2007-10-19 10:41:06 +04:00
|
|
|
#ifndef _LINUX_MARKER_H
|
|
|
|
#define _LINUX_MARKER_H
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Code markup for dynamic and static tracing.
|
|
|
|
*
|
|
|
|
* See Documentation/marker.txt.
|
|
|
|
*
|
|
|
|
* (C) Copyright 2006 Mathieu Desnoyers <mathieu.desnoyers@polymtl.ca>
|
|
|
|
*
|
|
|
|
* This file is released under the GPLv2.
|
|
|
|
* See the file COPYING for more details.
|
|
|
|
*/
|
|
|
|
|
2008-11-15 01:47:36 +03:00
|
|
|
#include <stdarg.h>
|
2007-10-19 10:41:06 +04:00
|
|
|
#include <linux/types.h>
|
|
|
|
|
|
|
|
struct module;
|
|
|
|
struct marker;
|
|
|
|
|
|
|
|
/**
|
|
|
|
* marker_probe_func - Type of a marker probe function
|
Linux Kernel Markers: support multiple probes
RCU style multiple probes support for the Linux Kernel Markers. Common case
(one probe) is still fast and does not require dynamic allocation or a
supplementary pointer dereference on the fast path.
- Move preempt disable from the marker site to the callback.
Since we now have an internal callback, move the preempt disable/enable to the
callback instead of the marker site.
Since the callback change is done asynchronously (passing from a handler that
supports arguments to a handler that does not setup the arguments is no
arguments are passed), we can safely update it even if it is outside the
preempt disable section.
- Move probe arm to probe connection. Now, a connected probe is automatically
armed.
Remove MARK_MAX_FORMAT_LEN, unused.
This patch modifies the Linux Kernel Markers API : it removes the probe
"arm/disarm" and changes the probe function prototype : it now expects a
va_list * instead of a "...".
If we want to have more than one probe connected to a marker at a given
time (LTTng, or blktrace, ssytemtap) then we need this patch. Without it,
connecting a second probe handler to a marker will fail.
It allow us, for instance, to do interesting combinations :
Do standard tracing with LTTng and, eventually, to compute statistics
with SystemTAP, or to have a special trigger on an event that would call
a systemtap script which would stop flight recorder tracing.
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@polymtl.ca>
Cc: Christoph Hellwig <hch@infradead.org>
Cc: Mike Mason <mmlnx@us.ibm.com>
Cc: Dipankar Sarma <dipankar@in.ibm.com>
Cc: David Smith <dsmith@redhat.com>
Cc: "Paul E. McKenney" <paulmck@us.ibm.com>
Cc: "Frank Ch. Eigler" <fche@redhat.com>
Cc: Steven Rostedt <rostedt@goodmis.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-02-14 02:03:37 +03:00
|
|
|
* @probe_private: probe private data
|
|
|
|
* @call_private: call site private data
|
2007-10-19 10:41:06 +04:00
|
|
|
* @fmt: format string
|
Linux Kernel Markers: support multiple probes
RCU style multiple probes support for the Linux Kernel Markers. Common case
(one probe) is still fast and does not require dynamic allocation or a
supplementary pointer dereference on the fast path.
- Move preempt disable from the marker site to the callback.
Since we now have an internal callback, move the preempt disable/enable to the
callback instead of the marker site.
Since the callback change is done asynchronously (passing from a handler that
supports arguments to a handler that does not setup the arguments is no
arguments are passed), we can safely update it even if it is outside the
preempt disable section.
- Move probe arm to probe connection. Now, a connected probe is automatically
armed.
Remove MARK_MAX_FORMAT_LEN, unused.
This patch modifies the Linux Kernel Markers API : it removes the probe
"arm/disarm" and changes the probe function prototype : it now expects a
va_list * instead of a "...".
If we want to have more than one probe connected to a marker at a given
time (LTTng, or blktrace, ssytemtap) then we need this patch. Without it,
connecting a second probe handler to a marker will fail.
It allow us, for instance, to do interesting combinations :
Do standard tracing with LTTng and, eventually, to compute statistics
with SystemTAP, or to have a special trigger on an event that would call
a systemtap script which would stop flight recorder tracing.
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@polymtl.ca>
Cc: Christoph Hellwig <hch@infradead.org>
Cc: Mike Mason <mmlnx@us.ibm.com>
Cc: Dipankar Sarma <dipankar@in.ibm.com>
Cc: David Smith <dsmith@redhat.com>
Cc: "Paul E. McKenney" <paulmck@us.ibm.com>
Cc: "Frank Ch. Eigler" <fche@redhat.com>
Cc: Steven Rostedt <rostedt@goodmis.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-02-14 02:03:37 +03:00
|
|
|
* @args: variable argument list pointer. Use a pointer to overcome C's
|
|
|
|
* inability to pass this around as a pointer in a portable manner in
|
|
|
|
* the callee otherwise.
|
2007-10-19 10:41:06 +04:00
|
|
|
*
|
|
|
|
* Type of marker probe functions. They receive the mdata and need to parse the
|
|
|
|
* format string to recover the variable argument list.
|
|
|
|
*/
|
Linux Kernel Markers: support multiple probes
RCU style multiple probes support for the Linux Kernel Markers. Common case
(one probe) is still fast and does not require dynamic allocation or a
supplementary pointer dereference on the fast path.
