2
0
mirror of https://github.com/edk2-porting/linux-next.git synced 2024-12-27 06:34:11 +08:00
linux-next/tools/perf/builtin-lock.c

1032 lines
23 KiB
C
Raw Normal View History

#include <errno.h>
#include <inttypes.h>
#include "builtin.h"
#include "perf.h"
#include "util/evlist.h"
#include "util/evsel.h"
#include "util/util.h"
#include "util/cache.h"
#include "util/symbol.h"
#include "util/thread.h"
#include "util/header.h"
#include <subcmd/parse-options.h>
#include "util/trace-event.h"
#include "util/debug.h"
#include "util/session.h"
#include "util/tool.h"
#include "util/data.h"
#include <sys/types.h>
#include <sys/prctl.h>
#include <semaphore.h>
#include <pthread.h>
#include <math.h>
#include <limits.h>
#include <linux/list.h>
#include <linux/hash.h>
#include <linux/kernel.h>
perf lock: Fix state machine to recognize lock sequence Previous state machine of perf lock was really broken. This patch improves it a little. This patch prepares the list of state machine that represents lock sequences for each threads. These state machines can be one of these sequences: 1) acquire -> acquired -> release 2) acquire -> contended -> acquired -> release 3) acquire (w/ try) -> release 4) acquire (w/ read) -> release The case of 4) is a little special. Double acquire of read lock is allowed, so the state machine counts read lock number, and permits double acquire and release. But, things are not so simple. Something in my model is still wrong. I counted the number of lock instances with bad sequence, and ratio is like this (case of tracing whoami): bad:233, total:2279 version 2: * threads are now identified with tid, not pid * prepared SEQ_STATE_READ_ACQUIRED for read lock. * bunch of struct lock_seq_stat is now linked list * debug information enhanced (this have to be removed someday) e.g. | === output for debug=== | | bad:233, total:2279 | bad rate:0.000000 | histogram of events caused bad sequence | acquire: 165 | acquired: 0 | contended: 0 | release: 68 Signed-off-by: Hitoshi Mitake <mitake@dcl.info.waseda.ac.jp> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Jens Axboe <jens.axboe@oracle.com> Cc: Jason Baron <jbaron@redhat.com> Cc: Xiao Guangrong <xiaoguangrong@cn.fujitsu.com> Cc: Ingo Molnar <mingo@elte.hu> LKML-Reference: <1271852634-9351-1-git-send-email-mitake@dcl.info.waseda.ac.jp> [rename SEQ_STATE_UNINITED to SEQ_STATE_UNINITIALIZED] Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>
2010-04-21 20:23:54 +08:00
static struct perf_session *session;
/* based on kernel/lockdep.c */
#define LOCKHASH_BITS 12
#define LOCKHASH_SIZE (1UL << LOCKHASH_BITS)
static struct list_head lockhash_table[LOCKHASH_SIZE];
#define __lockhashfn(key) hash_long((unsigned long)key, LOCKHASH_BITS)
#define lockhashentry(key) (lockhash_table + __lockhashfn((key)))
struct lock_stat {
struct list_head hash_entry;
struct rb_node rb; /* used for sorting */
/*
* FIXME: perf_evsel__intval() returns u64,
* so address of lockdep_map should be dealed as 64bit.
* Is there more better solution?
*/
void *addr; /* address of lockdep_map, used as ID */
char *name; /* for strcpy(), we cannot use const */
unsigned int nr_acquire;
perf lock: Fix state machine to recognize lock sequence Previous state machine of perf lock was really broken. This patch improves it a little. This patch prepares the list of state machine that represents lock sequences for each threads. These state machines can be one of these sequences: 1) acquire -> acquired -> release 2) acquire -> contended -> acquired -> release 3) acquire (w/ try) -> release 4) acquire (w/ read) -> release The case of 4) is a little special. Double acquire of read lock is allowed, so the state machine counts read lock number, and permits double acquire and release. But, things are not so simple. Something in my model is still wrong. I counted the number of lock instances with bad sequence, and ratio is like this (case of tracing whoami): bad:233, total:2279 version 2: * threads are now identified with tid, not pid * prepared SEQ_STATE_READ_ACQUIRED for read lock. * bunch of struct lock_seq_stat is now linked list * debug information enhanced (this have to be removed someday) e.g. | === output for debug=== | | bad:233, total:2279 | bad rate:0.000000 | histogram of events caused bad sequence | acquire: 165 | acquired: 0 | contended: 0 | release: 68 Signed-off-by: Hitoshi Mitake <mitake@dcl.info.waseda.ac.jp> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Jens Axboe <jens.axboe@oracle.com> Cc: Jason Baron <jbaron@redhat.com> Cc: Xiao Guangrong <xiaoguangrong@cn.fujitsu.com> Cc: Ingo Molnar <mingo@elte.hu> LKML-Reference: <1271852634-9351-1-git-send-email-mitake@dcl.info.waseda.ac.jp> [rename SEQ_STATE_UNINITED to SEQ_STATE_UNINITIALIZED] Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>
2010-04-21 20:23:54 +08:00
unsigned int nr_acquired;
unsigned int nr_contended;
unsigned int nr_release;
perf lock: Fix state machine to recognize lock sequence Previous state machine of perf lock was really broken. This patch improves it a little. This patch prepares the list of state machine that represents lock sequences for each threads. These state machines can be one of these sequences: 1) acquire -> acquired -> release 2) acquire -> contended -> acquired -> release 3) acquire (w/ try) -> release 4) acquire (w/ read) -> release The case of 4) is a little special. Double acquire of read lock is allowed, so the state machine counts read lock number, and permits double acquire and release. But, things are not so simple. Something in my model is still wrong. I counted the number of lock instances with bad sequence, and ratio is like this (case of tracing whoami): bad:233, total:2279 version 2: * threads are now identified with tid, not pid * prepared SEQ_STATE_READ_ACQUIRED for read lock. * bunch of struct lock_seq_stat is now linked list * debug information enhanced (this have to be removed someday) e.g. | === output for debug=== | | bad:233, total:2279 | bad rate:0.000000 | histogram of events caused bad sequence | acquire: 165 | acquired: 0 | contended: 0 | release: 68 Signed-off-by: Hitoshi Mitake <mitake@dcl.info.waseda.ac.jp> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Jens Axboe <jens.axboe@oracle.com> Cc: Jason Baron <jbaron@redhat.com> Cc: Xiao Guangrong <xiaoguangrong@cn.fujitsu.com> Cc: Ingo Molnar <mingo@elte.hu> LKML-Reference: <1271852634-9351-1-git-send-email-mitake@dcl.info.waseda.ac.jp> [rename SEQ_STATE_UNINITED to SEQ_STATE_UNINITIALIZED] Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>
2010-04-21 20:23:54 +08:00
unsigned int nr_readlock;
unsigned int nr_trylock;
/* these times are in nano sec. */
u64 avg_wait_time;
u64 wait_time_total;
u64 wait_time_min;
u64 wait_time_max;
perf lock: Fix state machine to recognize lock sequence Previous state machine of perf lock was really broken. This patch improves it a little. This patch prepares the list of state machine that represents lock sequences for each threads. These state machines can be one of these sequences: 1) acquire -> acquired -> release 2) acquire -> contended -> acquired -> release 3) acquire (w/ try) -> release 4) acquire (w/ read) -> release The case of 4) is a little special. Double acquire of read lock is allowed, so the state machine counts read lock number, and permits double acquire and release. But, things are not so simple. Something in my model is still wrong. I counted the number of lock instances with bad sequence, and ratio is like this (case of tracing whoami): bad:233, total:2279 version 2: * threads are now identified with tid, not pid * prepared SEQ_STATE_READ_ACQUIRED for read lock. * bunch of struct lock_seq_stat is now linked list * debug information enhanced (this have to be removed someday) e.g. | === output for debug=== | | bad:233, total:2279 | bad rate:0.000000 | histogram of events caused bad sequence | acquire: 165 | acquired: 0 | contended: 0 | release: 68 Signed-off-by: Hitoshi Mitake <mitake@dcl.info.waseda.ac.jp> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Jens Axboe <jens.axboe@oracle.com> Cc: Jason Baron <jbaron@redhat.com> Cc: Xiao Guangrong <xiaoguangrong@cn.fujitsu.com> Cc: Ingo Molnar <mingo@elte.hu> LKML-Reference: <1271852634-9351-1-git-send-email-mitake@dcl.info.waseda.ac.jp> [rename SEQ_STATE_UNINITED to SEQ_STATE_UNINITIALIZED] Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>
2010-04-21 20:23:54 +08:00
int discard; /* flag of blacklist */
};
/*
* States of lock_seq_stat
*
* UNINITIALIZED is required for detecting first event of acquire.
* As the nature of lock events, there is no guarantee
* that the first event for the locks are acquire,
* it can be acquired, contended or release.
*/
#define SEQ_STATE_UNINITIALIZED 0 /* initial state */
#define SEQ_STATE_RELEASED 1
#define SEQ_STATE_ACQUIRING 2
#define SEQ_STATE_ACQUIRED 3
#define SEQ_STATE_READ_ACQUIRED 4
#define SEQ_STATE_CONTENDED 5
/*
* MAX_LOCK_DEPTH
* Imported from include/linux/sched.h.
* Should this be synchronized?
*/
#define MAX_LOCK_DEPTH 48
/*
* struct lock_seq_stat:
* Place to put on state of one lock sequence
* 1) acquire -> acquired -> release
* 2) acquire -> contended -> acquired -> release
* 3) acquire (with read or try) -> release
* 4) Are there other patterns?
