2007-10-18 23:41:06 -07:00
										 
									 
								 
							 | 
							
								
							 | 
							
								
							 | 
							
							
								/*
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								 * Copyright (C) 2007 Mathieu Desnoyers
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								 *
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								 * This program is free software; you can redistribute it and/or modify
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								 * it under the terms of the GNU General Public License as published by
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								 * the Free Software Foundation; either version 2 of the License, or
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								 * (at your option) any later version.
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								 *
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								 * This program is distributed in the hope that it will be useful,
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								 * but WITHOUT ANY WARRANTY; without even the implied warranty of
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								 * GNU General Public License for more details.
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								 *
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								 * You should have received a copy of the GNU General Public License
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								 * along with this program; if not, write to the Free Software
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								 * Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA.
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								 */
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								#include <linux/module.h>
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								#include <linux/mutex.h>
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								#include <linux/types.h>
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								#include <linux/jhash.h>
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								#include <linux/list.h>
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								#include <linux/rcupdate.h>
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								#include <linux/marker.h>
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								#include <linux/err.h>
							 | 
						
					
						
							
								
									
										
										
										
											2008-04-29 00:59:25 -07:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
								#include <linux/slab.h>
							 | 
						
					
						
							
								
									
										
										
										
											2007-10-18 23:41:06 -07:00
										 
									 
								 
							 | 
							
								
							 | 
							
								
							 | 
							
							
								
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								extern struct marker __start___markers[];
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								extern struct marker __stop___markers[];
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								
							 | 
						
					
						
							
								
									
										
											 
										 
										
											
												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>
											
										 
										
											2008-02-13 15:03:37 -08:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
								/* Set to 1 to enable marker debug output */
							 | 
						
					
						
							
								
									
										
										
										
											2008-04-30 00:54:30 -07:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
								static const int marker_debug;
							 | 
						
					
						
							
								
									
										
											 
										 
										
											
												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>
											
										 
										
											2008-02-13 15:03:37 -08:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
								
							 | 
						
					
						
							
								
									
										
										
										
											2007-10-18 23:41:06 -07:00
										 
									 
								 
							 | 
							
								
							 | 
							
								
							 | 
							
							
								/*
							 | 
						
					
						
							
								
									
										
										
										
											2007-11-14 16:59:48 -08:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
								 * markers_mutex nests inside module_mutex. Markers mutex protects the builtin
							 | 
						
					
						
							
								
									
										
											 
										 
										
											
												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>
											
										 
										
											2008-02-13 15:03:37 -08:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
								 * and module markers and the hash table.
							 | 
						
					
						
							
								
									
										
										
										
											2007-10-18 23:41:06 -07:00
										 
									 
								 
							 | 
							
								
							 | 
							
								
							 | 
							
							
								 */
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								static DEFINE_MUTEX(markers_mutex);
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								/*
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								 * Marker hash table, containing the active markers.
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								 * Protected by module_mutex.
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								 */
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								#define MARKER_HASH_BITS 6
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								#define MARKER_TABLE_SIZE (1 << MARKER_HASH_BITS)
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								
							 | 
						
					
						
							
								
									
										
											 
										 
										
											
												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>
											
										 
										
											2008-02-13 15:03:37 -08:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
								/*
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								 * Note about RCU :
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								 * It is used to make sure every handler has finished using its private data
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								 * between two consecutive operation (add or remove) on a given marker.  It is
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								 * also used to delay the free of multiple probes array until a quiescent state
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								 * is reached.
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								 * marker entries modifications are protected by the markers_mutex.
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								 */
							 | 
						
					
						
							
								
									
										
										
										
											2007-10-18 23:41:06 -07:00
										 
									 
								 
							 | 
							
								
							 | 
							
								
							 | 
							
							
								struct marker_entry {
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									struct hlist_node hlist;
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									char *format;
							 | 
						
					
						
							
								
									
										
											 
										 
										
											
												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 21:21:09 +02:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
											/* 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>
											
										 
										
											2008-02-13 15:03:37 -08:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
									struct marker_probe_closure single;
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									struct marker_probe_closure *multi;
							 | 
						
					
						
							
								
									
										
										
										
											2007-10-18 23:41:06 -07:00
										 
									 
								 
							 | 
							
								
							 | 
							
								
							 | 
							
							
									int refcount;	/* Number of times armed. 0 if disarmed. */
							 | 
						
					
						
							
								
									
										
											 
										 
										
											
												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>
											
										 
										
											2008-02-13 15:03:37 -08:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
									struct rcu_head rcu;
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									void *oldptr;
							 | 
						
					
						
							
								
									
										
										
										
											2008-02-23 15:23:33 -08:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
									unsigned char rcu_pending:1;
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									unsigned char ptype:1;
							 | 
						
					
						
							
								
									
										
										
										
											2007-10-18 23:41:06 -07:00
										 
									 
								 
							 | 
							
								
							 | 
							
								
							 | 
							
							
									char name[0];	/* Contains name'\0'format'\0' */
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								};
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								static struct hlist_head marker_table[MARKER_TABLE_SIZE];
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								/**
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								 * __mark_empty_function - Empty probe callback
							 | 
						
					
						
							
								
									
										
											 
										 
										
											
												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>
											
										 
										
											2008-02-13 15:03:37 -08:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
								 * @probe_private: probe private data
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								 * @call_private: call site private data
							 | 
						
					
						
							
								
									
										
										
										
											2007-10-18 23:41:06 -07:00
										 
									 
								 
							 | 
							
								
							 | 
							
								
							 | 
							
							
								 * @fmt: format string
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								 * @...: variable argument list
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								 *
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								 * Empty callback provided as a probe to the markers. By providing this to a
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								 * disabled marker, we make sure the  execution flow is always valid even
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								 * though the function pointer change and the marker enabling are two distinct
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								 * operations that modifies the execution flow of preemptible code.
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								 */
							 | 
						
					
						
							
								
									
										
											 
										 
										
											
												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>
											
										 
										
											2008-02-13 15:03:37 -08:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
								void __mark_empty_function(void *probe_private, void *call_private,
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									const char *fmt, va_list *args)
							 | 
						
					
						
							
								
									
										
										
										
											2007-10-18 23:41:06 -07:00
										 
									 
								 
							 | 
							
								
							 | 
							
								
							 | 
							
							
								{
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								}
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								EXPORT_SYMBOL_GPL(__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>
											
										 
										
											2008-02-13 15:03:37 -08:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
								/*
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								 * marker_probe_cb Callback that prepares the variable argument list for probes.
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								 * @mdata: pointer of type struct marker
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								 * @call_private: caller site private data
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								 * @...:  Variable argument list.
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								 *
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								 * Since we do not use "typical" pointer based RCU in the 1 argument case, we
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								 * need to put a full smp_rmb() in this branch. This is why we do not use
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								 * rcu_dereference() for the pointer read.
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								 */
							 | 
						
					
						
							
								
									
										
											 
										 
										
											
												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 21:21:09 +02:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
								void marker_probe_cb(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>
											
										 
										
											2008-02-13 15:03:37 -08:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
								{
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									va_list args;
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									char ptype;
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									/*
							 | 
						
					
						
							
								
									
										
										
										
											2008-03-24 12:29:47 -07:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
									 * preempt_disable does two things : disabling preemption to make sure
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									 * the teardown of the callbacks can be done correctly when they are in
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									 * modules and they insure RCU read coherency.
							 | 
						