- Move preempt disable from the marker site to the callback.
Since we now have an internal callback, move the preempt disable/enable to the
callback instead of the marker site.
Since the callback change is done asynchronously (passing from a handler that
supports arguments to a handler that does not setup the arguments is no
arguments are passed), we can safely update it even if it is outside the
preempt disable section.
- Move probe arm to probe connection. Now, a connected probe is automatically
armed.
Remove MARK_MAX_FORMAT_LEN, unused.
This patch modifies the Linux Kernel Markers API : it removes the probe
"arm/disarm" and changes the probe function prototype : it now expects a
va_list * instead of a "...".
If we want to have more than one probe connected to a marker at a given
time (LTTng, or blktrace, ssytemtap) then we need this patch. Without it,
connecting a second probe handler to a marker will fail.
It allow us, for instance, to do interesting combinations :
Do standard tracing with LTTng and, eventually, to compute statistics
with SystemTAP, or to have a special trigger on an event that would call
a systemtap script which would stop flight recorder tracing.
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@polymtl.ca>
Cc: Christoph Hellwig <hch@infradead.org>
Cc: Mike Mason <mmlnx@us.ibm.com>
Cc: Dipankar Sarma <dipankar@in.ibm.com>
Cc: David Smith <dsmith@redhat.com>
Cc: "Paul E. McKenney" <paulmck@us.ibm.com>
Cc: "Frank Ch. Eigler" <fche@redhat.com>
Cc: Steven Rostedt <rostedt@goodmis.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-02-14 02:03:37 +03:00
|
|
|
typedef void marker_probe_func(void *probe_private, void *call_private,
|
|
|
|
const char *fmt, va_list *args);
|
|
|
|
|
|
|
|
struct marker_probe_closure {
|
|
|
|
marker_probe_func *func; /* Callback */
|
|
|
|
void *probe_private; /* Private probe data */
|
|
|
|
};
|
2007-10-19 10:41:06 +04:00
|
|
|
|
|
|
|
struct marker {
|
|
|
|
const char *name; /* Marker name */
|
|
|
|
const char *format; /* Marker format string, describing the
|
|
|
|
* variable argument list.
|
|
|
|
*/
|
|
|
|
char state; /* Marker state. */
|
Linux Kernel Markers: support multiple probes
RCU style multiple probes support for the Linux Kernel Markers. Common case
(one probe) is still fast and does not require dynamic allocation or a
supplementary pointer dereference on the fast path.
- Move preempt disable from the marker site to the callback.
Since we now have an internal callback, move the preempt disable/enable to the
callback instead of the marker site.
Since the callback change is done asynchronously (passing from a handler that
supports arguments to a handler that does not setup the arguments is no
arguments are passed), we can safely update it even if it is outside the
preempt disable section.
- Move probe arm to probe connection. Now, a connected probe is automatically
armed.
Remove MARK_MAX_FORMAT_LEN, unused.
This patch modifies the Linux Kernel Markers API : it removes the probe
"arm/disarm" and changes the probe function prototype : it now expects a
va_list * instead of a "...".
If we want to have more than one probe connected to a marker at a given
time (LTTng, or blktrace, ssytemtap) then we need this patch. Without it,
connecting a second probe handler to a marker will fail.
It allow us, for instance, to do interesting combinations :
Do standard tracing with LTTng and, eventually, to compute statistics
with SystemTAP, or to have a special trigger on an event that would call
a systemtap script which would stop flight recorder tracing.
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@polymtl.ca>
Cc: Christoph Hellwig <hch@infradead.org>
Cc: Mike Mason <mmlnx@us.ibm.com>
Cc: Dipankar Sarma <dipankar@in.ibm.com>
Cc: David Smith <dsmith@redhat.com>
Cc: "Paul E. McKenney" <paulmck@us.ibm.com>
Cc: "Frank Ch. Eigler" <fche@redhat.com>
Cc: Steven Rostedt <rostedt@goodmis.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-02-14 02:03:37 +03:00
|
|
|
char ptype; /* probe type : 0 : single, 1 : multi */
|
Markers - remove extra format argument
Denys Vlasenko <vda.linux@googlemail.com> :
> Not in this patch, but I noticed:
>
> #define __trace_mark(name, call_private, format, args...) \
> do { \
> static const char __mstrtab_##name[] \
> __attribute__((section("__markers_strings"))) \
> = #name "\0" format; \
> static struct marker __mark_##name \
> __attribute__((section("__markers"), aligned(8))) = \
> { __mstrtab_##name, &__mstrtab_##name[sizeof(#name)], \
> 0, 0, marker_probe_cb, \
> { __mark_empty_function, NULL}, NULL }; \
> __mark_check_format(format, ## args); \
> if (unlikely(__mark_##name.state)) { \
> (*__mark_##name.call) \
> (&__mark_##name, call_private, \
> format, ## args); \
> } \
> } while (0)
>
> In this call:
>
> (*__mark_##name.call) \
> (&__mark_##name, call_private, \
> format, ## args); \
>
> you make gcc allocate duplicate format string. You can use
> &__mstrtab_##name[sizeof(#name)] instead since it holds the same string,
> or drop ", format," above and "const char *fmt" from here:
>
> void (*call)(const struct marker *mdata, /* Probe wrapper */
> void *call_private, const char *fmt, ...);
>
> since mdata->format is the same and all callees which need it can take it there.