*/
struct lock_seq_stat {
struct list_head list;
int state;
u64 prev_event_time;
void *addr;
int read_count;
};
perf lock: Fix state machine to recognize lock sequence Previous state machine of perf lock was really broken. This patch improves it a little. This patch prepares the list of state machine that represents lock sequences for each threads. These state machines can be one of these sequences: 1) acquire -> acquired -> release 2) acquire -> contended -> acquired -> release 3) acquire (w/ try) -> release 4) acquire (w/ read) -> release The case of 4) is a little special. Double acquire of read lock is allowed, so the state machine counts read lock number, and permits double acquire and release. But, things are not so simple. Something in my model is still wrong. I counted the number of lock instances with bad sequence, and ratio is like this (case of tracing whoami): bad:233, total:2279 version 2: * threads are now identified with tid, not pid * prepared SEQ_STATE_READ_ACQUIRED for read lock. * bunch of struct lock_seq_stat is now linked list * debug information enhanced (this have to be removed someday) e.g. | === output for debug=== | | bad:233, total:2279 | bad rate:0.000000 | histogram of events caused bad sequence | acquire: 165 | acquired: 0 | contended: 0 | release: 68 Signed-off-by: Hitoshi Mitake <mitake@dcl.info.waseda.ac.jp> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Jens Axboe <jens.axboe@oracle.com> Cc: Jason Baron <jbaron@redhat.com> Cc: Xiao Guangrong <xiaoguangrong@cn.fujitsu.com> Cc: Ingo Molnar <mingo@elte.hu> LKML-Reference: <1271852634-9351-1-git-send-email-mitake@dcl.info.waseda.ac.jp> [rename SEQ_STATE_UNINITED to SEQ_STATE_UNINITIALIZED] Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>
2010-04-21 20:23:54 +08:00
struct thread_stat {
struct rb_node rb;
u32 tid;
struct list_head seq_list;
};
static struct rb_root thread_stats;
static struct thread_stat *thread_stat_find(u32 tid)
{
struct rb_node *node;
struct thread_stat *st;
node = thread_stats.rb_node;
while (node) {
st = container_of(node, struct thread_stat, rb);
if (st->tid == tid)
return st;
else if (tid < st->tid)
node = node->rb_left;
else
node = node->rb_right;
}
return NULL;
}
static void thread_stat_insert(struct thread_stat *new)
{
struct rb_node **rb = &thread_stats.rb_node;
struct rb_node *parent = NULL;
struct thread_stat *p;
while (*rb) {
p = container_of(*rb, struct thread_stat, rb);
parent = *rb;
if (new->tid < p->tid)
rb = &(*rb)->rb_left;
else if (new->tid > p->tid)
rb = &(*rb)->rb_right;
else
BUG_ON("inserting invalid thread_stat\n");
}
rb_link_node(&new->rb, parent, rb);
rb_insert_color(&new->rb, &thread_stats);
}
static struct thread_stat *thread_stat_findnew_after_first(u32 tid)
{
struct thread_stat *st;
st = thread_stat_find(tid);
if (st)
return st;
st = zalloc(sizeof(struct thread_stat));
if (!st) {
pr_err("memory allocation failed\n");
return NULL;
}
perf lock: Fix state machine to recognize lock sequence Previous state machine of perf lock was really broken. This patch improves it a little. This patch prepares the list of state machine that represents lock sequences for each threads. These state machines can be one of these sequences: 1) acquire -> acquired -> release 2) acquire -> contended -> acquired -> release 3) acquire (w/ try) -> release 4) acquire (w/ read) -> release The case of 4) is a little special. Double acquire of read lock is allowed, so the state machine counts read lock number, and permits double acquire and release. But, things are not so simple. Something in my model is still wrong. I counted the number of lock instances with bad sequence, and ratio is like this (case of tracing whoami): bad:233, total:2279 version 2: * threads are now identified with tid, not pid * prepared SEQ_STATE_READ_ACQUIRED for read lock. * bunch of struct lock_seq_stat is now linked list * debug information enhanced (this have to be removed someday) e.g. | === output for debug=== | | bad:233, total:2279 | bad rate:0.000000 | histogram of events caused bad sequence | acquire: 165 | acquired: 0 | contended: 0 | release: 68 Signed-off-by: Hitoshi Mitake <mitake@dcl.info.waseda.ac.jp> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Jens Axboe <jens.axboe@oracle.com> Cc: Jason Baron <jbaron@redhat.com> Cc: Xiao Guangrong <xiaoguangrong@cn.fujitsu.com> Cc: Ingo Molnar <mingo@elte.hu> LKML-Reference: <1271852634-9351-1-git-send-email-mitake@dcl.info.waseda.ac.jp> [rename SEQ_STATE_UNINITED to SEQ_STATE_UNINITIALIZED] Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>
2010-04-21 20:23:54 +08:00
st->tid = tid;
INIT_LIST_HEAD(&st->seq_list);
thread_stat_insert(st);
return st;
}
static struct thread_stat *thread_stat_findnew_first(u32 tid);
static struct thread_stat *(*thread_stat_findnew)(u32 tid) =
thread_stat_findnew_first;
static struct thread_stat *thread_stat_findnew_first(u32 tid)
{
struct thread_stat *st;
st = zalloc(sizeof(struct thread_stat));
if (!st) {
pr_err("memory allocation failed\n");
return NULL;
}
perf lock: Fix state machine to recognize lock sequence Previous state machine of perf lock was really broken. This patch improves it a little. This patch prepares the list of state machine that represents lock sequences for each threads. These state machines can be one of these sequences: 1) acquire -> acquired -> release 2) acquire -> contended -> acquired -> release 3) acquire (w/ try) -> release 4) acquire (w/ read) -> release The case of 4) is a little special. Double acquire of read lock is allowed, so the state machine counts read lock number, and permits double acquire and release. But, things are not so simple. Something in my model is still wrong. I counted the number of lock instances with bad sequence, and ratio is like this (case of tracing whoami): bad:233, total:2279 version 2: * threads are now identified with tid, not pid * prepared SEQ_STATE_READ_ACQUIRED for read lock. * bunch of struct lock_seq_stat is now linked list * debug information enhanced (this have to be removed someday) e.g. | === output for debug=== | | bad:233, total:2279 | bad rate:0.000000 | histogram of events caused bad sequence | acquire: 165 | acquired: 0 | contended: 0 | release: 68 Signed-off-by: Hitoshi Mitake <mitake@dcl.info.waseda.ac.jp> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Jens Axboe <jens.axboe@oracle.com> Cc: Jason Baron <jbaron@redhat.com> Cc: Xiao Guangrong <xiaoguangrong@cn.fujitsu.com> Cc: Ingo Molnar <mingo@elte.hu> LKML-Reference: <1271852634-9351-1-git-send-email-mitake@dcl.info.waseda.ac.jp> [rename SEQ_STATE_UNINITED to SEQ_STATE_UNINITIALIZED] Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>
2010-04-21 20:23:54 +08:00
st->tid = tid;
INIT_LIST_HEAD(&st->seq_list);
rb_link_node(&st->rb, NULL, &thread_stats.rb_node);
rb_insert_color(&st->rb, &thread_stats);
thread_stat_findnew = thread_stat_findnew_after_first;
return st;
}
/* build simple key function one is bigger than two */
#define SINGLE_KEY(member) \
static int lock_stat_key_ ## member(struct lock_stat *one, \
struct lock_stat *two) \
{ \
return one->member > two->member; \
}
SINGLE_KEY(nr_acquired)
SINGLE_KEY(nr_contended)
SINGLE_KEY(avg_wait_time)
SINGLE_KEY(wait_time_total)
SINGLE_KEY(wait_time_max)
static int lock_stat_key_wait_time_min(struct lock_stat *one,
struct lock_stat *two)
{
u64 s1 = one->wait_time_min;
u64 s2 = two->wait_time_min;
if (s1 == ULLONG_MAX)
s1 = 0;
if (s2 == ULLONG_MAX)
s2 = 0;
return s1 > s2;
}
struct lock_key {
/*
* name: the value for specify by user
* this should be simpler than raw name of member
* e.g. nr_acquired -> acquired, wait_time_total -> wait_total
*/
const char *name;
int (*key)(struct lock_stat*, struct lock_stat*);
};
static const char *sort_key = "acquired";
static int (*compare)(struct lock_stat *, struct lock_stat *);
static struct rb_root result; /* place to store sorted data */
#define DEF_KEY_LOCK(name, fn_suffix) \
{ #name, lock_stat_key_ ## fn_suffix }
struct lock_key keys[] = {
DEF_KEY_LOCK(acquired, nr_acquired),
DEF_KEY_LOCK(contended, nr_contended),
DEF_KEY_LOCK(avg_wait, avg_wait_time),
DEF_KEY_LOCK(wait_total, wait_time_total),
DEF_KEY_LOCK(wait_min, wait_time_min),
DEF_KEY_LOCK(wait_max, wait_time_max),
/* extra comparisons much complicated should be here */
{ NULL, NULL }
};
static int select_key(void)
{
int i;
for (i = 0; keys[i].name; i++) {
if (!strcmp(keys[i].name, sort_key)) {
compare = keys[i].key;
return 0;
}
}
pr_err("Unknown compare key: %s\n", sort_key);
return -1;
}
static void insert_to_result(struct lock_stat *st,
int (*bigger)(struct lock_stat *, struct lock_stat *))
{
struct rb_node **rb = &result.rb_node;
struct rb_node *parent = NULL;
struct lock_stat *p;
while (*rb) {
p = container_of(*rb, struct lock_stat, rb);
parent = *rb;
if (bigger(st, p))
rb = &(*rb)->rb_left;
else
rb = &(*rb)->rb_right;
}
rb_link_node(&st->rb, parent, rb);
rb_insert_color(&st->rb, &result);
}
/* returns left most element of result, and erase it */
static struct lock_stat *pop_from_result(void)
{
struct rb_node *node = result.rb_node;
if (!node)
return NULL;
while (node->rb_left)
node = node->rb_left;
rb_erase(node, &result);
return container_of(node, struct lock_stat, rb);
}
static struct lock_stat *lock_stat_findnew(void *addr, const char *name)
{
struct list_head *entry = lockhashentry(addr);
struct lock_stat *ret, *new;
list_for_each_entry(ret, entry, hash_entry) {
if (ret->addr == addr)
return ret;
}
new = zalloc(sizeof(struct lock_stat));
if (!new)
goto alloc_failed;
new->addr = addr;
new->name = zalloc(sizeof(char) * strlen(name) + 1);
if (!new->name) {
free(new);
goto alloc_failed;
}
strcpy(new->name, name);
new->wait_time_min = ULLONG_MAX;
list_add(&new->hash_entry, entry);
return new;
alloc_failed:
pr_err("memory allocation failed\n");
return NULL;
}
struct trace_lock_handler {
int (*acquire_event)(struct perf_evsel *evsel,
struct perf_sample *sample);
int (*acquired_event)(struct perf_evsel *evsel,
struct perf_sample *sample);
int (*contended_event)(struct perf_evsel *evsel,
struct perf_sample *sample);
int (*release_event)(struct perf_evsel *evsel,
struct perf_sample *sample);
};
perf lock: Fix state machine to recognize lock sequence Previous state machine of perf lock was really broken. This patch improves it a little. This patch prepares the list of state machine that represents lock sequences for each threads. These state machines can be one of these sequences: 1) acquire -> acquired -> release 2) acquire -> contended -> acquired -> release 3) acquire (w/ try) -> release 4) acquire (w/ read) -> release The case of 4) is a little special. Double acquire of read lock is allowed, so the state machine counts read lock number, and permits double acquire and release. But, things are not so simple. Something in my model is still wrong. I counted the number of lock instances with bad sequence, and ratio is like this (case of tracing whoami): bad:233, total:2279 version 2: * threads are now identified with tid, not pid * prepared SEQ_STATE_READ_ACQUIRED for read lock. * bunch of struct lock_seq_stat is now linked list * debug information enhanced (this have to be removed someday) e.g. | === output for debug=== | | bad:233, total:2279 | bad rate:0.000000 | histogram of events caused bad sequence | acquire: 165 | acquired: 0 | contended: 0 | release: 68 Signed-off-by: Hitoshi Mitake <mitake@dcl.info.waseda.ac.jp> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Jens Axboe <jens.axboe@oracle.com> Cc: Jason Baron <jbaron@redhat.com> Cc: Xiao Guangrong <xiaoguangrong@cn.fujitsu.com> Cc: Ingo Molnar <mingo@elte.hu> LKML-Reference: <1271852634-9351-1-git-send-email-mitake@dcl.info.waseda.ac.jp> [rename SEQ_STATE_UNINITED to SEQ_STATE_UNINITIALIZED] Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>