					
						
							
								
									
										
											 
										 
										
											
												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>
											
										 
										
											2008-02-13 15:03:37 -08:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
									 */
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									preempt_disable();
							 | 
						
					
						
							
								
									
										
										
										
											2008-03-24 12:29:49 -07:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
									ptype = mdata->ptype;
							 | 
						
					
						
							
								
									
										
											 
										 
										
											
												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>
											
										 
										
											2008-02-13 15:03:37 -08:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
									if (likely(!ptype)) {
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
										marker_probe_func *func;
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
										/* Must read the ptype before ptr. They are not data dependant,
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
										 * so we put an explicit smp_rmb() here. */
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
										smp_rmb();
							 | 
						
					
						
							
								
									
										
										
										
											2008-03-24 12:29:49 -07:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
										func = mdata->single.func;
							 | 
						
					
						
							
								
									
										
											 
										 
										
											
												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>
											
										 
										
											2008-02-13 15:03:37 -08:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
										/* Must read the ptr before private data. They are not data
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
										 * dependant, so we put an explicit smp_rmb() here. */
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
										smp_rmb();
							 | 
						
					
						
							
								
									
										
											 
										 
										
											
												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 21:21:09 +02:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
										va_start(args, call_private);
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
										func(mdata->single.probe_private, call_private, mdata->format,
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
											&args);
							 | 
						
					
						
							
								
									
										
											 
										 
										
											
												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>
											
										 
										
											2008-02-13 15:03:37 -08:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
										va_end(args);
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									} else {
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
										struct marker_probe_closure *multi;
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
										int i;
							 | 
						
					
						
							
								
									
										
										
										
											2008-07-29 22:33:31 -07:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
										/*
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
										 * Read mdata->ptype before mdata->multi.
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
										 */
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
										smp_rmb();
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
										multi = mdata->multi;
							 | 
						
					
						
							
								
									
										
											 
										 
										
											
												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>
											
										 
										
											2008-02-13 15:03:37 -08:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
										/*
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
										 * multi points to an array, therefore accessing the array
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
										 * depends on reading multi. However, even in this case,
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
										 * we must insure that the pointer is read _before_ the array
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
										 * data. Same as rcu_dereference, but we need a full smp_rmb()
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
										 * in the fast path, so put the explicit barrier here.
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
										 */
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
										smp_read_barrier_depends();
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
										for (i = 0; multi[i].func; i++) {
							 | 
						
					
						
							
								
									
										
											 
										 
										
											
												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 21:21:09 +02:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
											va_start(args, call_private);
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
											multi[i].func(multi[i].probe_private, call_private,
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
												mdata->format, &args);
							 | 
						
					
						
							
								
									
										
											 
										 
										
											
												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>
											
										 
										
											2008-02-13 15:03:37 -08:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
											va_end(args);
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
										}
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									}
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									preempt_enable();
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								}
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								EXPORT_SYMBOL_GPL(marker_probe_cb);
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								/*
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								 * marker_probe_cb Callback that does not prepare the variable argument list.
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								 * @mdata: pointer of type struct marker
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								 * @call_private: caller site private data
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								 * @...:  Variable argument list.
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								 *
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								 * Should be connected to markers "MARK_NOARGS".
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								 */
							 | 
						
					
						
							
								
									
										
											 
										 
										
											
												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 21:21:09 +02:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
								void marker_probe_cb_noarg(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>
											
										 
										
											2008-02-13 15:03:37 -08:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
								{
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									va_list args;	/* not initialized */
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									char ptype;
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									preempt_disable();
							 | 
						
					
						
							
								
									
										
										
										
											2008-03-24 12:29:49 -07:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
									ptype = mdata->ptype;
							 | 
						
					
						
							
								
									
										
											 
										 
										
											
												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>
											
										 
										
											2008-02-13 15:03:37 -08:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
									if (likely(!ptype)) {
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
										marker_probe_func *func;
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
										/* Must read the ptype before ptr. They are not data dependant,
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
										 * so we put an explicit smp_rmb() here. */
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
										smp_rmb();
							 | 
						
					
						
							
								
									
										
										
										
											2008-03-24 12:29:49 -07:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
										func = mdata->single.func;
							 | 
						
					
						
							
								
									
										
											 
										 
										
											
												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>
											
										 
										
											2008-02-13 15:03:37 -08:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
										/* Must read the ptr before private data. They are not data
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
										 * dependant, so we put an explicit smp_rmb() here. */
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
										smp_rmb();
							 | 
						
					
						
							
								
									
										
											 
										 
										
											
												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 21:21:09 +02:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
										func(mdata->single.probe_private, call_private, mdata->format,
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
											&args);
							 | 
						
					
						
							
								
									
										
											 
										 
										
											
												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>
											
										 
										
											2008-02-13 15:03:37 -08:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
									} else {
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
										struct marker_probe_closure *multi;
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
										int i;
							 | 
						
					
						
							
								
									
										
										
										
											2008-07-29 22:33:31 -07:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
										/*
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
										 * Read mdata->ptype before mdata->multi.
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
										 */
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
										smp_rmb();
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
										multi = mdata->multi;
							 | 
						
					
						
							
								
									
										
											 
										 
										
											
												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>
											
										 
										
											2008-02-13 15:03:37 -08:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
										/*
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
										 * multi points to an array, therefore accessing the array
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
										 * depends on reading multi. However, even in this case,
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
										 * we must insure that the pointer is read _before_ the array
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
										 * data. Same as rcu_dereference, but we need a full smp_rmb()
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
										 * in the fast path, so put the explicit barrier here.
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
										 */
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
										smp_read_barrier_depends();
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
										for (i = 0; multi[i].func; i++)
							 | 
						
					
						
							
								
									
										
											 
										 
										
											
												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 21:21:09 +02:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
											multi[i].func(multi[i].probe_private, call_private,
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
												mdata->format, &args);
							 | 
						
					
						
							
								
									
										
											 
										 
										
											
												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>
											
										 
										