Very good point. I actually thought about dropping it, since it would
remove an unnecessary argument from the stack. And actually, since I now
have the marker_probe_cb sitting between the marker site and the
callbacks, there is no API change required. Thanks :)
Mathieu
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@polymtl.ca>
CC: Denys Vlasenko <vda.linux@googlemail.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
2008-05-12 23:21:09 +04:00
|
|
|
/* Probe wrapper */
|
|
|
|
void (*call)(const struct marker *mdata, void *call_private, ...);
|
Linux Kernel Markers: support multiple probes
RCU style multiple probes support for the Linux Kernel Markers. Common case
(one probe) is still fast and does not require dynamic allocation or a
supplementary pointer dereference on the fast path.
- Move preempt disable from the marker site to the callback.
Since we now have an internal callback, move the preempt disable/enable to the
callback instead of the marker site.
Since the callback change is done asynchronously (passing from a handler that
supports arguments to a handler that does not setup the arguments is no
arguments are passed), we can safely update it even if it is outside the
preempt disable section.
- Move probe arm to probe connection. Now, a connected probe is automatically
armed.
Remove MARK_MAX_FORMAT_LEN, unused.
This patch modifies the Linux Kernel Markers API : it removes the probe
"arm/disarm" and changes the probe function prototype : it now expects a
va_list * instead of a "...".
If we want to have more than one probe connected to a marker at a given
time (LTTng, or blktrace, ssytemtap) then we need this patch. Without it,
connecting a second probe handler to a marker will fail.
It allow us, for instance, to do interesting combinations :
Do standard tracing with LTTng and, eventually, to compute statistics
with SystemTAP, or to have a special trigger on an event that would call
a systemtap script which would stop flight recorder tracing.
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@polymtl.ca>
Cc: Christoph Hellwig <hch@infradead.org>
Cc: Mike Mason <mmlnx@us.ibm.com>
Cc: Dipankar Sarma <dipankar@in.ibm.com>
Cc: David Smith <dsmith@redhat.com>
Cc: "Paul E. McKenney" <paulmck@us.ibm.com>
Cc: "Frank Ch. Eigler" <fche@redhat.com>
Cc: Steven Rostedt <rostedt@goodmis.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-02-14 02:03:37 +03:00
|
|
|
struct marker_probe_closure single;
|
|
|
|
struct marker_probe_closure *multi;
|
2008-11-15 01:47:39 +03:00
|
|
|
const char *tp_name; /* Optional tracepoint name */
|
|
|
|
void *tp_cb; /* Optional tracepoint callback */
|
2007-10-19 10:41:06 +04:00
|
|
|
} __attribute__((aligned(8)));
|
|
|
|
|
|
|
|
#ifdef CONFIG_MARKERS
|
|
|
|
|
2008-11-15 01:47:40 +03:00
|
|
|
#define _DEFINE_MARKER(name, tp_name_str, tp_cb, format) \
|
|
|
|
static const char __mstrtab_##name[] \
|
|
|
|
__attribute__((section("__markers_strings"))) \
|
|
|
|
= #name "\0" format; \
|
|
|
|
static struct marker __mark_##name \
|
|
|
|
__attribute__((section("__markers"), aligned(8))) = \
|
|
|
|
{ __mstrtab_##name, &__mstrtab_##name[sizeof(#name)], \
|
|
|
|
0, 0, marker_probe_cb, { __mark_empty_function, NULL},\
|
|
|
|
NULL, tp_name_str, tp_cb }
|
|
|
|
|
|
|
|
#define DEFINE_MARKER(name, format) \
|
|
|
|
_DEFINE_MARKER(name, NULL, NULL, format)
|
|
|
|
|
|
|
|
#define DEFINE_MARKER_TP(name, tp_name, tp_cb, format) \
|
|
|
|
_DEFINE_MARKER(name, #tp_name, tp_cb, format)
|
|
|
|
|
2007-10-19 10:41:06 +04:00
|
|
|
/*
|
|
|
|
* Note : the empty asm volatile with read constraint is used here instead of a
|
|
|
|
* "used" attribute to fix a gcc 4.1.x bug.
|
|
|
|
* Make sure the alignment of the structure in the __markers section will
|
|
|
|
* not add unwanted padding between the beginning of the section and the
|
|
|
|
* structure. Force alignment to the same alignment as the section start.
|
2008-05-12 23:21:10 +04:00
|
|
|
*
|
|
|
|
* The "generic" argument controls which marker enabling mechanism must be used.
|
|
|
|
* If generic is true, a variable read is used.
|
|
|
|
* If generic is false, immediate values are used.