2010-04-21 20:23:54 +08:00
static struct lock_seq_stat *get_seq(struct thread_stat *ts, void *addr)
{
struct lock_seq_stat *seq;
list_for_each_entry(seq, &ts->seq_list, list) {
if (seq->addr == addr)
return seq;
}
seq = zalloc(sizeof(struct lock_seq_stat));
if (!seq) {
pr_err("memory allocation failed\n");
return NULL;
}
perf lock: Fix state machine to recognize lock sequence Previous state machine of perf lock was really broken. This patch improves it a little. This patch prepares the list of state machine that represents lock sequences for each threads. These state machines can be one of these sequences: 1) acquire -> acquired -> release 2) acquire -> contended -> acquired -> release 3) acquire (w/ try) -> release 4) acquire (w/ read) -> release The case of 4) is a little special. Double acquire of read lock is allowed, so the state machine counts read lock number, and permits double acquire and release. But, things are not so simple. Something in my model is still wrong. I counted the number of lock instances with bad sequence, and ratio is like this (case of tracing whoami): bad:233, total:2279 version 2: * threads are now identified with tid, not pid * prepared SEQ_STATE_READ_ACQUIRED for read lock. * bunch of struct lock_seq_stat is now linked list * debug information enhanced (this have to be removed someday) e.g. | === output for debug=== | | bad:233, total:2279 | bad rate:0.000000 | histogram of events caused bad sequence | acquire: 165 | acquired: 0 | contended: 0 | release: 68 Signed-off-by: Hitoshi Mitake <mitake@dcl.info.waseda.ac.jp> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Jens Axboe <jens.axboe@oracle.com> Cc: Jason Baron <jbaron@redhat.com> Cc: Xiao Guangrong <xiaoguangrong@cn.fujitsu.com> Cc: Ingo Molnar <mingo@elte.hu> LKML-Reference: <1271852634-9351-1-git-send-email-mitake@dcl.info.waseda.ac.jp> [rename SEQ_STATE_UNINITED to SEQ_STATE_UNINITIALIZED] Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>
2010-04-21 20:23:54 +08:00
seq->state = SEQ_STATE_UNINITIALIZED;
seq->addr = addr;
list_add(&seq->list, &ts->seq_list);
return seq;
}
enum broken_state {
BROKEN_ACQUIRE,
BROKEN_ACQUIRED,
BROKEN_CONTENDED,
BROKEN_RELEASE,
BROKEN_MAX,
};
static int bad_hist[BROKEN_MAX];
perf lock: Fix state machine to recognize lock sequence Previous state machine of perf lock was really broken. This patch improves it a little. This patch prepares the list of state machine that represents lock sequences for each threads. These state machines can be one of these sequences: 1) acquire -> acquired -> release 2) acquire -> contended -> acquired -> release 3) acquire (w/ try) -> release 4) acquire (w/ read) -> release The case of 4) is a little special. Double acquire of read lock is allowed, so the state machine counts read lock number, and permits double acquire and release. But, things are not so simple. Something in my model is still wrong. I counted the number of lock instances with bad sequence, and ratio is like this (case of tracing whoami): bad:233, total:2279 version 2: * threads are now identified with tid, not pid * prepared SEQ_STATE_READ_ACQUIRED for read lock. * bunch of struct lock_seq_stat is now linked list * debug information enhanced (this have to be removed someday) e.g. | === output for debug=== | | bad:233, total:2279 | bad rate:0.000000 | histogram of events caused bad sequence | acquire: 165 | acquired: 0 | contended: 0 | release: 68 Signed-off-by: Hitoshi Mitake <mitake@dcl.info.waseda.ac.jp> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Jens Axboe <jens.axboe@oracle.com> Cc: Jason Baron <jbaron@redhat.com> Cc: Xiao Guangrong <xiaoguangrong@cn.fujitsu.com> Cc: Ingo Molnar <mingo@elte.hu> LKML-Reference: <1271852634-9351-1-git-send-email-mitake@dcl.info.waseda.ac.jp> [rename SEQ_STATE_UNINITED to SEQ_STATE_UNINITIALIZED] Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>
2010-04-21 20:23:54 +08:00
enum acquire_flags {
TRY_LOCK = 1,
READ_LOCK = 2,
};
static int report_lock_acquire_event(struct perf_evsel *evsel,
struct perf_sample *sample)
{
void *addr;
perf lock: Fix state machine to recognize lock sequence Previous state machine of perf lock was really broken. This patch improves it a little. This patch prepares the list of state machine that represents lock sequences for each threads. These state machines can be one of these sequences: 1) acquire -> acquired -> release 2) acquire -> contended -> acquired -> release 3) acquire (w/ try) -> release 4) acquire (w/ read) -> release The case of 4) is a little special. Double acquire of read lock is allowed, so the state machine counts read lock number, and permits double acquire and release. But, things are not so simple. Something in my model is still wrong. I counted the number of lock instances with bad sequence, and ratio is like this (case of tracing whoami): bad:233, total:2279 version 2: * threads are now identified with tid, not pid * prepared SEQ_STATE_READ_ACQUIRED for read lock. * bunch of struct lock_seq_stat is now linked list * debug information enhanced (this have to be removed someday) e.g. | === output for debug=== | | bad:233, total:2279 | bad rate:0.000000 | histogram of events caused bad sequence | acquire: 165 | acquired: 0 | contended: 0 | release: 68 Signed-off-by: Hitoshi Mitake <mitake@dcl.info.waseda.ac.jp> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Jens Axboe <jens.axboe@oracle.com> Cc: Jason Baron <jbaron@redhat.com> Cc: Xiao Guangrong <xiaoguangrong@cn.fujitsu.com> Cc: Ingo Molnar <mingo@elte.hu> LKML-Reference: <1271852634-9351-1-git-send-email-mitake@dcl.info.waseda.ac.jp> [rename SEQ_STATE_UNINITED to SEQ_STATE_UNINITIALIZED] Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>
2010-04-21 20:23:54 +08:00
struct lock_stat *ls;
struct thread_stat *ts;
struct lock_seq_stat *seq;
const char *name = perf_evsel__strval(evsel, sample, "name");
u64 tmp = perf_evsel__intval(evsel, sample, "lockdep_addr");
int flag = perf_evsel__intval(evsel, sample, "flag");
perf lock: Fix state machine to recognize lock sequence Previous state machine of perf lock was really broken. This patch improves it a little. This patch prepares the list of state machine that represents lock sequences for each threads. These state machines can be one of these sequences: 1) acquire -> acquired -> release 2) acquire -> contended -> acquired -> release 3) acquire (w/ try) -> release 4) acquire (w/ read) -> release The case of 4) is a little special. Double acquire of read lock is allowed, so the state machine counts read lock number, and permits double acquire and release. But, things are not so simple. Something in my model is still wrong. I counted the number of lock instances with bad sequence, and ratio is like this (case of tracing whoami): bad:233, total:2279 version 2: * threads are now identified with tid, not pid * prepared SEQ_STATE_READ_ACQUIRED for read lock. * bunch of struct lock_seq_stat is now linked list * debug information enhanced (this have to be removed someday) e.g. | === output for debug=== | | bad:233, total:2279 | bad rate:0.000000 | histogram of events caused bad sequence | acquire: 165 | acquired: 0 | contended: 0 | release: 68 Signed-off-by: Hitoshi Mitake <mitake@dcl.info.waseda.ac.jp> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Jens Axboe <jens.axboe@oracle.com> Cc: Jason Baron <jbaron@redhat.com> Cc: Xiao Guangrong <xiaoguangrong@cn.fujitsu.com> Cc: Ingo Molnar <mingo@elte.hu> LKML-Reference: <1271852634-9351-1-git-send-email-mitake@dcl.info.waseda.ac.jp> [rename SEQ_STATE_UNINITED to SEQ_STATE_UNINITIALIZED] Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>
2010-04-21 20:23:54 +08:00
memcpy(&addr, &tmp, sizeof(void *));
ls = lock_stat_findnew(addr, name);
if (!ls)
return -ENOMEM;
perf lock: Fix state machine to recognize lock sequence Previous state machine of perf lock was really broken. This patch improves it a little. This patch prepares the list of state machine that represents lock sequences for each threads. These state machines can be one of these sequences: 1) acquire -> acquired -> release 2) acquire -> contended -> acquired -> release 3) acquire (w/ try) -> release 4) acquire (w/ read) -> release The case of 4) is a little special. Double acquire of read lock is allowed, so the state machine counts read lock number, and permits double acquire and release. But, things are not so simple. Something in my model is still wrong. I counted the number of lock instances with bad sequence, and ratio is like this (case of tracing whoami): bad:233, total:2279 version 2: * threads are now identified with tid, not pid * prepared SEQ_STATE_READ_ACQUIRED for read lock. * bunch of struct lock_seq_stat is now linked list * debug information enhanced (this have to be removed someday) e.g. | === output for debug=== | | bad:233, total:2279 | bad rate:0.000000 | histogram of events caused bad sequence | acquire: 165 | acquired: 0 | contended: 0 | release: 68 Signed-off-by: Hitoshi Mitake <mitake@dcl.info.waseda.ac.jp> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Jens Axboe <jens.axboe@oracle.com> Cc: Jason Baron <jbaron@redhat.com> Cc: Xiao Guangrong <xiaoguangrong@cn.fujitsu.com> Cc: Ingo Molnar <mingo@elte.hu> LKML-Reference: <1271852634-9351-1-git-send-email-mitake@dcl.info.waseda.ac.jp> [rename SEQ_STATE_UNINITED to SEQ_STATE_UNINITIALIZED] Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>
2010-04-21 20:23:54 +08:00
if (ls->discard)
return 0;
ts = thread_stat_findnew(sample->tid);
if (!ts)
return -ENOMEM;
seq = get_seq(ts, addr);
if (!seq)
return -ENOMEM;
perf lock: Fix state machine to recognize lock sequence Previous state machine of perf lock was really broken. This patch improves it a little. This patch prepares the list of state machine that represents lock sequences for each threads. These state machines can be one of these sequences: 1) acquire -> acquired -> release 2) acquire -> contended -> acquired -> release 3) acquire (w/ try) -> release 4) acquire (w/ read) -> release The case of 4) is a little special. Double acquire of read lock is allowed, so the state machine counts read lock number, and permits double acquire and release. But, things are not so simple. Something in my model is still wrong. I counted the number of lock instances with bad sequence, and ratio is like this (case of tracing whoami): bad:233, total:2279 version 2: * threads are now identified with tid, not pid * prepared SEQ_STATE_READ_ACQUIRED for read lock. * bunch of struct lock_seq_stat is now linked list * debug information enhanced (this have to be removed someday) e.g. | === output for debug=== | | bad:233, total:2279 | bad rate:0.000000 | histogram of events caused bad sequence | acquire: 165 | acquired: 0 | contended: 0 | release: 68 Signed-off-by: Hitoshi Mitake <mitake@dcl.info.waseda.ac.jp> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Jens Axboe <jens.axboe@oracle.com> Cc: Jason Baron <jbaron@redhat.com> Cc: Xiao Guangrong <xiaoguangrong@cn.fujitsu.com> Cc: Ingo Molnar <mingo@elte.hu> LKML-Reference: <1271852634-9351-1-git-send-email-mitake@dcl.info.waseda.ac.jp> [rename SEQ_STATE_UNINITED to SEQ_STATE_UNINITIALIZED] Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>
2010-04-21 20:23:54 +08:00
switch (seq->state) {
case SEQ_STATE_UNINITIALIZED:
case SEQ_STATE_RELEASED:
if (!flag) {