											2008-02-13 15:03:37 -08:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
									}
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									preempt_enable();
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								}
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								EXPORT_SYMBOL_GPL(marker_probe_cb_noarg);
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								static void free_old_closure(struct rcu_head *head)
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								{
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									struct marker_entry *entry = container_of(head,
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
										struct marker_entry, rcu);
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									kfree(entry->oldptr);
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									/* Make sure we free the data before setting the pending flag to 0 */
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									smp_wmb();
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									entry->rcu_pending = 0;
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								}
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								static void debug_print_probes(struct marker_entry *entry)
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								{
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									int i;
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									if (!marker_debug)
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
										return;
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									if (!entry->ptype) {
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
										printk(KERN_DEBUG "Single probe : %p %p\n",
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
											entry->single.func,
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
											entry->single.probe_private);
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									} else {
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
										for (i = 0; entry->multi[i].func; i++)
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
											printk(KERN_DEBUG "Multi probe %d : %p %p\n", i,
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
												entry->multi[i].func,
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
												entry->multi[i].probe_private);
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									}
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								}
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								static struct marker_probe_closure *
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								marker_entry_add_probe(struct marker_entry *entry,
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
										marker_probe_func *probe, void *probe_private)
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								{
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									int nr_probes = 0;
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									struct marker_probe_closure *old, *new;
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									WARN_ON(!probe);
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									debug_print_probes(entry);
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									old = entry->multi;
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									if (!entry->ptype) {
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
										if (entry->single.func == probe &&
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
												entry->single.probe_private == probe_private)
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
											return ERR_PTR(-EBUSY);
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
										if (entry->single.func == __mark_empty_function) {
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
											/* 0 -> 1 probes */
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
											entry->single.func = probe;
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
											entry->single.probe_private = probe_private;
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
											entry->refcount = 1;
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
											entry->ptype = 0;
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
											debug_print_probes(entry);
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
											return NULL;
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
										} else {
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
											/* 1 -> 2 probes */
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
											nr_probes = 1;
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
											old = NULL;
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
										}
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									} else {
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
										/* (N -> N+1), (N != 0, 1) probes */
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
										for (nr_probes = 0; old[nr_probes].func; nr_probes++)
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
											if (old[nr_probes].func == probe
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
													&& old[nr_probes].probe_private
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
														== probe_private)
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
												return ERR_PTR(-EBUSY);
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									}
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									/* + 2 : one for new probe, one for NULL func */
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									new = kzalloc((nr_probes + 2) * sizeof(struct marker_probe_closure),
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
											GFP_KERNEL);
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									if (new == NULL)
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
										return ERR_PTR(-ENOMEM);
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									if (!old)
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
										new[0] = entry->single;
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									else
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
										memcpy(new, old,
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
											nr_probes * sizeof(struct marker_probe_closure));
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									new[nr_probes].func = probe;
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									new[nr_probes].probe_private = probe_private;
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									entry->refcount = nr_probes + 1;
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									entry->multi = new;
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									entry->ptype = 1;
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									debug_print_probes(entry);
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									return old;
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								}
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								static struct marker_probe_closure *
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								marker_entry_remove_probe(struct marker_entry *entry,
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
										marker_probe_func *probe, void *probe_private)
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								{
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									int nr_probes = 0, nr_del = 0, i;
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									struct marker_probe_closure *old, *new;
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									old = entry->multi;
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									debug_print_probes(entry);
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									if (!entry->ptype) {
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
										/* 0 -> N is an error */
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
										WARN_ON(entry->single.func == __mark_empty_function);
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
										/* 1 -> 0 probes */
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
										WARN_ON(probe && entry->single.func != probe);
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
										WARN_ON(entry->single.probe_private != probe_private);
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
										entry->single.func = __mark_empty_function;
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
										entry->refcount = 0;
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
										entry->ptype = 0;
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
										debug_print_probes(entry);
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
										return NULL;
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									} else {
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
										/* (N -> M), (N > 1, M >= 0) probes */
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
										for (nr_probes = 0; old[nr_probes].func; nr_probes++) {
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
											if ((!probe || old[nr_probes].func == probe)
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
													&& old[nr_probes].probe_private
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
														== probe_private)
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
												nr_del++;
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
										}
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									}
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									if (nr_probes - nr_del == 0) {
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
										/* N -> 0, (N > 1) */
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
										entry->single.func = __mark_empty_function;
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
										entry->refcount = 0;
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
										entry->ptype = 0;
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									} else if (nr_probes - nr_del == 1) {
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
										/* N -> 1, (N > 1) */
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
										for (i = 0; old[i].func; i++)
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
											if ((probe && old[i].func != probe) ||
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
													old[i].probe_private != probe_private)
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
												entry->single = old[i];
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
										entry->refcount = 1;
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
										entry->ptype = 0;
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									} else {
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
										int j = 0;
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
										/* N -> M, (N > 1, M > 1) */
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
										/* + 1 for NULL */
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
										new = kzalloc((nr_probes - nr_del + 1)
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
											* sizeof(struct marker_probe_closure), GFP_KERNEL);
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
										if (new == NULL)
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
											return ERR_PTR(-ENOMEM);
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
										for (i = 0; old[i].func; i++)
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
											if ((probe && old[i].func != probe) ||
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
													old[i].probe_private != probe_private)
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
												new[j++] = old[i];
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
										entry->refcount = nr_probes - nr_del;
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
										entry->ptype = 1;
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
										entry->multi = new;
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									}
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									debug_print_probes(entry);
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									return old;
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								}
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								
							 | 
						
					
						
							
								
									
										
										
										
											2007-10-18 23:41:06 -07:00
										 
									 
								 
							 | 
							
								
							 | 
							
								
							 | 
							
							
								/*
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								 * Get marker if the marker is present in the marker hash table.
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								 * Must be called with markers_mutex held.
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								 * Returns NULL if not present.
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								 */
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								static struct marker_entry *get_marker(const char *name)
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								{
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									struct hlist_head *head;
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									struct hlist_node *node;
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									struct marker_entry *e;
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									u32 hash = jhash(name, strlen(name), 0);
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									head = &marker_table[hash & ((1 << MARKER_HASH_BITS)-1)];
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									hlist_for_each_entry(e, node, head, hlist) {
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
										if (!strcmp(name, e->name))
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
											return e;
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									}
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									return NULL;
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								}
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								/*
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								 * Add the marker to the marker hash table. Must be called with markers_mutex
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								 * held.
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								 */
							 | 
						
					
						
							
								
									
										
											 
										 
										
											
												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>
											
										 
										
											2008-02-13 15:03:37 -08:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
								static struct marker_entry *add_marker(const char *name, const char *format)
							 | 
						
					
						
							
								
									
										
										
										
											2007-10-18 23:41:06 -07:00
										 
									 
								 
							 | 
							
								
							 | 
							
								
							 | 
							
							
								{
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									struct hlist_head *head;
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									struct hlist_node *node;
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									struct marker_entry *e;
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									size_t name_len = strlen(name) + 1;
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									size_t format_len = 0;
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									u32 hash = jhash(name, name_len-1, 0);
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									if (format)
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
										format_len = strlen(format) + 1;
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									head = &marker_table[hash & ((1 << MARKER_HASH_BITS)-1)];
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									hlist_for_each_entry(e, node, head, hlist) {
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
										if (!strcmp(name, e->name)) {
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
											printk(KERN_NOTICE
							 | 
						
					
						
							
								
									
										
											 
										 
										
											
												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>
											
										 
										
											2008-02-13 15:03:37 -08:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
												"Marker %s busy\n", name);
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
											return ERR_PTR(-EBUSY);	/* Already there */
							 | 
						
					
						
							
								
									
										
										
										
											2007-10-18 23:41:06 -07:00
										 
									 
								 
							 | 
							
								
							 | 
							
								
							 | 
							
							
										}
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									}
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									/*
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									 * Using kmalloc here to allocate a variable length element. Could
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									 * cause some memory fragmentation if overused.
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									 */
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									e = kmalloc(sizeof(struct marker_entry) + name_len + format_len,
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
											GFP_KERNEL);
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									if (!e)
							 | 
						
					
						
							
								
									
										
											 
										 
										
											
												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>
											
										 
										
											2008-02-13 15:03:37 -08:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
										return ERR_PTR(-ENOMEM);
							 | 
						
					
						
							
								
									
										
										
										
											2007-10-18 23:41:06 -07:00
										 
									 
								 
							 | 
							
								
							 | 
							
								
							 | 
							
							
									memcpy(&e->name[0], name, name_len);
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									if (format) {
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
										e->format = &e->name[name_len];
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
										memcpy(e->format, format, format_len);
							 | 
						
					
						
							
								
									
										
											 
										 
										
											
												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>
											
										 
										