|
2007-10-19 10:41:06 +04:00
|
|
|
*/
|
2008-05-12 23:21:10 +04:00
|
|
|
#define __trace_mark(generic, name, call_private, format, args...) \
|
2007-10-19 10:41:06 +04:00
|
|
|
do { \
|
2008-11-15 01:47:40 +03:00
|
|
|
DEFINE_MARKER(name, format); \
|
2007-10-19 10:41:06 +04:00
|
|
|
__mark_check_format(format, ## args); \
|
|
|
|
if (unlikely(__mark_##name.state)) { \
|
|
|
|
(*__mark_##name.call) \
|
Markers - remove extra format argument
Denys Vlasenko <vda.linux@googlemail.com> :
> Not in this patch, but I noticed:
>
> #define __trace_mark(name, call_private, format, args...) \
> do { \
> static const char __mstrtab_##name[] \
> __attribute__((section("__markers_strings"))) \
> = #name "\0" format; \
> static struct marker __mark_##name \
> __attribute__((section("__markers"), aligned(8))) = \
> { __mstrtab_##name, &__mstrtab_##name[sizeof(#name)], \
> 0, 0, marker_probe_cb, \
> { __mark_empty_function, NULL}, NULL }; \
> __mark_check_format(format, ## args); \
> if (unlikely(__mark_##name.state)) { \
> (*__mark_##name.call) \
> (&__mark_##name, call_private, \
> format, ## args); \
> } \
> } while (0)
>
> In this call:
>
> (*__mark_##name.call) \
> (&__mark_##name, call_private, \
> format, ## args); \
>
> you make gcc allocate duplicate format string. You can use
> &__mstrtab_##name[sizeof(#name)] instead since it holds the same string,
> or drop ", format," above and "const char *fmt" from here:
>
> void (*call)(const struct marker *mdata, /* Probe wrapper */
> void *call_private, const char *fmt, ...);
>
> since mdata->format is the same and all callees which need it can take it there.
Very good point. I actually thought about dropping it, since it would
remove an unnecessary argument from the stack. And actually, since I now
have the marker_probe_cb sitting between the marker site and the
callbacks, there is no API change required. Thanks :)
Mathieu
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@polymtl.ca>
CC: Denys Vlasenko <vda.linux@googlemail.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
2008-05-12 23:21:09 +04:00
|
|
|
(&__mark_##name, call_private, ## args);\
|
2007-10-19 10:41:06 +04:00
|
|
|
} \
|
|
|
|
} while (0)
|
|
|
|
|
2008-11-15 01:47:39 +03:00
|
|
|
#define __trace_mark_tp(name, call_private, tp_name, tp_cb, format, args...) \
|
|
|
|
do { \
|
|
|
|
void __check_tp_type(void) \
|
|
|
|
{ \
|
|
|
|
register_trace_##tp_name(tp_cb); \
|
|
|
|
} \
|
2008-11-15 01:47:40 +03:00
|
|
|
DEFINE_MARKER_TP(name, tp_name, tp_cb, format); \
|
2008-11-15 01:47:39 +03:00
|
|
|
__mark_check_format(format, ## args); \
|
|
|
|
(*__mark_##name.call)(&__mark_##name, call_private, \
|
|
|
|
## args); \
|
|
|
|
} while (0)
|
|
|
|
|
2007-10-19 10:41:06 +04:00
|
|
|
extern void marker_update_probe_range(struct marker *begin,
|
Linux Kernel Markers: support multiple probes
RCU style multiple probes support for the Linux Kernel Markers. Common case
(one probe) is still fast and does not require dynamic allocation or a
supplementary pointer dereference on the fast path.
- Move preempt disable from the marker site to the callback.
Since we now have an internal callback, move the preempt disable/enable to the
callback instead of the marker site.
Since the callback change is done asynchronously (passing from a handler that
supports arguments to a handler that does not setup the arguments is no
arguments are passed), we can safely update it even if it is outside the
preempt disable section.
- Move probe arm to probe connection. Now, a connected probe is automatically
armed.
Remove MARK_MAX_FORMAT_LEN, unused.
This patch modifies the Linux Kernel Markers API : it removes the probe
"arm/disarm" and changes the probe function prototype : it now expects a
va_list * instead of a "...".
If we want to have more than one probe connected to a marker at a given
time (LTTng, or blktrace, ssytemtap) then we need this patch. Without it,
connecting a second probe handler to a marker will fail.
It allow us, for instance, to do interesting combinations :
Do standard tracing with LTTng and, eventually, to compute statistics
with SystemTAP, or to have a special trigger on an event that would call
a systemtap script which would stop flight recorder tracing.
