)			\
 	do {								\
@@ -1486,15 +1481,12 @@ static int register_probe_event(struct trace_probe *tp)
 	int ret;
 
 	/* Initialize ftrace_event_call */
+	INIT_LIST_HEAD(&call->class->fields);
 	if (probe_is_return(tp)) {
-		INIT_LIST_HEAD(&call->class->fields);
 		call->event.funcs = ...
[ message continues ]
[view original message]
From: Steven Rostedt
Subject: [PATCH 8/9] tracing/function-graph: Use correct string size for snprintf
Date: Monday, June 28, 2010 - 8:08 pm

From: Chase Douglas <chase.douglas@canonical.com>

The nsecs_str string is a local variable defined as:

char nsecs_str[5];

It is possible for the snprintf call to use a size value larger than the
size of the string. This should not cause a buffer overrun as it is
written now due to the value for the string format "%03lu" can not be
larger than 1000. However, this change makes it correct. By making the
size correct we guard against potential future changes that could actually
cause a buffer overrun.

Signed-off-by: Chase Douglas <chase.douglas@canonical.com>
LKML-Reference: <1276619355-18116-1-git-send-email-chase.douglas@canonical.com>

[ added 'UL' to number 8 to fix gcc warning comparing it to sizeof() ]

Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
---
 kernel/trace/trace_functions_graph.c |    3 ++-
 1 files changed, 2 insertions(+), 1 deletions(-)

diff --git a/kernel/trace/trace_functions_graph.c b/kernel/trace/trace_functions_graph.c
index 79f4bac..6bff236 100644
--- a/kernel/trace/trace_functions_graph.c
+++ b/kernel/trace/trace_functions_graph.c
@@ -641,7 +641,8 @@ trace_print_graph_duration(unsigned long long duration, struct trace_seq *s)
 
 	/* Print nsecs (we don't want to exceed 7 numbers) */
 	if (len < 7) {
-		snprintf(nsecs_str, 8 - len, "%03lu", nsecs_rem);
+		snprintf(nsecs_str, min(sizeof(nsecs_str), 8UL - len), "%03lu",
+			 nsecs_rem);
 		ret = trace_seq_printf(s, ".%s", nsecs_str);
 		if (!ret)
 			return TRACE_TYPE_PARTIAL_LINE;
-- 
1.7.1


--


[ message continues ]
[view original message]
From: Steven Rostedt
Subject: [PATCH 2/9] tracing: Dont allocate common fields for every trace events
Date: Monday, June 28, 2010 - 8:08 pm

From: Li Zefan <lizf@cn.fujitsu.com>

Every event has the same common fields, so it's a big waste of
memory to have a copy of those fields for every event.

Signed-off-by: Li Zefan <lizf@cn.fujitsu.com>
LKML-Reference: <4BFA3759.30105@cn.fujitsu.com>
Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
---
 kernel/trace/trace.h               |    2 +
 kernel/trace/trace_events.c        |  113 ++++++++++++++++++++----------------
 kernel/trace/trace_events_filter.c |   18 +++++-
 3 files changed, 81 insertions(+), 52 deletions(-)

diff --git a/kernel/trace/trace.h b/kernel/trace/trace.h
index 01ce088..cc90ccd 100644
--- a/kernel/trace/trace.h
+++ b/kernel/trace/trace.h
@@ -698,6 +698,8 @@ struct filter_pred {
 	int 			pop_n;
 };
 
+extern struct list_head ftrace_common_fields;
+
 extern enum regex_type
 filter_parse_regex(char *buff, int len, char **search, int *not);
 extern void print_event_filter(struct ftrace_event_call *call,
diff --git a/kernel/trace/trace_events.c b/kernel/trace/trace_events.c
index a594f9a..d3b4bdf 100644
--- a/kernel/trace/trace_events.c
+++ b/kernel/trace/trace_events.c
@@ -28,6 +28,7 @@
 DEFINE_MUTEX(event_mutex);
 