											2008-02-13 15:03:37 -08:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
										if (strcmp(e->format, MARK_NOARGS) == 0)
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
											e->call = marker_probe_cb_noarg;
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
										else
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
											e->call = marker_probe_cb;
							 | 
						
					
						
							
								
									
										
										
										
											2007-10-18 23:41:06 -07:00
										 
									 
								 
							 | 
							
								
							 | 
							
								
							 | 
							
							
										trace_mark(core_marker_format, "name %s format %s",
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
												e->name, e->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>
											
										 
										
											2008-02-13 15:03:37 -08:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
									} else {
							 | 
						
					
						
							
								
									
										
										
										
											2007-10-18 23:41:06 -07:00
										 
									 
								 
							 | 
							
								
							 | 
							
								
							 | 
							
							
										e->format = NULL;
							 | 
						
					
						
							
								
									
										
											 
										 
										
											
												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>
											
										 
										
											2008-02-13 15:03:37 -08:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
										e->call = marker_probe_cb;
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									}
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									e->single.func = __mark_empty_function;
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									e->single.probe_private = NULL;
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									e->multi = NULL;
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									e->ptype = 0;
							 | 
						
					
						
							
								
									
										
										
										
											2007-10-18 23:41:06 -07:00
										 
									 
								 
							 | 
							
								
							 | 
							
								
							 | 
							
							
									e->refcount = 0;
							 | 
						
					
						
							
								
									
										
											 
										 
										
											
												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>
											
										 
										
											2008-02-13 15:03:37 -08:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
									e->rcu_pending = 0;
							 | 
						
					
						
							
								
									
										
										
										
											2007-10-18 23:41:06 -07:00
										 
									 
								 
							 | 
							
								
							 | 
							
								
							 | 
							
							
									hlist_add_head(&e->hlist, head);
							 | 
						
					
						
							
								
									
										
											 
										 
										
											
												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>
											
										 
										
											2008-02-13 15:03:37 -08:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
									return e;
							 | 
						
					
						
							
								
									
										
										
										
											2007-10-18 23:41:06 -07:00
										 
									 
								 
							 | 
							
								
							 | 
							
								
							 | 
							
							
								}
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								/*
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								 * Remove the marker from the marker hash table. Must be called with mutex_lock
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								 * held.
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								 */
							 | 
						
					
						
							
								
									
										
											 
										 
										
											
												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>
											
										 
										
											2008-02-13 15:03:37 -08:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
								static int remove_marker(const char *name)
							 | 
						
					
						
							
								
									
										
										
										
											2007-10-18 23:41:06 -07:00
										 
									 
								 
							 | 
							
								
							 | 
							
								
							 | 
							
							
								{
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									struct hlist_head *head;
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									struct hlist_node *node;
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									struct marker_entry *e;
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									int found = 0;
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									size_t len = strlen(name) + 1;
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									u32 hash = jhash(name, len-1, 0);
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									head = &marker_table[hash & ((1 << MARKER_HASH_BITS)-1)];
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									hlist_for_each_entry(e, node, head, hlist) {
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
										if (!strcmp(name, e->name)) {
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
											found = 1;
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
											break;
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
										}
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									}
							 | 
						
					
						
							
								
									
										
											 
										 
										
											
												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>
											
										 
										
											2008-02-13 15:03:37 -08:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
									if (!found)
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
										return -ENOENT;
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									if (e->single.func != __mark_empty_function)
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
										return -EBUSY;
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									hlist_del(&e->hlist);
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									/* Make sure the call_rcu has been executed */
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									if (e->rcu_pending)
							 | 
						
					
						
							
								
									
										
										
										
											2008-07-25 01:48:38 -07:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
										rcu_barrier_sched();
							 | 
						
					
						
							
								
									
										
											 
										 
										
											
												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>
											
										 
										
											2008-02-13 15:03:37 -08:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
									kfree(e);
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									return 0;
							 | 
						
					
						
							
								
									
										
										
										
											2007-10-18 23:41:06 -07:00
										 
									 
								 
							 | 
							
								
							 | 
							
								
							 | 
							
							
								}
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								/*
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								 * Set the mark_entry format to the format found in the element.
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								 */
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								static int marker_set_format(struct marker_entry **entry, const char *format)
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								{
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									struct marker_entry *e;
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									size_t name_len = strlen((*entry)->name) + 1;
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									size_t format_len = strlen(format) + 1;
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								
							 | 
						
					
						
							
								
									
										
											 
										 
										
											
												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>
											
										 
										
											2008-02-13 15:03:37 -08:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
								
							 | 
						
					
						
							
								
									
										
										
										
											2007-10-18 23:41:06 -07:00
										 
									 
								 
							 | 
							
								
							 | 
							
								
							 | 
							
							
									e = kmalloc(sizeof(struct marker_entry) + name_len + format_len,
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
											GFP_KERNEL);
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									if (!e)
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
										return -ENOMEM;
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									memcpy(&e->name[0], (*entry)->name, name_len);
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									e->format = &e->name[name_len];
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									memcpy(e->format, format, format_len);
							 | 
						
					
						
							
								
									
										
											 
										 
										
											
												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>
											
										 
										
											2008-02-13 15:03:37 -08:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
									if (strcmp(e->format, MARK_NOARGS) == 0)
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
										e->call = marker_probe_cb_noarg;
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									else
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
										e->call = marker_probe_cb;
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									e->single = (*entry)->single;
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									e->multi = (*entry)->multi;
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									e->ptype = (*entry)->ptype;
							 | 
						
					
						
							
								
									
										
										
										
											2007-10-18 23:41:06 -07:00
										 
									 
								 
							 | 
							
								
							 | 
							
								
							 | 
							
							
									e->refcount = (*entry)->refcount;
							 | 
						
					
						
							
								
									
										
											 
										 
										
											
												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>
											
										 
										
											2008-02-13 15:03:37 -08:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
									e->rcu_pending = 0;
							 | 
						
					
						
							
								
									
										
										
										
											2007-10-18 23:41:06 -07:00
										 
									 
								 
							 | 
							
								
							 | 
							
								
							 | 
							
							
									hlist_add_before(&e->hlist, &(*entry)->hlist);
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									hlist_del(&(*entry)->hlist);
							 | 
						
					
						
							
								
									
										
											 
										 
										
											
												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>
											
										 
										
											2008-02-13 15:03:37 -08:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
									/* Make sure the call_rcu has been executed */
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									if ((*entry)->rcu_pending)
							 | 
						
					
						
							
								
									
										
										
										
											2008-07-25 01:48:38 -07:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
										rcu_barrier_sched();
							 | 
						
					
						
							
								
									
										
										
										
											2007-10-18 23:41:06 -07:00
										 
									 
								 
							 | 
							
								
							 | 
							
								
							 | 
							
							
									kfree(*entry);
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									*entry = e;
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									trace_mark(core_marker_format, "name %s format %s",
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
											e->name, e->format);
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									return 0;
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								}
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								/*
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								 * Sets the probe callback corresponding to one marker.
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								 */
							 | 
						
					
						
							
								
									
										
											 
										 
										
											
												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>
											
										 
										
											2008-02-13 15:03:37 -08:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
								static int set_marker(struct marker_entry **entry, struct marker *elem,
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
										int active)
							 | 
						
					
						
							
								
									
										
										
										