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@polymtl.ca>
Cc: Christoph Hellwig <hch@infradead.org>
Cc: Mike Mason <mmlnx@us.ibm.com>
Cc: Dipankar Sarma <dipankar@in.ibm.com>
Cc: David Smith <dsmith@redhat.com>
Cc: "Paul E. McKenney" <paulmck@us.ibm.com>
Cc: "Frank Ch. Eigler" <fche@redhat.com>
Cc: Steven Rostedt <rostedt@goodmis.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-02-14 02:03:37 +03:00
|
|
|
struct marker *end);
|
2008-11-15 01:47:40 +03:00
|
|
|
|
|
|
|
#define GET_MARKER(name) (__mark_##name)
|
|
|
|
|
2007-10-19 10:41:06 +04:00
|
|
|
#else /* !CONFIG_MARKERS */
|
2008-11-15 01:47:40 +03:00
|
|
|
#define DEFINE_MARKER(name, tp_name, tp_cb, format)
|
2008-05-12 23:21:10 +04:00
|
|
|
#define __trace_mark(generic, name, call_private, format, args...) \
|
2007-10-19 10:41:06 +04:00
|
|
|
__mark_check_format(format, ## args)
|
2008-11-15 01:47:39 +03:00
|
|
|
#define __trace_mark_tp(name, call_private, tp_name, tp_cb, format, args...) \
|
|
|
|
do { \
|
|
|
|
void __check_tp_type(void) \
|
|
|
|
{ \
|
|
|
|
register_trace_##tp_name(tp_cb); \
|
|
|
|
} \
|
|
|
|
__mark_check_format(format, ## args); \
|
|
|
|
} while (0)
|
2007-10-19 10:41:06 +04:00
|
|
|
static inline void marker_update_probe_range(struct marker *begin,
|
Linux Kernel Markers: support multiple probes
RCU style multiple probes support for the Linux Kernel Markers. Common case
(one probe) is still fast and does not require dynamic allocation or a
supplementary pointer dereference on the fast path.
- Move preempt disable from the marker site to the callback.
Since we now have an internal callback, move the preempt disable/enable to the
callback instead of the marker site.
Since the callback change is done asynchronously (passing from a handler that
supports arguments to a handler that does not setup the arguments is no
arguments are passed), we can safely update it even if it is outside the
preempt disable section.
- Move probe arm to probe connection. Now, a connected probe is automatically
armed.
Remove MARK_MAX_FORMAT_LEN, unused.
This patch modifies the Linux Kernel Markers API : it removes the probe
"arm/disarm" and changes the probe function prototype : it now expects a
va_list * instead of a "...".
If we want to have more than one probe connected to a marker at a given
time (LTTng, or blktrace, ssytemtap) then we need this patch. Without it,
connecting a second probe handler to a marker will fail.
It allow us, for instance, to do interesting combinations :
Do standard tracing with LTTng and, eventually, to compute statistics
with SystemTAP, or to have a special trigger on an event that would call
a systemtap script which would stop flight recorder tracing.
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@polymtl.ca>
Cc: Christoph Hellwig <hch@infradead.org>
Cc: Mike Mason <mmlnx@us.ibm.com>
Cc: Dipankar Sarma <dipankar@in.ibm.com>
Cc: David Smith <dsmith@redhat.com>
Cc: "Paul E. McKenney" <paulmck@us.ibm.com>
Cc: "Frank Ch. Eigler" <fche@redhat.com>
Cc: Steven Rostedt <rostedt@goodmis.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-02-14 02:03:37 +03:00
|
|
|
struct marker *end)
|
2007-10-19 10:41:06 +04:00
|
|
|
{ }
|
2008-11-15 01:47:40 +03:00
|
|
|
#define GET_MARKER(name)
|
2007-10-19 10:41:06 +04:00
|
|
|
#endif /* CONFIG_MARKERS */
|
|
|
|
|
|
|
|
/**
|
2008-05-12 23:21:10 +04:00
|
|
|
* trace_mark - Marker using code patching
|
2007-10-19 10:41:06 +04:00
|
|
|
* @name: marker name, not quoted.
|
|
|
|
* @format: format string
|
|
|
|
* @args...: variable argument list
|
|
|
|
*
|
2008-05-12 23:21:10 +04:00
|
|
|
* Places a marker using optimized code patching technique (imv_read())
|
|
|
|
* to be enabled when immediate values are present.
|
2007-10-19 10:41:06 +04:00
|
|
|
*/
|
|
|
|
#define trace_mark(name, format, args...) \
|
2008-05-12 23:21:10 +04:00
|
|
|
__trace_mark(0, name, NULL, format, ## args)
|
|
|
|
|
|
|
|
/**
|
|
|
|
* _trace_mark - Marker using variable read
|
|
|
|
* @name: marker name, not quoted.
|
|
|
|
* @format: format string
|
|
|
|
* @args...: variable argument list
|
|
|
|
*
|
|
|
|
* Places a marker using a standard memory read (_imv_read()) to be
|
|
|
|
* enabled. Should be used for markers in code paths where instruction
|
|
|
|
* modification based enabling is not welcome. (__init and __exit functions,
|
|
|
|
* lockdep, some traps, printk).
|
|
|
|
*/
|
|
|
|
#define _trace_mark(name, format, args...) \
|
|
|
|
__trace_mark(1, name, NULL, format, ## args)
|
2007-10-19 10:41:06 +04:00
|
|
|
|
2008-11-15 01:47:39 +03:00
|
|
|
/**
|
|
|
|
* trace_mark_tp - Marker in a tracepoint callback
|
|
|
|
* @name: marker name, not quoted.
|
|
|
|
* @tp_name: tracepoint name, not quoted.
|
|
|
|
* @tp_cb: tracepoint callback. Should have an associated global symbol so it
|
|
|
|
* is not optimized away by the compiler (should not be static).
|
|
|
|
* @format: format string
|
|
|
|
* @args...: variable argument list
|
|
|
|
*
|
|
|
|
* Places a marker in a tracepoint callback.