perf lock: Fix state machine to recognize lock sequence Previous state machine of perf lock was really broken. This patch improves it a little. This patch prepares the list of state machine that represents lock sequences for each threads. These state machines can be one of these sequences: 1) acquire -> acquired -> release 2) acquire -> contended -> acquired -> release 3) acquire (w/ try) -> release 4) acquire (w/ read) -> release The case of 4) is a little special. Double acquire of read lock is allowed, so the state machine counts read lock number, and permits double acquire and release. But, things are not so simple. Something in my model is still wrong. I counted the number of lock instances with bad sequence, and ratio is like this (case of tracing whoami): bad:233, total:2279 version 2: * threads are now identified with tid, not pid * prepared SEQ_STATE_READ_ACQUIRED for read lock. * bunch of struct lock_seq_stat is now linked list * debug information enhanced (this have to be removed someday) e.g. | === output for debug=== | | bad:233, total:2279 | bad rate:0.000000 | histogram of events caused bad sequence | acquire: 165 | acquired: 0 | contended: 0 | release: 68 Signed-off-by: Hitoshi Mitake <mitake@dcl.info.waseda.ac.jp> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Jens Axboe <jens.axboe@oracle.com> Cc: Jason Baron <jbaron@redhat.com> Cc: Xiao Guangrong <xiaoguangrong@cn.fujitsu.com> Cc: Ingo Molnar <mingo@elte.hu> LKML-Reference: <1271852634-9351-1-git-send-email-mitake@dcl.info.waseda.ac.jp> [rename SEQ_STATE_UNINITED to SEQ_STATE_UNINITIALIZED] Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>
2010-04-21 20:23:54 +08:00
seq->state = SEQ_STATE_ACQUIRING;
} else {
if (flag & TRY_LOCK)
perf lock: Fix state machine to recognize lock sequence Previous state machine of perf lock was really broken. This patch improves it a little. This patch prepares the list of state machine that represents lock sequences for each threads. These state machines can be one of these sequences: 1) acquire -> acquired -> release 2) acquire -> contended -> acquired -> release 3) acquire (w/ try) -> release 4) acquire (w/ read) -> release The case of 4) is a little special. Double acquire of read lock is allowed, so the state machine counts read lock number, and permits double acquire and release. But, things are not so simple. Something in my model is still wrong. I counted the number of lock instances with bad sequence, and ratio is like this (case of tracing whoami): bad:233, total:2279 version 2: * threads are now identified with tid, not pid * prepared SEQ_STATE_READ_ACQUIRED for read lock. * bunch of struct lock_seq_stat is now linked list * debug information enhanced (this have to be removed someday) e.g. | === output for debug=== | | bad:233, total:2279 | bad rate:0.000000 | histogram of events caused bad sequence | acquire: 165 | acquired: 0 | contended: 0 | release: 68 Signed-off-by: Hitoshi Mitake <mitake@dcl.info.waseda.ac.jp> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Jens Axboe <jens.axboe@oracle.com> Cc: Jason Baron <jbaron@redhat.com> Cc: Xiao Guangrong <xiaoguangrong@cn.fujitsu.com> Cc: Ingo Molnar <mingo@elte.hu> LKML-Reference: <1271852634-9351-1-git-send-email-mitake@dcl.info.waseda.ac.jp> [rename SEQ_STATE_UNINITED to SEQ_STATE_UNINITIALIZED] Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>
2010-04-21 20:23:54 +08:00
ls->nr_trylock++;
if (flag & READ_LOCK)
perf lock: Fix state machine to recognize lock sequence Previous state machine of perf lock was really broken. This patch improves it a little. This patch prepares the list of state machine that represents lock sequences for each threads. These state machines can be one of these sequences: 1) acquire -> acquired -> release 2) acquire -> contended -> acquired -> release 3) acquire (w/ try) -> release 4) acquire (w/ read) -> release The case of 4) is a little special. Double acquire of read lock is allowed, so the state machine counts read lock number, and permits double acquire and release. But, things are not so simple. Something in my model is still wrong. I counted the number of lock instances with bad sequence, and ratio is like this (case of tracing whoami): bad:233, total:2279 version 2: * threads are now identified with tid, not pid * prepared SEQ_STATE_READ_ACQUIRED for read lock. * bunch of struct lock_seq_stat is now linked list * debug information enhanced (this have to be removed someday) e.g. | === output for debug=== | | bad:233, total:2279 | bad rate:0.000000 | histogram of events caused bad sequence | acquire: 165 | acquired: 0 | contended: 0 | release: 68 Signed-off-by: Hitoshi Mitake <mitake@dcl.info.waseda.ac.jp> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Jens Axboe <jens.axboe@oracle.com> Cc: Jason Baron <jbaron@redhat.com> Cc: Xiao Guangrong <xiaoguangrong@cn.fujitsu.com> Cc: Ingo Molnar <mingo@elte.hu> LKML-Reference: <1271852634-9351-1-git-send-email-mitake@dcl.info.waseda.ac.jp> [rename SEQ_STATE_UNINITED to SEQ_STATE_UNINITIALIZED] Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>
2010-04-21 20:23:54 +08:00
ls->nr_readlock++;
seq->state = SEQ_STATE_READ_ACQUIRED;
seq->read_count = 1;
ls->nr_acquired++;
}
break;
case SEQ_STATE_READ_ACQUIRED:
if (flag & READ_LOCK) {
perf lock: Fix state machine to recognize lock sequence Previous state machine of perf lock was really broken. This patch improves it a little. This patch prepares the list of state machine that represents lock sequences for each threads. These state machines can be one of these sequences: 1) acquire -> acquired -> release 2) acquire -> contended -> acquired -> release 3) acquire (w/ try) -> release 4) acquire (w/ read) -> release The case of 4) is a little special. Double acquire of read lock is allowed, so the state machine counts read lock number, and permits double acquire and release. But, things are not so simple. Something in my model is still wrong. I counted the number of lock instances with bad sequence, and ratio is like this (case of tracing whoami): bad:233, total:2279 version 2: * threads are now identified with tid, not pid * prepared SEQ_STATE_READ_ACQUIRED for read lock. * bunch of struct lock_seq_stat is now linked list * debug information enhanced (this have to be removed someday) e.g. | === output for debug=== | | bad:233, total:2279 | bad rate:0.000000 | histogram of events caused bad sequence | acquire: 165 | acquired: 0 | contended: 0 | release: 68 Signed-off-by: Hitoshi Mitake <mitake@dcl.info.waseda.ac.jp> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Jens Axboe <jens.axboe@oracle.com> Cc: Jason Baron <jbaron@redhat.com> Cc: Xiao Guangrong <xiaoguangrong@cn.fujitsu.com> Cc: Ingo Molnar <mingo@elte.hu> LKML-Reference: <1271852634-9351-1-git-send-email-mitake@dcl.info.waseda.ac.jp> [rename SEQ_STATE_UNINITED to SEQ_STATE_UNINITIALIZED] Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>
2010-04-21 20:23:54 +08:00
seq->read_count++;
ls->nr_acquired++;
goto end;
} else {
goto broken;
}
break;
perf lock: Fix state machine to recognize lock sequence Previous state machine of perf lock was really broken. This patch improves it a little. This patch prepares the list of state machine that represents lock sequences for each threads. These state machines can be one of these sequences: 1) acquire -> acquired -> release 2) acquire -> contended -> acquired -> release 3) acquire (w/ try) -> release 4) acquire (w/ read) -> release The case of 4) is a little special. Double acquire of read lock is allowed, so the state machine counts read lock number, and permits double acquire and release. But, things are not so simple. Something in my model is still wrong. I counted the number of lock instances with bad sequence, and ratio is like this (case of tracing whoami): bad:233, total:2279 version 2: * threads are now identified with tid, not pid * prepared SEQ_STATE_READ_ACQUIRED for read lock. * bunch of struct lock_seq_stat is now linked list * debug information enhanced (this have to be removed someday) e.g. | === output for debug=== | | bad:233, total:2279 | bad rate:0.000000 | histogram of events caused bad sequence | acquire: 165 | acquired: 0 | contended: 0 | release: 68 Signed-off-by: Hitoshi Mitake <mitake@dcl.info.waseda.ac.jp> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Jens Axboe <jens.axboe@oracle.com> Cc: Jason Baron <jbaron@redhat.com> Cc: Xiao Guangrong <xiaoguangrong@cn.fujitsu.com> Cc: Ingo Molnar <mingo@elte.hu> LKML-Reference: <1271852634-9351-1-git-send-email-mitake@dcl.info.waseda.ac.jp> [rename SEQ_STATE_UNINITED to SEQ_STATE_UNINITIALIZED] Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>
2010-04-21 20:23:54 +08:00
case SEQ_STATE_ACQUIRED:
case SEQ_STATE_ACQUIRING:
case SEQ_STATE_CONTENDED:
broken:
/* broken lock sequence, discard it */
ls->discard = 1;
bad_hist[BROKEN_ACQUIRE]++;
perf lock: Fix state machine to recognize lock sequence Previous state machine of perf lock was really broken. This patch improves it a little. This patch prepares the list of state machine that represents lock sequences for each threads. These state machines can be one of these sequences: 1) acquire -> acquired -> release 2) acquire -> contended -> acquired -> release 3) acquire (w/ try) -> release 4) acquire (w/ read) -> release The case of 4) is a little special. Double acquire of read lock is allowed, so the state machine counts read lock number, and permits double acquire and release. But, things are not so simple. Something in my model is still wrong. I counted the number of lock instances with bad sequence, and ratio is like this (case of tracing whoami): bad:233, total:2279 version 2: * threads are now identified with tid, not pid * prepared SEQ_STATE_READ_ACQUIRED for read lock. * bunch of struct lock_seq_stat is now linked list * debug information enhanced (this have to be removed someday) e.g. | === output for debug=== | | bad:233, total:2279 | bad rate:0.000000 | histogram of events caused bad sequence | acquire: 165 | acquired: 0 | contended: 0 | release: 68 Signed-off-by: Hitoshi Mitake <mitake@dcl.info.waseda.ac.jp> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Jens Axboe <jens.axboe@oracle.com> Cc: Jason Baron <jbaron@redhat.com> Cc: Xiao Guangrong <xiaoguangrong@cn.fujitsu.com> Cc: Ingo Molnar <mingo@elte.hu> LKML-Reference: <1271852634-9351-1-git-send-email-mitake@dcl.info.waseda.ac.jp> [rename SEQ_STATE_UNINITED to SEQ_STATE_UNINITIALIZED] Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>
2010-04-21 20:23:54 +08:00
list_del(&seq->list);
free(seq);
goto end;
default:
perf lock: Fix state machine to recognize lock sequence Previous state machine of perf lock was really broken. This patch improves it a little. This patch prepares the list of state machine that represents lock sequences for each threads. These state machines can be one of these sequences: 1) acquire -> acquired -> release 2) acquire -> contended -> acquired -> release 3) acquire (w/ try) -> release 4) acquire (w/ read) -> release The case of 4) is a little special. Double acquire of read lock is allowed, so the state machine counts read lock number, and permits double acquire and release. But, things are not so simple. Something in my model is still wrong. I counted the number of lock instances with bad sequence, and ratio is like this (case of tracing whoami): bad:233, total:2279 version 2: * threads are now identified with tid, not pid * prepared SEQ_STATE_READ_ACQUIRED for read lock. * bunch of struct lock_seq_stat is now linked list * debug information enhanced (this have to be removed someday) e.g. | === output for debug=== | | bad:233, total:2279 | bad rate:0.000000 | histogram of events caused bad sequence | acquire: 165 | acquired: 0 | contended: 0 | release: 68 Signed-off-by: Hitoshi Mitake <mitake@dcl.info.waseda.ac.jp> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Jens Axboe <jens.axboe@oracle.com> Cc: Jason Baron <jbaron@redhat.com> Cc: Xiao Guangrong <xiaoguangrong@cn.fujitsu.com> Cc: Ingo Molnar <mingo@elte.hu> LKML-Reference: <1271852634-9351-1-git-send-email-mitake@dcl.info.waseda.ac.jp> [rename SEQ_STATE_UNINITED to SEQ_STATE_UNINITIALIZED] Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>