											2007-10-18 23:41:06 -07:00
										 
									 
								 
							 | 
							
								
							 | 
							
								
							 | 
							
							
								{
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									int ret;
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									WARN_ON(strcmp((*entry)->name, elem->name) != 0);
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									if ((*entry)->format) {
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
										if (strcmp((*entry)->format, elem->format) != 0) {
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
											printk(KERN_NOTICE
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
												"Format mismatch for probe %s "
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
												"(%s), marker (%s)\n",
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
												(*entry)->name,
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
												(*entry)->format,
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
												elem->format);
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
											return -EPERM;
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
										}
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									} else {
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
										ret = marker_set_format(entry, elem->format);
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
										if (ret)
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
											return ret;
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									}
							 | 
						
					
						
							
								
									
										
											 
										 
										
											
												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>
											
										 
										
											2008-02-13 15:03:37 -08:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
								
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									/*
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									 * probe_cb setup (statically known) is done here. It is
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									 * asynchronous with the rest of execution, therefore we only
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									 * pass from a "safe" callback (with argument) to an "unsafe"
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									 * callback (does not set arguments).
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									 */
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									elem->call = (*entry)->call;
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									/*
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									 * Sanity check :
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									 * We only update the single probe private data when the ptr is
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									 * set to a _non_ single probe! (0 -> 1 and N -> 1, N != 1)
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									 */
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									WARN_ON(elem->single.func != __mark_empty_function
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
										&& elem->single.probe_private
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
										!= (*entry)->single.probe_private &&
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
										!elem->ptype);
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									elem->single.probe_private = (*entry)->single.probe_private;
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									/*
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									 * Make sure the private data is valid when we update the
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									 * single probe ptr.
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									 */
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									smp_wmb();
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									elem->single.func = (*entry)->single.func;
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									/*
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									 * We also make sure that the new probe callbacks array is consistent
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									 * before setting a pointer to it.
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									 */
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									rcu_assign_pointer(elem->multi, (*entry)->multi);
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									/*
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									 * Update the function or multi probe array pointer before setting the
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									 * ptype.
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									 */
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									smp_wmb();
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									elem->ptype = (*entry)->ptype;
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									elem->state = active;
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								
							 | 
						
					
						
							
								
									
										
										
										
											2007-10-18 23:41:06 -07:00
										 
									 
								 
							 | 
							
								
							 | 
							
								
							 | 
							
							
									return 0;
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								}
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								/*
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								 * Disable a marker and its probe callback.
							 | 
						
					
						
							
								
									
										
										
										
											2008-03-24 12:29:47 -07:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
								 * Note: only waiting an RCU period after setting elem->call to the empty
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								 * function insures that the original callback is not used anymore. This insured
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								 * by preempt_disable around the call site.
							 | 
						
					
						
							
								
									
										
										
										
											2007-10-18 23:41:06 -07:00
										 
									 
								 
							 | 
							
								
							 | 
							
								
							 | 
							
							
								 */
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								static void disable_marker(struct marker *elem)
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								{
							 | 
						
					
						
							
								
									
										
											 
										 
										
											
												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>
											
										 
										
											2008-02-13 15:03:37 -08:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
									/* leave "call" as is. It is known statically. */
							 | 
						
					
						
							
								
									
										
										
										
											2007-10-18 23:41:06 -07:00
										 
									 
								 
							 | 
							
								
							 | 
							
								
							 | 
							
							
									elem->state = 0;
							 | 
						
					
						
							
								
									
										
											 
										 
										
											
												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>
											
										 
										
											2008-02-13 15:03:37 -08:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
									elem->single.func = __mark_empty_function;
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									/* Update the function before setting the ptype */
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									smp_wmb();
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									elem->ptype = 0;	/* single probe */
							 | 
						
					
						
							
								
									
										
										
										
											2007-10-18 23:41:06 -07:00
										 
									 
								 
							 | 
							
								
							 | 
							
								
							 | 
							
							
									/*
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									 * Leave the private data and id there, because removal is racy and
							 | 
						
					
						
							
								
									
										
										
										
											2008-03-24 12:29:47 -07:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
									 * should be done only after an RCU period. These are never used until
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									 * the next initialization anyway.
							 | 
						
					
						
							
								
									
										
										
										
											2007-10-18 23:41:06 -07:00
										 
									 
								 
							 | 
							
								
							 | 
							
								
							 | 
							
							
									 */
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								}
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								/**
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								 * marker_update_probe_range - Update a probe range
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								 * @begin: beginning of the range
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								 * @end: end of the range
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								 *
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								 * Updates the probe callback corresponding to a range of markers.
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								 */
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								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>
											
										 
										
											2008-02-13 15:03:37 -08:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
									struct marker *end)
							 | 
						
					
						
							
								
									
										
										
										
											2007-10-18 23:41:06 -07:00
										 
									 
								 
							 | 
							
								
							 | 
							
								
							 | 
							
							
								{
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									struct marker *iter;
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									struct marker_entry *mark_entry;
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								
							 | 
						
					
						
							
								
									
										
										
										
											2007-11-14 16:59:48 -08:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
									mutex_lock(&markers_mutex);
							 | 
						
					
						
							
								
									
										
										
										
											2007-10-18 23:41:06 -07:00
										 
									 
								 
							 | 
							
								
							 | 
							
								
							 | 
							
							
									for (iter = begin; iter < end; iter++) {
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
										mark_entry = get_marker(iter->name);
							 | 
						
					
						
							
								
									
										
											 
										 
										
											
												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>
											
										 
										
											2008-02-13 15:03:37 -08:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
										if (mark_entry) {
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
											set_marker(&mark_entry, iter,
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
													!!mark_entry->refcount);
							 | 
						
					
						
							
								
									
										
										
										
											2007-10-18 23:41:06 -07:00
										 
									 
								 
							 | 
							
								
							 | 
							
								
							 | 
							
							
											/*
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
											 * ignore error, continue
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
											 */
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
										} else {
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
											disable_marker(iter);
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
										}
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									}
							 | 
						
					
						
							
								
									
										
										
										
											2007-11-14 16:59:48 -08:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
									mutex_unlock(&markers_mutex);
							 | 
						
					
						
							
								
									
										
										
										
											2007-10-18 23:41:06 -07:00
										 
									 
								 
							 | 
							
								
							 | 
							
								
							 | 
							
							
								}
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								/*
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								 * Update probes, removing the faulty probes.
							 | 
						
					
						
							
								
									
										
											 
										 
										
											
												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>
											
										 
										
											2008-02-13 15:03:37 -08:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
								 *
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								 * Internal callback only changed before the first probe is connected to it.
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								 * Single probe private data can only be changed on 0 -> 1 and 2 -> 1
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								 * transitions.  All other transitions will leave the old private data valid.
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								 * This makes the non-atomicity of the callback/private data updates valid.
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								 *
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								 * "special case" updates :
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								 * 0 -> 1 callback
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								 * 1 -> 0 callback
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								 * 1 -> 2 callbacks
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								 * 2 -> 1 callbacks
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								 * Other updates all behave the same, just like the 2 -> 3 or 3 -> 2 updates.
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								 * Site effect : marker_set_format may delete the marker entry (creating a
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								 * replacement).
							 | 
						
					
						
							
								
									
										
										
										
											2007-10-18 23:41:06 -07:00
										 
									 
								 
							 | 
							
								
							 | 
							
								
							 | 
							
							
								 */
							 | 
						
					
						
							
								
									
										
											 
										 
										
											
												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>
											
										 
										
											2008-02-13 15:03:37 -08:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
								static void marker_update_probes(void)
							 | 
						
					
						