|
|
|
|
*/
|
|
|
|
#define trace_mark_tp(name, tp_name, tp_cb, format, args...) \
|
|
|
|
__trace_mark_tp(name, NULL, tp_name, tp_cb, format, ## args)
|
|
|
|
|
2007-10-19 10:41:06 +04:00
|
|
|
/**
|
|
|
|
* MARK_NOARGS - Format string for a marker with no argument.
|
|
|
|
*/
|
|
|
|
#define MARK_NOARGS " "
|
|
|
|
|
|
|
|
/* To be used for string format validity checking with gcc */
|
2008-03-05 01:29:00 +03:00
|
|
|
static inline void __printf(1, 2) ___mark_check_format(const char *fmt, ...)
|
2007-10-19 10:41:06 +04:00
|
|
|
{
|
|
|
|
}
|
|
|
|
|
2008-03-05 01:29:00 +03:00
|
|
|
#define __mark_check_format(format, args...) \
|
|
|
|
do { \
|
|
|
|
if (0) \
|
|
|
|
___mark_check_format(format, ## args); \
|
|
|
|
} while (0)
|
|
|
|
|
2007-10-19 10:41:06 +04:00
|
|
|
extern marker_probe_func __mark_empty_function;
|
|
|
|
|
Linux Kernel Markers: support multiple probes
RCU style multiple probes support for the Linux Kernel Markers. Common case
(one probe) is still fast and does not require dynamic allocation or a
supplementary pointer dereference on the fast path.
- Move preempt disable from the marker site to the callback.
Since we now have an internal callback, move the preempt disable/enable to the
callback instead of the marker site.
Since the callback change is done asynchronously (passing from a handler that
supports arguments to a handler that does not setup the arguments is no
arguments are passed), we can safely update it even if it is outside the
preempt disable section.
- Move probe arm to probe connection. Now, a connected probe is automatically
armed.
Remove MARK_MAX_FORMAT_LEN, unused.
This patch modifies the Linux Kernel Markers API : it removes the probe
"arm/disarm" and changes the probe function prototype : it now expects a
va_list * instead of a "...".
If we want to have more than one probe connected to a marker at a given
time (LTTng, or blktrace, ssytemtap) then we need this patch. Without it,
connecting a second probe handler to a marker will fail.
It allow us, for instance, to do interesting combinations :
Do standard tracing with LTTng and, eventually, to compute statistics
with SystemTAP, or to have a special trigger on an event that would call
a systemtap script which would stop flight recorder tracing.
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@polymtl.ca>
Cc: Christoph Hellwig <hch@infradead.org>
Cc: Mike Mason <mmlnx@us.ibm.com>
Cc: Dipankar Sarma <dipankar@in.ibm.com>
Cc: David Smith <dsmith@redhat.com>
Cc: "Paul E. McKenney" <paulmck@us.ibm.com>
Cc: "Frank Ch. Eigler" <fche@redhat.com>
Cc: Steven Rostedt <rostedt@goodmis.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-02-14 02:03:37 +03:00
|
|
|
extern void marker_probe_cb(const struct marker *mdata,
|
Markers - remove extra format argument
Denys Vlasenko <vda.linux@googlemail.com> :
> Not in this patch, but I noticed:
>
> #define __trace_mark(name, call_private, format, args...) \
> do { \
> static const char __mstrtab_##name[] \
> __attribute__((section("__markers_strings"))) \
> = #name "\0" format; \
> static struct marker __mark_##name \
> __attribute__((section("__markers"), aligned(8))) = \
> { __mstrtab_##name, &__mstrtab_##name[sizeof(#name)], \
> 0, 0, marker_probe_cb, \
> { __mark_empty_function, NULL}, NULL }; \
> __mark_check_format(format, ## args); \
> if (unlikely(__mark_##name.state)) { \
> (*__mark_##name.call) \
> (&__mark_##name, call_private, \
> format, ## args); \
> } \
> } while (0)
>
> In this call:
>
> (*__mark_##name.call) \
> (&__mark_##name, call_private, \
> format, ## args); \
>
> you make gcc allocate duplicate format string. You can use
> &__mstrtab_##name[sizeof(#name)] instead since it holds the same string,
> or drop ", format," above and "const char *fmt" from here:
>
> void (*call)(const struct marker *mdata, /* Probe wrapper */
> void *call_private, const char *fmt, ...);
>
> since mdata->format is the same and all callees which need it can take it there.
Very good point. I actually thought about dropping it, since it would
remove an unnecessary argument from the stack. And actually, since I now
have the marker_probe_cb sitting between the marker site and the
callbacks, there is no API change required. Thanks :)
Mathieu
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@polymtl.ca>
CC: Denys Vlasenko <vda.linux@googlemail.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
2008-05-12 23:21:09 +04:00
|
|
|
void *call_private, ...);
|
Linux Kernel Markers: support multiple probes
RCU style multiple probes support for the Linux Kernel Markers. Common case
(one probe) is still fast and does not require dynamic allocation or a
supplementary pointer dereference on the fast path.
- Move preempt disable from the marker site to the callback.
Since we now have an internal callback, move the preempt disable/enable to the
callback instead of the marker site.