2010-04-21 20:23:54 +08:00
BUG_ON("Unknown state of lock sequence found!\n");
break;
}
perf lock: Fix state machine to recognize lock sequence Previous state machine of perf lock was really broken. This patch improves it a little. This patch prepares the list of state machine that represents lock sequences for each threads. These state machines can be one of these sequences: 1) acquire -> acquired -> release 2) acquire -> contended -> acquired -> release 3) acquire (w/ try) -> release 4) acquire (w/ read) -> release The case of 4) is a little special. Double acquire of read lock is allowed, so the state machine counts read lock number, and permits double acquire and release. But, things are not so simple. Something in my model is still wrong. I counted the number of lock instances with bad sequence, and ratio is like this (case of tracing whoami): bad:233, total:2279 version 2: * threads are now identified with tid, not pid * prepared SEQ_STATE_READ_ACQUIRED for read lock. * bunch of struct lock_seq_stat is now linked list * debug information enhanced (this have to be removed someday) e.g. | === output for debug=== | | bad:233, total:2279 | bad rate:0.000000 | histogram of events caused bad sequence | acquire: 165 | acquired: 0 | contended: 0 | release: 68 Signed-off-by: Hitoshi Mitake <mitake@dcl.info.waseda.ac.jp> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Jens Axboe <jens.axboe@oracle.com> Cc: Jason Baron <jbaron@redhat.com> Cc: Xiao Guangrong <xiaoguangrong@cn.fujitsu.com> Cc: Ingo Molnar <mingo@elte.hu> LKML-Reference: <1271852634-9351-1-git-send-email-mitake@dcl.info.waseda.ac.jp> [rename SEQ_STATE_UNINITED to SEQ_STATE_UNINITIALIZED] Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>
2010-04-21 20:23:54 +08:00
ls->nr_acquire++;
seq->prev_event_time = sample->time;
perf lock: Fix state machine to recognize lock sequence Previous state machine of perf lock was really broken. This patch improves it a little. This patch prepares the list of state machine that represents lock sequences for each threads. These state machines can be one of these sequences: 1) acquire -> acquired -> release 2) acquire -> contended -> acquired -> release 3) acquire (w/ try) -> release 4) acquire (w/ read) -> release The case of 4) is a little special. Double acquire of read lock is allowed, so the state machine counts read lock number, and permits double acquire and release. But, things are not so simple. Something in my model is still wrong. I counted the number of lock instances with bad sequence, and ratio is like this (case of tracing whoami): bad:233, total:2279 version 2: * threads are now identified with tid, not pid * prepared SEQ_STATE_READ_ACQUIRED for read lock. * bunch of struct lock_seq_stat is now linked list * debug information enhanced (this have to be removed someday) e.g. | === output for debug=== | | bad:233, total:2279 | bad rate:0.000000 | histogram of events caused bad sequence | acquire: 165 | acquired: 0 | contended: 0 | release: 68 Signed-off-by: Hitoshi Mitake <mitake@dcl.info.waseda.ac.jp> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Jens Axboe <jens.axboe@oracle.com> Cc: Jason Baron <jbaron@redhat.com> Cc: Xiao Guangrong <xiaoguangrong@cn.fujitsu.com> Cc: Ingo Molnar <mingo@elte.hu> LKML-Reference: <1271852634-9351-1-git-send-email-mitake@dcl.info.waseda.ac.jp> [rename SEQ_STATE_UNINITED to SEQ_STATE_UNINITIALIZED] Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>
2010-04-21 20:23:54 +08:00
end:
return 0;
}
static int report_lock_acquired_event(struct perf_evsel *evsel,
struct perf_sample *sample)
{
void *addr;
perf lock: Fix state machine to recognize lock sequence Previous state machine of perf lock was really broken. This patch improves it a little. This patch prepares the list of state machine that represents lock sequences for each threads. These state machines can be one of these sequences: 1) acquire -> acquired -> release 2) acquire -> contended -> acquired -> release 3) acquire (w/ try) -> release 4) acquire (w/ read) -> release The case of 4) is a little special. Double acquire of read lock is allowed, so the state machine counts read lock number, and permits double acquire and release. But, things are not so simple. Something in my model is still wrong. I counted the number of lock instances with bad sequence, and ratio is like this (case of tracing whoami): bad:233, total:2279 version 2: * threads are now identified with tid, not pid * prepared SEQ_STATE_READ_ACQUIRED for read lock. * bunch of struct lock_seq_stat is now linked list * debug information enhanced (this have to be removed someday) e.g. | === output for debug=== | | bad:233, total:2279 | bad rate:0.000000 | histogram of events caused bad sequence | acquire: 165 | acquired: 0 | contended: 0 | release: 68 Signed-off-by: Hitoshi Mitake <mitake@dcl.info.waseda.ac.jp> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Jens Axboe <jens.axboe@oracle.com> Cc: Jason Baron <jbaron@redhat.com> Cc: Xiao Guangrong <xiaoguangrong@cn.fujitsu.com> Cc: Ingo Molnar <mingo@elte.hu> LKML-Reference: <1271852634-9351-1-git-send-email-mitake@dcl.info.waseda.ac.jp> [rename SEQ_STATE_UNINITED to SEQ_STATE_UNINITIALIZED] Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>
2010-04-21 20:23:54 +08:00
struct lock_stat *ls;
struct thread_stat *ts;
struct lock_seq_stat *seq;
u64 contended_term;
const char *name = perf_evsel__strval(evsel, sample, "name");
u64 tmp = perf_evsel__intval(evsel, sample, "lockdep_addr");
memcpy(&addr, &tmp, sizeof(void *));
ls = lock_stat_findnew(addr, name);
if (!ls)
return -ENOMEM;
perf lock: Fix state machine to recognize lock sequence Previous state machine of perf lock was really broken. This patch improves it a little. This patch prepares the list of state machine that represents lock sequences for each threads. These state machines can be one of these sequences: 1) acquire -> acquired -> release 2) acquire -> contended -> acquired -> release 3) acquire (w/ try) -> release 4) acquire (w/ read) -> release The case of 4) is a little special. Double acquire of read lock is allowed, so the state machine counts read lock number, and permits double acquire and release. But, things are not so simple. Something in my model is still wrong. I counted the number of lock instances with bad sequence, and ratio is like this (case of tracing whoami): bad:233, total:2279 version 2: * threads are now identified with tid, not pid * prepared SEQ_STATE_READ_ACQUIRED for read lock. * bunch of struct lock_seq_stat is now linked list * debug information enhanced (this have to be removed someday) e.g. | === output for debug=== | | bad:233, total:2279 | bad rate:0.000000 | histogram of events caused bad sequence | acquire: 165 | acquired: 0 | contended: 0 | release: 68 Signed-off-by: Hitoshi Mitake <mitake@dcl.info.waseda.ac.jp> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Jens Axboe <jens.axboe@oracle.com> Cc: Jason Baron <jbaron@redhat.com> Cc: Xiao Guangrong <xiaoguangrong@cn.fujitsu.com> Cc: Ingo Molnar <mingo@elte.hu> LKML-Reference: <1271852634-9351-1-git-send-email-mitake@dcl.info.waseda.ac.jp> [rename SEQ_STATE_UNINITED to SEQ_STATE_UNINITIALIZED] Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>
2010-04-21 20:23:54 +08:00
if (ls->discard)
return 0;
perf lock: Fix state machine to recognize lock sequence Previous state machine of perf lock was really broken. This patch improves it a little. This patch prepares the list of state machine that represents lock sequences for each threads. These state machines can be one of these sequences: 1) acquire -> acquired -> release 2) acquire -> contended -> acquired -> release 3) acquire (w/ try) -> release 4) acquire (w/ read) -> release The case of 4) is a little special. Double acquire of read lock is allowed, so the state machine counts read lock number, and permits double acquire and release. But, things are not so simple. Something in my model is still wrong. I counted the number of lock instances with bad sequence, and ratio is like this (case of tracing whoami): bad:233, total:2279 version 2: * threads are now identified with tid, not pid * prepared SEQ_STATE_READ_ACQUIRED for read lock. * bunch of struct lock_seq_stat is now linked list * debug information enhanced (this have to be removed someday) e.g. | === output for debug=== | | bad:233, total:2279 | bad rate:0.000000 | histogram of events caused bad sequence | acquire: 165 | acquired: 0 | contended: 0 | release: 68 Signed-off-by: Hitoshi Mitake <mitake@dcl.info.waseda.ac.jp> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Jens Axboe <jens.axboe@oracle.com> Cc: Jason Baron <jbaron@redhat.com> Cc: Xiao Guangrong <xiaoguangrong@cn.fujitsu.com> Cc: Ingo Molnar <mingo@elte.hu> LKML-Reference: <1271852634-9351-1-git-send-email-mitake@dcl.info.waseda.ac.jp> [rename SEQ_STATE_UNINITED to SEQ_STATE_UNINITIALIZED] Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>
2010-04-21 20:23:54 +08:00
ts = thread_stat_findnew(sample->tid);
if (!ts)
return -ENOMEM;
seq = get_seq(ts, addr);
if (!seq)
return -ENOMEM;
perf lock: Fix state machine to recognize lock sequence Previous state machine of perf lock was really broken. This patch improves it a little. This patch prepares the list of state machine that represents lock sequences for each threads. These state machines can be one of these sequences: 1) acquire -> acquired -> release 2) acquire -> contended -> acquired -> release 3) acquire (w/ try) -> release 4) acquire (w/ read) -> release The case of 4) is a little special. Double acquire of read lock is allowed, so the state machine counts read lock number, and permits double acquire and release. But, things are not so simple. Something in my model is still wrong. I counted the number of lock instances with bad sequence, and ratio is like this (case of tracing whoami): bad:233, total:2279 version 2: * threads are now identified with tid, not pid * prepared SEQ_STATE_READ_ACQUIRED for read lock. * bunch of struct lock_seq_stat is now linked list * debug information enhanced (this have to be removed someday) e.g. | === output for debug=== | | bad:233, total:2279 | bad rate:0.000000 | histogram of events caused bad sequence | acquire: 165 | acquired: 0 | contended: 0 | release: 68 Signed-off-by: Hitoshi Mitake <mitake@dcl.info.waseda.ac.jp> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Jens Axboe <jens.axboe@oracle.com> Cc: Jason Baron <jbaron@redhat.com> Cc: Xiao Guangrong <xiaoguangrong@cn.fujitsu.com> Cc: Ingo Molnar <mingo@elte.hu> LKML-Reference: <1271852634-9351-1-git-send-email-mitake@dcl.info.waseda.ac.jp> [rename SEQ_STATE_UNINITED to SEQ_STATE_UNINITIALIZED] Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>
2010-04-21 20:23:54 +08:00
switch (seq->state) {
case SEQ_STATE_UNINITIALIZED:
/* orphan event, do nothing */
return 0;