 LIST_HEAD(ftrace_events);
+LIST_HEAD(ftrace_common_fields);
 
 struct list_head *
 trace_get_fields(struct ftrace_event_call *event_call)
@@ -37,15 +38,11 @@ trace_get_fields(struct ftrace_event_call *event_call)
 	return event_call->class->get_fields(event_call);
 }
 
-int trace_define_field(struct ftrace_event_call *call, const char *type,
-		       const char *name, int offset, int size, int is_signed,
-		       int filter_type)
+static int __trace_define_field(struct list_head *head, const char *type,
+				const char *name, int offset, int size,
+				int is_signed, int filter_type)
 {
 	struct ftrace_event_field *field;
-	struct list_head *head;
-
-	if (WARN_ON(!call->class))
-		return 0;
 
 	field = kzalloc(sizeof(*field), GFP_KERNEL);
 	if (!field)
@@ -68,7 +65,6 @@ int trace_define_field(struct ...
[ message continues ]
[view original message]
From: Steven Rostedt
Subject: [PATCH 5/9] tracing: Remove test of NULL define_fields callback
Date: Monday, June 28, 2010 - 8:08 pm

From: Li Zefan <lizf@cn.fujitsu.com>

Every event (or event class) has it's define_fields callback,
so the test is redundant.

Signed-off-by: Li Zefan <lizf@cn.fujitsu.com>
LKML-Reference: <4BFA37BC.8080707@cn.fujitsu.com>
Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
---
 kernel/trace/trace_events.c        |   28 +++++++++++++---------------
 kernel/trace/trace_events_filter.c |    9 ---------
 2 files changed, 13 insertions(+), 24 deletions(-)

diff --git a/kernel/trace/trace_events.c b/kernel/trace/trace_events.c
index d3b4bdf..5bad9cb 100644
--- a/kernel/trace/trace_events.c
+++ b/kernel/trace/trace_events.c
@@ -987,23 +987,21 @@ event_create_dir(struct ftrace_event_call *call, struct dentry *d_events,
 		 		  id);
 #endif
 
-	if (call->class->define_fields) {
-		/*
-		 * Other events may have the same class. Only update
-		 * the fields if they are not already defined.
-		 */
-		head = trace_get_fields(call);
-		if (list_empty(head)) {
-			ret = call->class->define_fields(call);
-			if (ret < 0) {
-				pr_warning("Could not initialize trace point"
-					   " events/%s\n", call->name);
-				return ret;
-			}
+	/*
+	 * Other events may have the same class. Only update
+	 * the fields if they are not already defined.
+	 */
+	head = trace_get_fields(call);
+	if (list_empty(head)) {
+		ret = call->class->define_fields(call);
+		if (ret < 0) {
+			pr_warning("Could not initialize trace point"
+				   " events/%s\n", call->name);
+			return ret;
 		}
-		trace_create_file("filter", 0644, call->dir, call,
-				  filter);
 	}
+	trace_create_file("filter", 0644, call->dir, call,
+			  filter);
 
 	trace_create_file("format", 0444, call->dir, call,
 			  format);
diff --git a/kernel/trace/trace_events_filter.c b/kernel/trace/trace_events_filter.c
index 330fefd..36d4010 100644
--- a/kernel/trace/trace_events_filter.c
+++ b/kernel/trace/trace_events_filter.c
@@ -639,9 +639,6 @@ static int init_subsystem_preds(struct event_subsystem *system)
 	int err;
 ...
[ message continues ]
[view original message]
From: Ingo Molnar
Subject: Re: [PATCH 0/9] [GI PULL][2.6.36] tracing: various updates
Date: Tuesday, June 29, 2010 - 1:42 am

Pulled, thanks a lot Steve!

	Ingo
--


[ message continues ]
Previous thread: [PATCH 1/9] tracing: Use a global field list for all syscall exit events by Steven Rostedt on Monday, June 28, 2010 - 8:08 pm. (1 message)

Next thread: linux-next: build failure after merge of the spi tree by Stephen Rothwell on Monday, June 28, 2010 - 8:29 pm. (2 messages)