Since the callback change is done asynchronously (passing from a handler that
supports arguments to a handler that does not setup the arguments is no
arguments are passed), we can safely update it even if it is outside the
preempt disable section.
- Move probe arm to probe connection. Now, a connected probe is automatically
armed.
Remove MARK_MAX_FORMAT_LEN, unused.
This patch modifies the Linux Kernel Markers API : it removes the probe
"arm/disarm" and changes the probe function prototype : it now expects a
va_list * instead of a "...".
If we want to have more than one probe connected to a marker at a given
time (LTTng, or blktrace, ssytemtap) then we need this patch. Without it,
connecting a second probe handler to a marker will fail.
It allow us, for instance, to do interesting combinations :
Do standard tracing with LTTng and, eventually, to compute statistics
with SystemTAP, or to have a special trigger on an event that would call
a systemtap script which would stop flight recorder tracing.
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@polymtl.ca>
Cc: Christoph Hellwig <hch@infradead.org>
Cc: Mike Mason <mmlnx@us.ibm.com>
Cc: Dipankar Sarma <dipankar@in.ibm.com>
Cc: David Smith <dsmith@redhat.com>
Cc: "Paul E. McKenney" <paulmck@us.ibm.com>
Cc: "Frank Ch. Eigler" <fche@redhat.com>
Cc: Steven Rostedt <rostedt@goodmis.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-02-14 02:03:37 +03:00
|
|
|
|
2007-10-19 10:41:06 +04:00
|
|
|
/*
|
|
|
|
* Connect a probe to a marker.
|
|
|
|
* private data pointer must be a valid allocated memory address, or NULL.
|
|
|
|
*/
|
|
|
|
extern int marker_probe_register(const char *name, const char *format,
|
Linux Kernel Markers: support multiple probes
RCU style multiple probes support for the Linux Kernel Markers. Common case
(one probe) is still fast and does not require dynamic allocation or a
supplementary pointer dereference on the fast path.
- Move preempt disable from the marker site to the callback.
Since we now have an internal callback, move the preempt disable/enable to the
callback instead of the marker site.
Since the callback change is done asynchronously (passing from a handler that
supports arguments to a handler that does not setup the arguments is no
arguments are passed), we can safely update it even if it is outside the
preempt disable section.
- Move probe arm to probe connection. Now, a connected probe is automatically
armed.
Remove MARK_MAX_FORMAT_LEN, unused.
This patch modifies the Linux Kernel Markers API : it removes the probe
"arm/disarm" and changes the probe function prototype : it now expects a
va_list * instead of a "...".
If we want to have more than one probe connected to a marker at a given
time (LTTng, or blktrace, ssytemtap) then we need this patch. Without it,
connecting a second probe handler to a marker will fail.
It allow us, for instance, to do interesting combinations :
Do standard tracing with LTTng and, eventually, to compute statistics
with SystemTAP, or to have a special trigger on an event that would call
a systemtap script which would stop flight recorder tracing.
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@polymtl.ca>
Cc: Christoph Hellwig <hch@infradead.org>
Cc: Mike Mason <mmlnx@us.ibm.com>
Cc: Dipankar Sarma <dipankar@in.ibm.com>
Cc: David Smith <dsmith@redhat.com>
Cc: "Paul E. McKenney" <paulmck@us.ibm.com>
Cc: "Frank Ch. Eigler" <fche@redhat.com>
Cc: Steven Rostedt <rostedt@goodmis.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-02-14 02:03:37 +03:00
|
|
|
marker_probe_func *probe, void *probe_private);
|
2007-10-19 10:41:06 +04:00
|
|
|
|
|
|
|
/*
|
|
|
|
* Returns the private data given to marker_probe_register.
|
|
|
|
*/
|
Linux Kernel Markers: support multiple probes
RCU style multiple probes support for the Linux Kernel Markers. Common case
(one probe) is still fast and does not require dynamic allocation or a
supplementary pointer dereference on the fast path.
- Move preempt disable from the marker site to the callback.
Since we now have an internal callback, move the preempt disable/enable to the
callback instead of the marker site.
Since the callback change is done asynchronously (passing from a handler that
supports arguments to a handler that does not setup the arguments is no
arguments are passed), we can safely update it even if it is outside the
preempt disable section.
- Move probe arm to probe connection. Now, a connected probe is automatically
armed.
Remove MARK_MAX_FORMAT_LEN, unused.
This patch modifies the Linux Kernel Markers API : it removes the probe
"arm/disarm" and changes the probe function prototype : it now expects a
va_list * instead of a "...".
If we want to have more than one probe connected to a marker at a given
time (LTTng, or blktrace, ssytemtap) then we need this patch. Without it,
connecting a second probe handler to a marker will fail.
It allow us, for instance, to do interesting combinations :
Do standard tracing with LTTng and, eventually, to compute statistics
with SystemTAP, or to have a special trigger on an event that would call
a systemtap script which would stop flight recorder tracing.