perf lock: Fix state machine to recognize lock sequence Previous state machine of perf lock was really broken. This patch improves it a little. This patch prepares the list of state machine that represents lock sequences for each threads. These state machines can be one of these sequences: 1) acquire -> acquired -> release 2) acquire -> contended -> acquired -> release 3) acquire (w/ try) -> release 4) acquire (w/ read) -> release The case of 4) is a little special. Double acquire of read lock is allowed, so the state machine counts read lock number, and permits double acquire and release. But, things are not so simple. Something in my model is still wrong. I counted the number of lock instances with bad sequence, and ratio is like this (case of tracing whoami): bad:233, total:2279 version 2: * threads are now identified with tid, not pid * prepared SEQ_STATE_READ_ACQUIRED for read lock. * bunch of struct lock_seq_stat is now linked list * debug information enhanced (this have to be removed someday) e.g. | === output for debug=== | | bad:233, total:2279 | bad rate:0.000000 | histogram of events caused bad sequence | acquire: 165 | acquired: 0 | contended: 0 | release: 68 Signed-off-by: Hitoshi Mitake <mitake@dcl.info.waseda.ac.jp> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Jens Axboe <jens.axboe@oracle.com> Cc: Jason Baron <jbaron@redhat.com> Cc: Xiao Guangrong <xiaoguangrong@cn.fujitsu.com> Cc: Ingo Molnar <mingo@elte.hu> LKML-Reference: <1271852634-9351-1-git-send-email-mitake@dcl.info.waseda.ac.jp> [rename SEQ_STATE_UNINITED to SEQ_STATE_UNINITIALIZED] Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>
2010-04-21 20:23:54 +08:00
case SEQ_STATE_ACQUIRING:
break;
perf lock: Fix state machine to recognize lock sequence Previous state machine of perf lock was really broken. This patch improves it a little. This patch prepares the list of state machine that represents lock sequences for each threads. These state machines can be one of these sequences: 1) acquire -> acquired -> release 2) acquire -> contended -> acquired -> release 3) acquire (w/ try) -> release 4) acquire (w/ read) -> release The case of 4) is a little special. Double acquire of read lock is allowed, so the state machine counts read lock number, and permits double acquire and release. But, things are not so simple. Something in my model is still wrong. I counted the number of lock instances with bad sequence, and ratio is like this (case of tracing whoami): bad:233, total:2279 version 2: * threads are now identified with tid, not pid * prepared SEQ_STATE_READ_ACQUIRED for read lock. * bunch of struct lock_seq_stat is now linked list * debug information enhanced (this have to be removed someday) e.g. | === output for debug=== | | bad:233, total:2279 | bad rate:0.000000 | histogram of events caused bad sequence | acquire: 165 | acquired: 0 | contended: 0 | release: 68 Signed-off-by: Hitoshi Mitake <mitake@dcl.info.waseda.ac.jp> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Jens Axboe <jens.axboe@oracle.com> Cc: Jason Baron <jbaron@redhat.com> Cc: Xiao Guangrong <xiaoguangrong@cn.fujitsu.com> Cc: Ingo Molnar <mingo@elte.hu> LKML-Reference: <1271852634-9351-1-git-send-email-mitake@dcl.info.waseda.ac.jp> [rename SEQ_STATE_UNINITED to SEQ_STATE_UNINITIALIZED] Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>
2010-04-21 20:23:54 +08:00
case SEQ_STATE_CONTENDED:
contended_term = sample->time - seq->prev_event_time;
perf lock: Fix state machine to recognize lock sequence Previous state machine of perf lock was really broken. This patch improves it a little. This patch prepares the list of state machine that represents lock sequences for each threads. These state machines can be one of these sequences: 1) acquire -> acquired -> release 2) acquire -> contended -> acquired -> release 3) acquire (w/ try) -> release 4) acquire (w/ read) -> release The case of 4) is a little special. Double acquire of read lock is allowed, so the state machine counts read lock number, and permits double acquire and release. But, things are not so simple. Something in my model is still wrong. I counted the number of lock instances with bad sequence, and ratio is like this (case of tracing whoami): bad:233, total:2279 version 2: * threads are now identified with tid, not pid * prepared SEQ_STATE_READ_ACQUIRED for read lock. * bunch of struct lock_seq_stat is now linked list * debug information enhanced (this have to be removed someday) e.g. | === output for debug=== | | bad:233, total:2279 | bad rate:0.000000 | histogram of events caused bad sequence | acquire: 165 | acquired: 0 | contended: 0 | release: 68 Signed-off-by: Hitoshi Mitake <mitake@dcl.info.waseda.ac.jp> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Jens Axboe <jens.axboe@oracle.com> Cc: Jason Baron <jbaron@redhat.com> Cc: Xiao Guangrong <xiaoguangrong@cn.fujitsu.com> Cc: Ingo Molnar <mingo@elte.hu> LKML-Reference: <1271852634-9351-1-git-send-email-mitake@dcl.info.waseda.ac.jp> [rename SEQ_STATE_UNINITED to SEQ_STATE_UNINITIALIZED] Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>
2010-04-21 20:23:54 +08:00
ls->wait_time_total += contended_term;
if (contended_term < ls->wait_time_min)
ls->wait_time_min = contended_term;
if (ls->wait_time_max < contended_term)
perf lock: Fix state machine to recognize lock sequence Previous state machine of perf lock was really broken. This patch improves it a little. This patch prepares the list of state machine that represents lock sequences for each threads. These state machines can be one of these sequences: 1) acquire -> acquired -> release 2) acquire -> contended -> acquired -> release 3) acquire (w/ try) -> release 4) acquire (w/ read) -> release The case of 4) is a little special. Double acquire of read lock is allowed, so the state machine counts read lock number, and permits double acquire and release. But, things are not so simple. Something in my model is still wrong. I counted the number of lock instances with bad sequence, and ratio is like this (case of tracing whoami): bad:233, total:2279 version 2: * threads are now identified with tid, not pid * prepared SEQ_STATE_READ_ACQUIRED for read lock. * bunch of struct lock_seq_stat is now linked list * debug information enhanced (this have to be removed someday) e.g. | === output for debug=== | | bad:233, total:2279 | bad rate:0.000000 | histogram of events caused bad sequence | acquire: 165 | acquired: 0 | contended: 0 | release: 68 Signed-off-by: Hitoshi Mitake <mitake@dcl.info.waseda.ac.jp> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Jens Axboe <jens.axboe@oracle.com> Cc: Jason Baron <jbaron@redhat.com> Cc: Xiao Guangrong <xiaoguangrong@cn.fujitsu.com> Cc: Ingo Molnar <mingo@elte.hu> LKML-Reference: <1271852634-9351-1-git-send-email-mitake@dcl.info.waseda.ac.jp> [rename SEQ_STATE_UNINITED to SEQ_STATE_UNINITIALIZED] Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>
2010-04-21 20:23:54 +08:00
ls->wait_time_max = contended_term;
break;
perf lock: Fix state machine to recognize lock sequence Previous state machine of perf lock was really broken. This patch improves it a little. This patch prepares the list of state machine that represents lock sequences for each threads. These state machines can be one of these sequences: 1) acquire -> acquired -> release 2) acquire -> contended -> acquired -> release 3) acquire (w/ try) -> release 4) acquire (w/ read) -> release The case of 4) is a little special. Double acquire of read lock is allowed, so the state machine counts read lock number, and permits double acquire and release. But, things are not so simple. Something in my model is still wrong. I counted the number of lock instances with bad sequence, and ratio is like this (case of tracing whoami): bad:233, total:2279 version 2: * threads are now identified with tid, not pid * prepared SEQ_STATE_READ_ACQUIRED for read lock. * bunch of struct lock_seq_stat is now linked list * debug information enhanced (this have to be removed someday) e.g. | === output for debug=== | | bad:233, total:2279 | bad rate:0.000000 | histogram of events caused bad sequence | acquire: 165 | acquired: 0 | contended: 0 | release: 68 Signed-off-by: Hitoshi Mitake <mitake@dcl.info.waseda.ac.jp> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Jens Axboe <jens.axboe@oracle.com> Cc: Jason Baron <jbaron@redhat.com> Cc: Xiao Guangrong <xiaoguangrong@cn.fujitsu.com> Cc: Ingo Molnar <mingo@elte.hu> LKML-Reference: <1271852634-9351-1-git-send-email-mitake@dcl.info.waseda.ac.jp> [rename SEQ_STATE_UNINITED to SEQ_STATE_UNINITIALIZED] Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>
2010-04-21 20:23:54 +08:00
case SEQ_STATE_RELEASED:
case SEQ_STATE_ACQUIRED:
case SEQ_STATE_READ_ACQUIRED:
/* broken lock sequence, discard it */
ls->discard = 1;
bad_hist[BROKEN_ACQUIRED]++;
perf lock: Fix state machine to recognize lock sequence Previous state machine of perf lock was really broken. This patch improves it a little. This patch prepares the list of state machine that represents lock sequences for each threads. These state machines can be one of these sequences: 1) acquire -> acquired -> release 2) acquire -> contended -> acquired -> release 3) acquire (w/ try) -> release 4) acquire (w/ read) -> release The case of 4) is a little special. Double acquire of read lock is allowed, so the state machine counts read lock number, and permits double acquire and release. But, things are not so simple. Something in my model is still wrong. I counted the number of lock instances with bad sequence, and ratio is like this (case of tracing whoami): bad:233, total:2279 version 2: * threads are now identified with tid, not pid * prepared SEQ_STATE_READ_ACQUIRED for read lock. * bunch of struct lock_seq_stat is now linked list * debug information enhanced (this have to be removed someday) e.g. | === output for debug=== | | bad:233, total:2279 | bad rate:0.000000 | histogram of events caused bad sequence | acquire: 165 | acquired: 0 | contended: 0 | release: 68 Signed-off-by: Hitoshi Mitake <mitake@dcl.info.waseda.ac.jp> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Jens Axboe <jens.axboe@oracle.com> Cc: Jason Baron <jbaron@redhat.com> Cc: Xiao Guangrong <xiaoguangrong@cn.fujitsu.com> Cc: Ingo Molnar <mingo@elte.hu> LKML-Reference: <1271852634-9351-1-git-send-email-mitake@dcl.info.waseda.ac.jp> [rename SEQ_STATE_UNINITED to SEQ_STATE_UNINITIALIZED] Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>
2010-04-21 20:23:54 +08:00
list_del(&seq->list);
free(seq);
goto end;
default:
perf lock: Fix state machine to recognize lock sequence Previous state machine of perf lock was really broken. This patch improves it a little. This patch prepares the list of state machine that represents lock sequences for each threads. These state machines can be one of these sequences: 1) acquire -> acquired -> release 2) acquire -> contended -> acquired -> release 3) acquire (w/ try) -> release 4) acquire (w/ read) -> release The case of 4) is a little special. Double acquire of read lock is allowed, so the state machine counts read lock number, and permits double acquire and release. But, things are not so simple. Something in my model is still wrong. I counted the number of lock instances with bad sequence, and ratio is like this (case of tracing whoami): bad:233, total:2279 version 2: * threads are now identified with tid, not pid * prepared SEQ_STATE_READ_ACQUIRED for read lock. * bunch of struct lock_seq_stat is now linked list * debug information enhanced (this have to be removed someday) e.g. | === output for debug=== | | bad:233, total:2279 | bad rate:0.000000 | histogram of events caused bad sequence | acquire: 165 | acquired: 0 | contended: 0 | release: 68 Signed-off-by: Hitoshi Mitake <mitake@dcl.info.waseda.ac.jp> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Jens Axboe <jens.axboe@oracle.com> Cc: Jason Baron <jbaron@redhat.com> Cc: Xiao Guangrong <xiaoguangrong@cn.fujitsu.com> Cc: Ingo Molnar <mingo@elte.hu> LKML-Reference: <1271852634-9351-1-git-send-email-mitake@dcl.info.waseda.ac.jp> [rename SEQ_STATE_UNINITED to SEQ_STATE_UNINITIALIZED] Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>
2010-04-21 20:23:54 +08:00
BUG_ON("Unknown state of lock sequence found!\n");
break;
}