							
								
									
										
										
										
											2007-10-18 23:41:06 -07:00
										 
									 
								 
							 | 
							
								
							 | 
							
								
							 | 
							
							
								{
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									/* Core kernel markers */
							 | 
						
					
						
							
								
									
										
											 
										 
										
											
												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>
											
										 
										
											2008-02-13 15:03:37 -08:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
									marker_update_probe_range(__start___markers, __stop___markers);
							 | 
						
					
						
							
								
									
										
										
										
											2007-10-18 23:41:06 -07:00
										 
									 
								 
							 | 
							
								
							 | 
							
								
							 | 
							
							
									/* Markers in modules. */
							 | 
						
					
						
							
								
									
										
											 
										 
										
											
												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>
											
										 
										
											2008-02-13 15:03:37 -08:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
									module_update_markers();
							 | 
						
					
						
							
								
									
										
										
										
											2007-10-18 23:41:06 -07:00
										 
									 
								 
							 | 
							
								
							 | 
							
								
							 | 
							
							
								}
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								/**
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								 * marker_probe_register -  Connect a probe to a marker
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								 * @name: marker name
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								 * @format: format string
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								 * @probe: probe handler
							 | 
						
					
						
							
								
									
										
											 
										 
										
											
												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>
											
										 
										
											2008-02-13 15:03:37 -08:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
								 * @probe_private: probe private data
							 | 
						
					
						
							
								
									
										
										
										
											2007-10-18 23:41:06 -07:00
										 
									 
								 
							 | 
							
								
							 | 
							
								
							 | 
							
							
								 *
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								 * private data must be a valid allocated memory address, or NULL.
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								 * Returns 0 if ok, error value on error.
							 | 
						
					
						
							
								
									
										
											 
										 
										
											
												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>
											
										 
										
											2008-02-13 15:03:37 -08:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
								 * The probe address must at least be aligned on the architecture pointer size.
							 | 
						
					
						
							
								
									
										
										
										
											2007-10-18 23:41:06 -07:00
										 
									 
								 
							 | 
							
								
							 | 
							
								
							 | 
							
							
								 */
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								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>
											
										 
										
											2008-02-13 15:03:37 -08:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
											marker_probe_func *probe, void *probe_private)
							 | 
						
					
						
							
								
									
										
										
										
											2007-10-18 23:41:06 -07:00
										 
									 
								 
							 | 
							
								
							 | 
							
								
							 | 
							
							
								{
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									struct marker_entry *entry;
							 | 
						
					
						
							
								
									
										
										
										
											2007-11-14 16:59:48 -08:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
									int ret = 0;
							 | 
						
					
						
							
								
									
										
											 
										 
										
											
												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>
											
										 
										
											2008-02-13 15:03:37 -08:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
									struct marker_probe_closure *old;
							 | 
						
					
						
							
								
									
										
										
										
											2007-10-18 23:41:06 -07:00
										 
									 
								 
							 | 
							
								
							 | 
							
								
							 | 
							
							
								
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									mutex_lock(&markers_mutex);
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									entry = get_marker(name);
							 | 
						
					
						
							
								
									
										
											 
										 
										
											
												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>
											
										 
										
											2008-02-13 15:03:37 -08:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
									if (!entry) {
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
										entry = add_marker(name, format);
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
										if (IS_ERR(entry)) {
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
											ret = PTR_ERR(entry);
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
											goto end;
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
										}
							 | 
						
					
						
							
								
									
										
										
										
											2007-10-18 23:41:06 -07:00
										 
									 
								 
							 | 
							
								
							 | 
							
								
							 | 
							
							
									}
							 | 
						
					
						
							
								
									
										
											 
										 
										
											
												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>
											
										 
										
											2008-02-13 15:03:37 -08:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
									/*
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									 * If we detect that a call_rcu is pending for this marker,
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									 * make sure it's executed now.
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									 */
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									if (entry->rcu_pending)
							 | 
						
					
						
							
								
									
										
										
										
											2008-07-25 01:48:38 -07:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
										rcu_barrier_sched();
							 | 
						
					
						
							
								
									
										
											 
										 
										
											
												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>
											
										 
										
											2008-02-13 15:03:37 -08:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
									old = marker_entry_add_probe(entry, probe, probe_private);
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									if (IS_ERR(old)) {
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
										ret = PTR_ERR(old);
							 | 
						
					
						
							
								
									
										
										
										
											2007-10-18 23:41:06 -07:00
										 
									 
								 
							 | 
							
								
							 | 
							
								
							 | 
							
							
										goto end;
							 | 
						
					
						
							
								
									
										
											 
										 
										
											
												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>
											
										 
										
											2008-02-13 15:03:37 -08:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
									}
							 | 
						
					
						
							
								
									
										
										
										
											2007-11-14 16:59:48 -08:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
									mutex_unlock(&markers_mutex);
							 | 
						
					
						
							
								
									
										
											 
										 
										
											
												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>
											
										 
										
											2008-02-13 15:03:37 -08:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
									marker_update_probes();		/* may update entry */
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									mutex_lock(&markers_mutex);
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									entry = get_marker(name);
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									WARN_ON(!entry);
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									entry->oldptr = old;
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									entry->rcu_pending = 1;
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									/* write rcu_pending before calling the RCU callback */
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									smp_wmb();
							 | 
						
					
						
							
								
									
										
										
										
											2008-07-25 01:48:38 -07:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
									call_rcu_sched(&entry->rcu, free_old_closure);
							 | 
						
					
						
							
								
									
										
										
										
											2007-10-18 23:41:06 -07:00
										 
									 
								 
							 | 
							
								
							 | 
							
								
							 | 
							
							
								end:
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									mutex_unlock(&markers_mutex);
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									return ret;
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								}
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								EXPORT_SYMBOL_GPL(marker_probe_register);
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								/**
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								 * marker_probe_unregister -  Disconnect a probe from a marker
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								 * @name: marker name
							 | 
						
					
						
							
								
									
										
											 
										 
										
											
												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>
											
										 
										
											2008-02-13 15:03:37 -08:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
								 * @probe: probe function pointer
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								 * @probe_private: probe private data
							 | 
						
					
						
							
								
									
										
										
										
											2007-10-18 23:41:06 -07:00
										 
									 
								 
							 | 
							
								
							 | 
							
								
							 | 
							
							
								 *
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								 * Returns the private data given to marker_probe_register, or an ERR_PTR().
							 | 
						
					
						
							
								
									
										
											 
										 
										
											
												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>
											
										 
										
											2008-02-13 15:03:37 -08:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
								 * We do not need to call a synchronize_sched to make sure the probes have
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								 * finished running before doing a module unload, because the module unload
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								 * itself uses stop_machine(), which insures that every preempt disabled section
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								 * have finished.
							 | 
						
					
						
							
								
									
										
										
										
											2007-10-18 23:41:06 -07:00
										 
									 
								 
							 | 
							
								
							 | 
							
								
							 | 
							
							
								 */
							 | 
						
					
						
							
								
									
										
											 
										 
										
											
												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>
											
										 
										
											2008-02-13 15:03:37 -08:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
								int marker_probe_unregister(const char *name,
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									marker_probe_func *probe, void *probe_private)
							 | 
						
					
						
							
								
									
										
										
										
											2007-10-18 23:41:06 -07:00
										 
									 
								 
							 | 
							
								
							 | 
							
								
							 | 
							
							
								{
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									struct marker_entry *entry;
							 | 
						