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@polymtl.ca>
Cc: Christoph Hellwig <hch@infradead.org>
Cc: Mike Mason <mmlnx@us.ibm.com>
Cc: Dipankar Sarma <dipankar@in.ibm.com>
Cc: David Smith <dsmith@redhat.com>
Cc: "Paul E. McKenney" <paulmck@us.ibm.com>
Cc: "Frank Ch. Eigler" <fche@redhat.com>
Cc: Steven Rostedt <rostedt@goodmis.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-02-14 02:03:37 +03:00
|
|
|
extern int marker_probe_unregister(const char *name,
|
|
|
|
marker_probe_func *probe, void *probe_private);
|
2007-10-19 10:41:06 +04:00
|
|
|
/*
|
|
|
|
* Unregister a marker by providing the registered private data.
|
|
|
|
*/
|
Linux Kernel Markers: support multiple probes
RCU style multiple probes support for the Linux Kernel Markers. Common case
(one probe) is still fast and does not require dynamic allocation or a
supplementary pointer dereference on the fast path.
- Move preempt disable from the marker site to the callback.
Since we now have an internal callback, move the preempt disable/enable to the
callback instead of the marker site.
Since the callback change is done asynchronously (passing from a handler that
supports arguments to a handler that does not setup the arguments is no
arguments are passed), we can safely update it even if it is outside the
preempt disable section.
- Move probe arm to probe connection. Now, a connected probe is automatically
armed.
Remove MARK_MAX_FORMAT_LEN, unused.
This patch modifies the Linux Kernel Markers API : it removes the probe
"arm/disarm" and changes the probe function prototype : it now expects a
va_list * instead of a "...".
If we want to have more than one probe connected to a marker at a given
time (LTTng, or blktrace, ssytemtap) then we need this patch. Without it,
connecting a second probe handler to a marker will fail.
It allow us, for instance, to do interesting combinations :
Do standard tracing with LTTng and, eventually, to compute statistics
with SystemTAP, or to have a special trigger on an event that would call
a systemtap script which would stop flight recorder tracing.
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@polymtl.ca>
Cc: Christoph Hellwig <hch@infradead.org>
Cc: Mike Mason <mmlnx@us.ibm.com>
Cc: Dipankar Sarma <dipankar@in.ibm.com>
Cc: David Smith <dsmith@redhat.com>
Cc: "Paul E. McKenney" <paulmck@us.ibm.com>
Cc: "Frank Ch. Eigler" <fche@redhat.com>
Cc: Steven Rostedt <rostedt@goodmis.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-02-14 02:03:37 +03:00
|
|
|
extern int marker_probe_unregister_private_data(marker_probe_func *probe,
|
|
|
|
void *probe_private);
|
2007-10-19 10:41:06 +04:00
|
|
|
|
Linux Kernel Markers: support multiple probes
RCU style multiple probes support for the Linux Kernel Markers. Common case
(one probe) is still fast and does not require dynamic allocation or a
supplementary pointer dereference on the fast path.
- Move preempt disable from the marker site to the callback.
Since we now have an internal callback, move the preempt disable/enable to the
callback instead of the marker site.
Since the callback change is done asynchronously (passing from a handler that
supports arguments to a handler that does not setup the arguments is no
arguments are passed), we can safely update it even if it is outside the
preempt disable section.
- Move probe arm to probe connection. Now, a connected probe is automatically
armed.
Remove MARK_MAX_FORMAT_LEN, unused.
This patch modifies the Linux Kernel Markers API : it removes the probe
"arm/disarm" and changes the probe function prototype : it now expects a
va_list * instead of a "...".
If we want to have more than one probe connected to a marker at a given
time (LTTng, or blktrace, ssytemtap) then we need this patch. Without it,
connecting a second probe handler to a marker will fail.
It allow us, for instance, to do interesting combinations :
Do standard tracing with LTTng and, eventually, to compute statistics
with SystemTAP, or to have a special trigger on an event that would call
a systemtap script which would stop flight recorder tracing.
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@polymtl.ca>
Cc: Christoph Hellwig <hch@infradead.org>
Cc: Mike Mason <mmlnx@us.ibm.com>
Cc: Dipankar Sarma <dipankar@in.ibm.com>
Cc: David Smith <dsmith@redhat.com>
Cc: "Paul E. McKenney" <paulmck@us.ibm.com>
Cc: "Frank Ch. Eigler" <fche@redhat.com>
Cc: Steven Rostedt <rostedt@goodmis.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2008-02-14 02:03:37 +03:00
|
|
|
extern void *marker_get_private_data(const char *name, marker_probe_func *probe,
|
|
|
|
int num);
|
2007-10-19 10:41:06 +04:00
|
|
|
|
2008-09-29 19:05:13 +04:00
|
|
|
/*
|
|
|
|
* marker_synchronize_unregister must be called between the last marker probe
|
2008-11-27 11:14:44 +03:00
|
|
|
* unregistration and the first one of
|
|
|
|
* - the end of module exit function
|
|
|
|
* - the free of any resource used by the probes
|
|
|
|
* to ensure the code and data are valid for any possibly running probes.
|
2008-09-29 19:05:13 +04:00
|
|
|
*/
|
2008-10-10 11:48:25 +04:00
|
|
|
#define marker_synchronize_unregister() synchronize_sched()
|
2008-09-29 19:05:13 +04:00
|
|
|
|
2007-10-19 10:41:06 +04:00
|
|
|
#endif
|