perf lock: Fix state machine to recognize lock sequence Previous state machine of perf lock was really broken. This patch improves it a little. This patch prepares the list of state machine that represents lock sequences for each threads. These state machines can be one of these sequences: 1) acquire -> acquired -> release 2) acquire -> contended -> acquired -> release 3) acquire (w/ try) -> release 4) acquire (w/ read) -> release The case of 4) is a little special. Double acquire of read lock is allowed, so the state machine counts read lock number, and permits double acquire and release. But, things are not so simple. Something in my model is still wrong. I counted the number of lock instances with bad sequence, and ratio is like this (case of tracing whoami): bad:233, total:2279 version 2: * threads are now identified with tid, not pid * prepared SEQ_STATE_READ_ACQUIRED for read lock. * bunch of struct lock_seq_stat is now linked list * debug information enhanced (this have to be removed someday) e.g. | === output for debug=== | | bad:233, total:2279 | bad rate:0.000000 | histogram of events caused bad sequence | acquire: 165 | acquired: 0 | contended: 0 | release: 68 Signed-off-by: Hitoshi Mitake <mitake@dcl.info.waseda.ac.jp> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Jens Axboe <jens.axboe@oracle.com> Cc: Jason Baron <jbaron@redhat.com> Cc: Xiao Guangrong <xiaoguangrong@cn.fujitsu.com> Cc: Ingo Molnar <mingo@elte.hu> LKML-Reference: <1271852634-9351-1-git-send-email-mitake@dcl.info.waseda.ac.jp> [rename SEQ_STATE_UNINITED to SEQ_STATE_UNINITIALIZED] Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>
2010-04-21 20:23:54 +08:00
seq->state = SEQ_STATE_ACQUIRED;
ls->nr_acquired++;
ls->avg_wait_time = ls->nr_contended ? ls->wait_time_total/ls->nr_contended : 0;
seq->prev_event_time = sample->time;
perf lock: Fix state machine to recognize lock sequence Previous state machine of perf lock was really broken. This patch improves it a little. This patch prepares the list of state machine that represents lock sequences for each threads. These state machines can be one of these sequences: 1) acquire -> acquired -> release 2) acquire -> contended -> acquired -> release 3) acquire (w/ try) -> release 4) acquire (w/ read) -> release The case of 4) is a little special. Double acquire of read lock is allowed, so the state machine counts read lock number, and permits double acquire and release. But, things are not so simple. Something in my model is still wrong. I counted the number of lock instances with bad sequence, and ratio is like this (case of tracing whoami): bad:233, total:2279 version 2: * threads are now identified with tid, not pid * prepared SEQ_STATE_READ_ACQUIRED for read lock. * bunch of struct lock_seq_stat is now linked list * debug information enhanced (this have to be removed someday) e.g. | === output for debug=== | | bad:233, total:2279 | bad rate:0.000000 | histogram of events caused bad sequence | acquire: 165 | acquired: 0 | contended: 0 | release: 68 Signed-off-by: Hitoshi Mitake <mitake@dcl.info.waseda.ac.jp> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Jens Axboe <jens.axboe@oracle.com> Cc: Jason Baron <jbaron@redhat.com> Cc: Xiao Guangrong <xiaoguangrong@cn.fujitsu.com> Cc: Ingo Molnar <mingo@elte.hu> LKML-Reference: <1271852634-9351-1-git-send-email-mitake@dcl.info.waseda.ac.jp> [rename SEQ_STATE_UNINITED to SEQ_STATE_UNINITIALIZED] Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>
2010-04-21 20:23:54 +08:00
end:
return 0;
}
static int report_lock_contended_event(struct perf_evsel *evsel,
struct perf_sample *sample)
{
void *addr;
perf lock: Fix state machine to recognize lock sequence Previous state machine of perf lock was really broken. This patch improves it a little. This patch prepares the list of state machine that represents lock sequences for each threads. These state machines can be one of these sequences: 1) acquire -> acquired -> release 2) acquire -> contended -> acquired -> release 3) acquire (w/ try) -> release 4) acquire (w/ read) -> release The case of 4) is a little special. Double acquire of read lock is allowed, so the state machine counts read lock number, and permits double acquire and release. But, things are not so simple. Something in my model is still wrong. I counted the number of lock instances with bad sequence, and ratio is like this (case of tracing whoami): bad:233, total:2279 version 2: * threads are now identified with tid, not pid * prepared SEQ_STATE_READ_ACQUIRED for read lock. * bunch of struct lock_seq_stat is now linked list * debug information enhanced (this have to be removed someday) e.g. | === output for debug=== | | bad:233, total:2279 | bad rate:0.000000 | histogram of events caused bad sequence | acquire: 165 | acquired: 0 | contended: 0 | release: 68 Signed-off-by: Hitoshi Mitake <mitake@dcl.info.waseda.ac.jp> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Jens Axboe <jens.axboe@oracle.com> Cc: Jason Baron <jbaron@redhat.com> Cc: Xiao Guangrong <xiaoguangrong@cn.fujitsu.com> Cc: Ingo Molnar <mingo@elte.hu> LKML-Reference: <1271852634-9351-1-git-send-email-mitake@dcl.info.waseda.ac.jp> [rename SEQ_STATE_UNINITED to SEQ_STATE_UNINITIALIZED] Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>
2010-04-21 20:23:54 +08:00
struct lock_stat *ls;
struct thread_stat *ts;
struct lock_seq_stat *seq;
const char *name = perf_evsel__strval(evsel, sample, "name");
u64 tmp = perf_evsel__intval(evsel, sample, "lockdep_addr");
memcpy(&addr, &tmp, sizeof(void *));
perf lock: Fix state machine to recognize lock sequence Previous state machine of perf lock was really broken. This patch improves it a little. This patch prepares the list of state machine that represents lock sequences for each threads. These state machines can be one of these sequences: 1) acquire -> acquired -> release 2) acquire -> contended -> acquired -> release 3) acquire (w/ try) -> release 4) acquire (w/ read) -> release The case of 4) is a little special. Double acquire of read lock is allowed, so the state machine counts read lock number, and permits double acquire and release. But, things are not so simple. Something in my model is still wrong. I counted the number of lock instances with bad sequence, and ratio is like this (case of tracing whoami): bad:233, total:2279 version 2: * threads are now identified with tid, not pid * prepared SEQ_STATE_READ_ACQUIRED for read lock. * bunch of struct lock_seq_stat is now linked list * debug information enhanced (this have to be removed someday) e.g. | === output for debug=== | | bad:233, total:2279 | bad rate:0.000000 | histogram of events caused bad sequence | acquire: 165 | acquired: 0 | contended: 0 | release: 68 Signed-off-by: Hitoshi Mitake <mitake@dcl.info.waseda.ac.jp> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Jens Axboe <jens.axboe@oracle.com> Cc: Jason Baron <jbaron@redhat.com> Cc: Xiao Guangrong <xiaoguangrong@cn.fujitsu.com> Cc: Ingo Molnar <mingo@elte.hu> LKML-Reference: <1271852634-9351-1-git-send-email-mitake@dcl.info.waseda.ac.jp> [rename SEQ_STATE_UNINITED to SEQ_STATE_UNINITIALIZED] Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>
2010-04-21 20:23:54 +08:00
ls = lock_stat_findnew(addr, name);
if (!ls)
return -ENOMEM;
perf lock: Fix state machine to recognize lock sequence Previous state machine of perf lock was really broken. This patch improves it a little. This patch prepares the list of state machine that represents lock sequences for each threads. These state machines can be one of these sequences: 1) acquire -> acquired -> release 2) acquire -> contended -> acquired -> release 3) acquire (w/ try) -> release 4) acquire (w/ read) -> release The case of 4) is a little special. Double acquire of read lock is allowed, so the state machine counts read lock number, and permits double acquire and release. But, things are not so simple. Something in my model is still wrong. I counted the number of lock instances with bad sequence, and ratio is like this (case of tracing whoami): bad:233, total:2279 version 2: * threads are now identified with tid, not pid * prepared SEQ_STATE_READ_ACQUIRED for read lock. * bunch of struct lock_seq_stat is now linked list * debug information enhanced (this have to be removed someday) e.g. | === output for debug=== | | bad:233, total:2279 | bad rate:0.000000 | histogram of events caused bad sequence | acquire: 165 | acquired: 0 | contended: 0 | release: 68 Signed-off-by: Hitoshi Mitake <mitake@dcl.info.waseda.ac.jp> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Jens Axboe <jens.axboe@oracle.com> Cc: Jason Baron <jbaron@redhat.com> Cc: Xiao Guangrong <xiaoguangrong@cn.fujitsu.com> Cc: Ingo Molnar <mingo@elte.hu> LKML-Reference: <1271852634-9351-1-git-send-email-mitake@dcl.info.waseda.ac.jp> [rename SEQ_STATE_UNINITED to SEQ_STATE_UNINITIALIZED] Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>
2010-04-21 20:23:54 +08:00
if (ls->discard)
return 0;
ts = thread_stat_findnew(sample->tid);
if (!ts)
return -ENOMEM;
seq = get_seq(ts, addr);
if (!seq)
return -ENOMEM;
perf lock: Fix state machine to recognize lock sequence Previous state machine of perf lock was really broken. This patch improves it a little. This patch prepares the list of state machine that represents lock sequences for each threads. These state machines can be one of these sequences: 1) acquire -> acquired -> release 2) acquire -> contended -> acquired -> release 3) acquire (w/ try) -> release 4) acquire (w/ read) -> release The case of 4) is a little special. Double acquire of read lock is allowed, so the state machine counts read lock number, and permits double acquire and release. But, things are not so simple. Something in my model is still wrong. I counted the number of lock instances with bad sequence, and ratio is like this (case of tracing whoami): bad:233, total:2279 version 2: * threads are now identified with tid, not pid * prepared SEQ_STATE_READ_ACQUIRED for read lock. * bunch of struct lock_seq_stat is now linked list * debug information enhanced (this have to be removed someday) e.g. | === output for debug=== | | bad:233, total:2279 | bad rate:0.000000 | histogram of events caused bad sequence | acquire: 165 | acquired: 0 | contended: 0 | release: 68 Signed-off-by: Hitoshi Mitake <mitake@dcl.info.waseda.ac.jp> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Jens Axboe <jens.axboe@oracle.com> Cc: Jason Baron <jbaron@redhat.com> Cc: Xiao Guangrong <xiaoguangrong@cn.fujitsu.com> Cc: Ingo Molnar <mingo@elte.hu> LKML-Reference: <1271852634-9351-1-git-send-email-mitake@dcl.info.waseda.ac.jp> [rename SEQ_STATE_UNINITED to SEQ_STATE_UNINITIALIZED] Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>
2010-04-21 20:23:54 +08:00
switch (seq->state) {
case SEQ_STATE_UNINITIALIZED:
/* orphan event, do nothing */
return 0;
perf lock: Fix state machine to recognize lock sequence Previous state machine of perf lock was really broken. This patch improves it a little. This patch prepares the list of state machine that represents lock sequences for each threads. These state machines can be one of these sequences: 1) acquire -> acquired -> release 2) acquire -> contended -> acquired -> release 3) acquire (w/ try) -> release 4) acquire (w/ read) -> release The case of 4) is a little special. Double acquire of read lock is allowed, so the state machine counts read lock number, and permits double acquire and release. But, things are not so simple. Something in my model is still wrong. I counted the number of lock instances with bad sequence, and ratio is like this (case of tracing whoami): bad:233, total:2279 version 2: * threads are now identified with tid, not pid * prepared SEQ_STATE_READ_ACQUIRED for read lock. * bunch of struct lock_seq_stat is now linked list * debug information enhanced (this have to be removed someday) e.g. | === output for debug=== | | bad:233, total:2279 | bad rate:0.000000 | histogram of events caused bad sequence | acquire: 165 | acquired: 0 | contended: 0 | release: 68 Signed-off-by: Hitoshi Mitake <mitake@dcl.info.waseda.ac.jp> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Jens Axboe <jens.axboe@oracle.com> Cc: Jason Baron <jbaron@redhat.com> Cc: Xiao Guangrong <xiaoguangrong@cn.fujitsu.com> Cc: Ingo Molnar <mingo@elte.hu> LKML-Reference: <1271852634-9351-1-git-send-email-mitake@dcl.info.waseda.ac.jp> [rename SEQ_STATE_UNINITED to SEQ_STATE_UNINITIALIZED] Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>