					
						
							
								
									
										
											 
										 
										
											
												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>
											
										 
										
											2008-02-13 15:03:37 -08:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
									struct marker_probe_closure *old;
							 | 
						
					
						
							
								
									
										
										
										
											2008-03-04 14:29:00 -08:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
									int ret = -ENOENT;
							 | 
						
					
						
							
								
									
										
										
										
											2007-10-18 23:41:06 -07:00
										 
									 
								 
							 | 
							
								
							 | 
							
								
							 | 
							
							
								
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									mutex_lock(&markers_mutex);
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									entry = get_marker(name);
							 | 
						
					
						
							
								
									
										
										
										
											2008-03-04 14:29:00 -08:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
									if (!entry)
							 | 
						
					
						
							
								
									
										
										
										
											2007-10-18 23:41:06 -07:00
										 
									 
								 
							 | 
							
								
							 | 
							
								
							 | 
							
							
										goto end;
							 | 
						
					
						
							
								
									
										
											 
										 
										
											
												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>
											
										 
										
											2008-02-13 15:03:37 -08:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
									if (entry->rcu_pending)
							 | 
						
					
						
							
								
									
										
										
										
											2008-07-25 01:48:38 -07:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
										rcu_barrier_sched();
							 | 
						
					
						
							
								
									
										
											 
										 
										
											
												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>
											
										 
										
											2008-02-13 15:03:37 -08:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
									old = marker_entry_remove_probe(entry, probe, probe_private);
							 | 
						
					
						
							
								
									
										
										
										
											2007-11-14 16:59:48 -08:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
									mutex_unlock(&markers_mutex);
							 | 
						
					
						
							
								
									
										
											 
										 
										
											
												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>
											
										 
										
											2008-02-13 15:03:37 -08:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
									marker_update_probes();		/* may update entry */
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									mutex_lock(&markers_mutex);
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									entry = get_marker(name);
							 | 
						
					
						
							
								
									
										
										
										
											2008-03-04 14:29:00 -08:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
									if (!entry)
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
										goto end;
							 | 
						
					
						
							
								
									
										
											 
										 
										
											
												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>
											
										 
										
											2008-02-13 15:03:37 -08:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
									entry->oldptr = old;
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									entry->rcu_pending = 1;
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									/* write rcu_pending before calling the RCU callback */
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									smp_wmb();
							 | 
						
					
						
							
								
									
										
										
										
											2008-07-25 01:48:38 -07:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
									call_rcu_sched(&entry->rcu, free_old_closure);
							 | 
						
					
						
							
								
									
										
											 
										 
										
											
												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>
											
										 
										
											2008-02-13 15:03:37 -08:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
									remove_marker(name);	/* Ignore busy error message */
							 | 
						
					
						
							
								
									
										
										
										
											2008-03-04 14:29:00 -08:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
									ret = 0;
							 | 
						
					
						
							
								
									
										
										
										
											2007-10-18 23:41:06 -07:00
										 
									 
								 
							 | 
							
								
							 | 
							
								
							 | 
							
							
								end:
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									mutex_unlock(&markers_mutex);
							 | 
						
					
						
							
								
									
										
											 
										 
										
											
												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>
											
										 
										
											2008-02-13 15:03:37 -08:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
									return ret;
							 | 
						
					
						
							
								
									
										
										
										
											2007-10-18 23:41:06 -07:00
										 
									 
								 
							 | 
							
								
							 | 
							
								
							 | 
							
							
								}
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								EXPORT_SYMBOL_GPL(marker_probe_unregister);
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								
							 | 
						
					
						
							
								
									
										
											 
										 
										
											
												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>
											
										 
										
											2008-02-13 15:03:37 -08:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
								static struct marker_entry *
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								get_marker_from_private_data(marker_probe_func *probe, void *probe_private)
							 | 
						
					
						
							
								
									
										
										
										
											2007-10-18 23:41:06 -07:00
										 
									 
								 
							 | 
							
								
							 | 
							
								
							 | 
							
							
								{
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									struct marker_entry *entry;
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									unsigned int i;
							 | 
						
					
						
							
								
									
										
											 
										 
										
											
												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>
											
										 
										
											2008-02-13 15:03:37 -08:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
									struct hlist_head *head;
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									struct hlist_node *node;
							 | 
						
					
						
							
								
									
										
										
										
											2007-10-18 23:41:06 -07:00
										 
									 
								 
							 | 
							
								
							 | 
							
								
							 | 
							
							
								
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									for (i = 0; i < MARKER_TABLE_SIZE; i++) {
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
										head = &marker_table[i];
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
										hlist_for_each_entry(entry, node, head, hlist) {
							 | 
						
					
						
							
								
									
										
											 
										 
										
											
												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>
											
										 
										
											2008-02-13 15:03:37 -08:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
											if (!entry->ptype) {
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
												if (entry->single.func == probe
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
														&& entry->single.probe_private
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
														== probe_private)
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
													return entry;
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
											} else {
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
												struct marker_probe_closure *closure;
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
												closure = entry->multi;
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
												for (i = 0; closure[i].func; i++) {
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
													if (closure[i].func == probe &&
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
															closure[i].probe_private
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
															== probe_private)
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
														return entry;
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
												}
							 | 
						
					
						
							
								
									
										
										
										
											2007-10-18 23:41:06 -07:00
										 
									 
								 
							 | 
							
								
							 | 
							
								
							 | 
							
							
											}
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
										}
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									}
							 | 
						
					
						
							
								
									
										
											 
										 
										
											
												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>
											
										 
										
											2008-02-13 15:03:37 -08:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
									return NULL;
							 | 
						
					
						
							
								
									
										
										
										
											2007-10-18 23:41:06 -07:00
										 
									 
								 
							 | 
							
								
							 | 
							
								
							 | 
							
							
								}
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								/**
							 | 
						
					
						
							
								
									
										
											 
										 
										
											
												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>
											
										 
										
											2008-02-13 15:03:37 -08:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
								 * marker_probe_unregister_private_data -  Disconnect a probe from a marker
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								 * @probe: probe function
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								 * @probe_private: probe private data
							 | 
						
					
						
							
								
									
										
										
										
											2007-10-18 23:41:06 -07:00
										 
									 
								 
							 | 
							
								
							 | 
							
								
							 | 
							
							
								 *
							 | 
						
					
						
							
								
									
										
											 
										 
										
											
												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>
											
										 
										
											2008-02-13 15:03:37 -08:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
								 * Unregister a probe by providing the registered private data.
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								 * Only removes the first marker found in hash table.
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								 * Return 0 on success or error value.
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								 * We do not need to call a synchronize_sched to make sure the probes have
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								 * finished running before doing a module unload, because the module unload
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								 * itself uses stop_machine(), which insures that every preempt disabled section
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								 * have finished.
							 | 
						
					
						
							
								
									
										
										
										
											2007-10-18 23:41:06 -07:00
										 
									 
								 
							 | 
							
								
							 | 
							
								
							 | 
							
							
								 */
							 | 
						
					
						
							
								
									
										
											 
										 
										
											
												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>
											
										 
										
											2008-02-13 15:03:37 -08:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
								int marker_probe_unregister_private_data(marker_probe_func *probe,
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
										void *probe_private)
							 | 
						
					
						
							
								
									
										
										
										
											2007-10-18 23:41:06 -07:00
										 
									 
								 
							 | 
							
								
							 | 
							
								
							 | 
							
							
								{
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									struct marker_entry *entry;
							 | 
						
					
						
							
								
									
										
										
										
											2007-11-14 16:59:48 -08:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
									int ret = 0;
							 | 
						
					
						
							
								
									
										
											 
										 
										
											
												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>
											
										 
										
											2008-02-13 15:03:37 -08:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
									struct marker_probe_closure *old;
							 | 
						
					
						
							
								
									
										
										
										
											2007-10-18 23:41:06 -07:00
										 
									 
								 
							 | 
							
								
							 | 
							
								
							 | 
							
							
								
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									mutex_lock(&markers_mutex);
							 | 
						
					
						
							
								
									
										
											 
										 
										
											
												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>
											
										 
										
											2008-02-13 15:03:37 -08:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
									entry = get_marker_from_private_data(probe, probe_private);
							 | 
						
					
						
							
								
									
										
										
										
											2007-10-18 23:41:06 -07:00
										 
									 
								 
							 | 
							
								
							 | 
							
								
							 | 
							
							
									if (!entry) {
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
										ret = -ENOENT;
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
										goto end;
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									}
							 | 
						
					
						
							
								
									
										
											 
										 
										
											
												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>
											
										 
										
											2008-02-13 15:03:37 -08:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
									if (entry->rcu_pending)
							 | 
						
					
						
							
								
									
										
										
										
											2008-07-25 01:48:38 -07:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
										rcu_barrier_sched();
							 | 
						
					
						
							
								
									
										
											 
										 
										
											
												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>
											
										 
										
											2008-02-13 15:03:37 -08:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
									old = marker_entry_remove_probe(entry, NULL, probe_private);
							 | 
						
					
						
							
								
									
										
										
										
											2007-10-18 23:41:06 -07:00
										 
									 
								 
							 | 
							
								
							 | 
							
								
							 | 
							
							
									mutex_unlock(&markers_mutex);
							 | 
						
					
						
							
								
									
										
											 
										 
										
											
												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>
											
										 
										
											2008-02-13 15:03:37 -08:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
									marker_update_probes();		/* may update entry */
							 | 
						
					
						
							
								
									
										
										
										
											2007-10-18 23:41:06 -07:00
										 
									 
								 
							 | 
							
								
							 | 
							
								
							 | 
							
							
									mutex_lock(&markers_mutex);
							 | 
						
					
						
							
								
									
										
											 
										 
										
											
												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>
											
										 
										
											2008-02-13 15:03:37 -08:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
									entry = get_marker_from_private_data(probe, probe_private);
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									WARN_ON(!entry);
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									entry->oldptr = old;
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									entry->rcu_pending = 1;
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									/* write rcu_pending before calling the RCU callback */
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									smp_wmb();
							 | 
						
					
						
							
								
									
										
										
										
											2008-07-25 01:48:38 -07:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
									call_rcu_sched(&entry->rcu, free_old_closure);
							 | 
						
					
						
							
								
									
										
											 
										 
										
											
												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>
											
										 
										
											2008-02-13 15:03:37 -08:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
									remove_marker(entry->name);	/* Ignore busy error message */
							 | 
						
					
						
							
								
									
										
										
										
											2007-10-18 23:41:06 -07:00
										 
									 
								 
							 | 
							
								
							 | 
							
								
							 | 
							
							
								end:
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									mutex_unlock(&markers_mutex);
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									return ret;
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								}
							 | 
						
					
						
							
								
									
										
											 
										 
										
											
												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>
											
										 
										
											2008-02-13 15:03:37 -08:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
								EXPORT_SYMBOL_GPL(marker_probe_unregister_private_data);
							 | 
						
					
						
							
								
									
										
										
										
											2007-10-18 23:41:06 -07:00
										 
									 
								 
							 | 
							
								
							 | 
							
								
							 | 
							
							
								
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								/**
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								 * marker_get_private_data - Get a marker's probe private data
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								 * @name: marker name
							 | 
						
					
						
							
								
									
										
											 
										 
										
											
												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>
											
										 
										
											2008-02-13 15:03:37 -08:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
								 * @probe: probe to match
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								 * @num: get the nth matching probe's private data
							 | 
						
					
						
							
								
									
										
										
										
											2007-10-18 23:41:06 -07:00
										 
									 
								 
							 | 
							
								
							 | 
							
								
							 | 
							
							
								 *
							 | 
						
					
						
							
								
									
										
											 
										 
										
											
												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>
											
										 
										
											2008-02-13 15:03:37 -08:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
								 * Returns the nth private data pointer (starting from 0) matching, or an
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								 * ERR_PTR.
							 | 
						
					
						
							
								
									
										
										
										
											2007-10-18 23:41:06 -07:00
										 
									 
								 
							 | 
							
								
							 | 
							
								
							 | 
							
							
								 * Returns the private data pointer, or an ERR_PTR.
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								 * The private data pointer should _only_ be dereferenced if the caller is the
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								 * owner of the data, or its content could vanish. This is mostly used to
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								 * confirm that a caller is the owner of a registered probe.
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								 */
							 | 
						
					
						
							
								
									
										
											 
										 
										
											
												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>
											
										 
										
											2008-02-13 15:03:37 -08:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
								void *marker_get_private_data(const char *name, marker_probe_func *probe,
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
										int num)
							 | 
						
					
						
							
								
									
										
										
										
											2007-10-18 23:41:06 -07:00
										 
									 
								 
							 | 
							
								
							 | 
							
								
							 | 
							
							
								{
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									struct hlist_head *head;
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									struct hlist_node *node;
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									struct marker_entry *e;
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									size_t name_len = strlen(name) + 1;
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									u32 hash = jhash(name, name_len-1, 0);
							 | 
						
					
						
							
								
									
										
											 
										 
										
											
												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>
											
										 
										
											2008-02-13 15:03:37 -08:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
									int i;
							 | 
						
					
						
							
								
									
										
										
										
											2007-10-18 23:41:06 -07:00
										 
									 
								 
							 | 
							
								
							 | 
							
								
							 | 
							
							
								
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									head = &marker_table[hash & ((1 << MARKER_HASH_BITS)-1)];
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									hlist_for_each_entry(e, node, head, hlist) {
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
										if (!strcmp(name, e->name)) {
							 | 
						
					
						
							
								
									
										
											 
										 
										
											
												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>
											
										 
										
											2008-02-13 15:03:37 -08:00
										 
									 
								 
							 | 
							
								
									
										
									
								
							 | 
							
								
							 | 
							
							
											if (!e->ptype) {
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
												if (num == 0 && e->single.func == probe)
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
													return e->single.probe_private;
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
												else
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
													break;
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
											} else {
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
												struct marker_probe_closure *closure;
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
												int match = 0;
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
												closure = e->multi;
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
												for (i = 0; closure[i].func; i++) {
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
													if (closure[i].func != probe)
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
														continue;
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
													if (match++ == num)
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
														return closure[i].probe_private;
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
												}
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
											}
							 | 
						
					
						
							
								
									
										
										
										
											2007-10-18 23:41:06 -07:00
										 
									 
								 
							 | 
							
								
							 | 
							
								
							 | 
							
							
										}
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									}
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
									return ERR_PTR(-ENOENT);
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								}
							 | 
						
					
						
							| 
								
							 | 
							
								
							 | 
							
								
							 | 
							
							
								EXPORT_SYMBOL_GPL(marker_get_private_data);
							 |