2010-04-21 20:23:54 +08:00
case SEQ_STATE_ACQUIRING:
break;
perf lock: Fix state machine to recognize lock sequence Previous state machine of perf lock was really broken. This patch improves it a little. This patch prepares the list of state machine that represents lock sequences for each threads. These state machines can be one of these sequences: 1) acquire -> acquired -> release 2) acquire -> contended -> acquired -> release 3) acquire (w/ try) -> release 4) acquire (w/ read) -> release The case of 4) is a little special. Double acquire of read lock is allowed, so the state machine counts read lock number, and permits double acquire and release. But, things are not so simple. Something in my model is still wrong. I counted the number of lock instances with bad sequence, and ratio is like this (case of tracing whoami): bad:233, total:2279 version 2: * threads are now identified with tid, not pid * prepared SEQ_STATE_READ_ACQUIRED for read lock. * bunch of struct lock_seq_stat is now linked list * debug information enhanced (this have to be removed someday) e.g. | === output for debug=== | | bad:233, total:2279 | bad rate:0.000000 | histogram of events caused bad sequence | acquire: 165 | acquired: 0 | contended: 0 | release: 68 Signed-off-by: Hitoshi Mitake <mitake@dcl.info.waseda.ac.jp> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Jens Axboe <jens.axboe@oracle.com> Cc: Jason Baron <jbaron@redhat.com> Cc: Xiao Guangrong <xiaoguangrong@cn.fujitsu.com> Cc: Ingo Molnar <mingo@elte.hu> LKML-Reference: <1271852634-9351-1-git-send-email-mitake@dcl.info.waseda.ac.jp> [rename SEQ_STATE_UNINITED to SEQ_STATE_UNINITIALIZED] Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>
2010-04-21 20:23:54 +08:00
case SEQ_STATE_RELEASED:
case SEQ_STATE_ACQUIRED:
case SEQ_STATE_READ_ACQUIRED:
case SEQ_STATE_CONTENDED:
/* broken lock sequence, discard it */
ls->discard = 1;
bad_hist[BROKEN_CONTENDED]++;
perf lock: Fix state machine to recognize lock sequence Previous state machine of perf lock was really broken. This patch improves it a little. This patch prepares the list of state machine that represents lock sequences for each threads. These state machines can be one of these sequences: 1) acquire -> acquired -> release 2) acquire -> contended -> acquired -> release 3) acquire (w/ try) -> release 4) acquire (w/ read) -> release The case of 4) is a little special. Double acquire of read lock is allowed, so the state machine counts read lock number, and permits double acquire and release. But, things are not so simple. Something in my model is still wrong. I counted the number of lock instances with bad sequence, and ratio is like this (case of tracing whoami): bad:233, total:2279 version 2: * threads are now identified with tid, not pid * prepared SEQ_STATE_READ_ACQUIRED for read lock. * bunch of struct lock_seq_stat is now linked list * debug information enhanced (this have to be removed someday) e.g. | === output for debug=== | | bad:233, total:2279 | bad rate:0.000000 | histogram of events caused bad sequence | acquire: 165 | acquired: 0 | contended: 0 | release: 68 Signed-off-by: Hitoshi Mitake <mitake@dcl.info.waseda.ac.jp> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Jens Axboe <jens.axboe@oracle.com> Cc: Jason Baron <jbaron@redhat.com> Cc: Xiao Guangrong <xiaoguangrong@cn.fujitsu.com> Cc: Ingo Molnar <mingo@elte.hu> LKML-Reference: <1271852634-9351-1-git-send-email-mitake@dcl.info.waseda.ac.jp> [rename SEQ_STATE_UNINITED to SEQ_STATE_UNINITIALIZED] Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>
2010-04-21 20:23:54 +08:00
list_del(&seq->list);
free(seq);
goto end;
default:
perf lock: Fix state machine to recognize lock sequence Previous state machine of perf lock was really broken. This patch improves it a little. This patch prepares the list of state machine that represents lock sequences for each threads. These state machines can be one of these sequences: 1) acquire -> acquired -> release 2) acquire -> contended -> acquired -> release 3) acquire (w/ try) -> release 4) acquire (w/ read) -> release The case of 4) is a little special. Double acquire of read lock is allowed, so the state machine counts read lock number, and permits double acquire and release. But, things are not so simple. Something in my model is still wrong. I counted the number of lock instances with bad sequence, and ratio is like this (case of tracing whoami): bad:233, total:2279 version 2: * threads are now identified with tid, not pid * prepared SEQ_STATE_READ_ACQUIRED for read lock. * bunch of struct lock_seq_stat is now linked list * debug information enhanced (this have to be removed someday) e.g. | === output for debug=== | | bad:233, total:2279 | bad rate:0.000000 | histogram of events caused bad sequence | acquire: 165 | acquired: 0 | contended: 0 | release: 68 Signed-off-by: Hitoshi Mitake <mitake@dcl.info.waseda.ac.jp> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Jens Axboe <jens.axboe@oracle.com> Cc: Jason Baron <jbaron@redhat.com> Cc: Xiao Guangrong <xiaoguangrong@cn.fujitsu.com> Cc: Ingo Molnar <mingo@elte.hu> LKML-Reference: <1271852634-9351-1-git-send-email-mitake@dcl.info.waseda.ac.jp> [rename SEQ_STATE_UNINITED to SEQ_STATE_UNINITIALIZED] Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>
2010-04-21 20:23:54 +08:00
BUG_ON("Unknown state of lock sequence found!\n");
break;
}
perf lock: Fix state machine to recognize lock sequence Previous state machine of perf lock was really broken. This patch improves it a little. This patch prepares the list of state machine that represents lock sequences for each threads. These state machines can be one of these sequences: 1) acquire -> acquired -> release 2) acquire -> contended -> acquired -> release 3) acquire (w/ try) -> release 4) acquire (w/ read) -> release The case of 4) is a little special. Double acquire of read lock is allowed, so the state machine counts read lock number, and permits double acquire and release. But, things are not so simple. Something in my model is still wrong. I counted the number of lock instances with bad sequence, and ratio is like this (case of tracing whoami): bad:233, total:2279 version 2: * threads are now identified with tid, not pid * prepared SEQ_STATE_READ_ACQUIRED for read lock. * bunch of struct lock_seq_stat is now linked list * debug information enhanced (this have to be removed someday) e.g. | === output for debug=== | | bad:233, total:2279 | bad rate:0.000000 | histogram of events caused bad sequence | acquire: 165 | acquired: 0 | contended: 0 | release: 68 Signed-off-by: Hitoshi Mitake <mitake@dcl.info.waseda.ac.jp> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Jens Axboe <jens.axboe@oracle.com> Cc: Jason Baron <jbaron@redhat.com> Cc: Xiao Guangrong <xiaoguangrong@cn.fujitsu.com> Cc: Ingo Molnar <mingo@elte.hu> LKML-Reference: <1271852634-9351-1-git-send-email-mitake@dcl.info.waseda.ac.jp> [rename SEQ_STATE_UNINITED to SEQ_STATE_UNINITIALIZED] Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>
2010-04-21 20:23:54 +08:00
seq->state = SEQ_STATE_CONTENDED;
ls->nr_contended++;
ls->avg_wait_time = ls->wait_time_total/ls->nr_contended;
seq->prev_event_time = sample->time;
perf lock: Fix state machine to recognize lock sequence Previous state machine of perf lock was really broken. This patch improves it a little. This patch prepares the list of state machine that represents lock sequences for each threads. These state machines can be one of these sequences: 1) acquire -> acquired -> release 2) acquire -> contended -> acquired -> release 3) acquire (w/ try) -> release 4) acquire (w/ read) -> release The case of 4) is a little special. Double acquire of read lock is allowed, so the state machine counts read lock number, and permits double acquire and release. But, things are not so simple. Something in my model is still wrong. I counted the number of lock instances with bad sequence, and ratio is like this (case of tracing whoami): bad:233, total:2279 version 2: * threads are now identified with tid, not pid * prepared SEQ_STATE_READ_ACQUIRED for read lock. * bunch of struct lock_seq_stat is now linked list * debug information enhanced (this have to be removed someday) e.g. | === output for debug=== | | bad:233, total:2279 | bad rate:0.000000 | histogram of events caused bad sequence | acquire: 165 | acquired: 0 | contended: 0 | release: 68 Signed-off-by: Hitoshi Mitake <mitake@dcl.info.waseda.ac.jp> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Jens Axboe <jens.axboe@oracle.com> Cc: Jason Baron <jbaron@redhat.com> Cc: Xiao Guangrong <xiaoguangrong@cn.fujitsu.com> Cc: Ingo Molnar <mingo@elte.hu> LKML-Reference: <1271852634-9351-1-git-send-email-mitake@dcl.info.waseda.ac.jp> [rename SEQ_STATE_UNINITED to SEQ_STATE_UNINITIALIZED] Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>
2010-04-21 20:23:54 +08:00
end:
return 0;
}
static int report_lock_release_event(struct perf_evsel *evsel,
struct perf_sample *sample)
{
void *addr;
perf lock: Fix state machine to recognize lock sequence Previous state machine of perf lock was really broken. This patch improves it a little. This patch prepares the list of state machine that represents lock sequences for each threads. These state machines can be one of these sequences: 1) acquire -> acquired -> release 2) acquire -> contended -> acquired -> release 3) acquire (w/ try) -> release 4) acquire (w/ read) -> release The case of 4) is a little special. Double acquire of read lock is allowed, so the state machine counts read lock number, and permits double acquire and release. But, things are not so simple. Something in my model is still wrong. I counted the number of lock instances with bad sequence, and ratio is like this (case of tracing whoami): bad:233, total:2279 version 2: * threads are now identified with tid, not pid * prepared SEQ_STATE_READ_ACQUIRED for read lock. * bunch of struct lock_seq_stat is now linked list * debug information enhanced (this have to be removed someday) e.g. | === output for debug=== | | bad:233, total:2279 | bad rate:0.000000 | histogram of events caused bad sequence | acquire: 165 | acquired: 0 | contended: 0 | release: 68 Signed-off-by: Hitoshi Mitake <mitake@dcl.info.waseda.ac.jp> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Jens Axboe <jens.axboe@oracle.com> Cc: Jason Baron <jbaron@redhat.com> Cc: Xiao Guangrong <xiaoguangrong@cn.fujitsu.com> Cc: Ingo Molnar <mingo@elte.hu> LKML-Reference: <1271852634-9351-1-git-send-email-mitake@dcl.info.waseda.ac.jp> [rename SEQ_STATE_UNINITED to SEQ_STATE_UNINITIALIZED] Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>
2010-04-21 20:23:54 +08:00
struct lock_stat *ls;
struct thread_stat *ts;
struct lock_seq_stat *seq;
const char *name = perf_evsel__strval(evsel, sample, "name");
u64 tmp = perf_evsel__intval(evsel, sample, "lockdep_addr");
memcpy(&addr, &tmp, sizeof(void *));
ls = lock_stat_findnew(addr, name);
if (!ls)
return -ENOMEM;
perf lock: Fix state machine to recognize lock sequence Previous state machine of perf lock was really broken. This patch improves it a little. This patch prepares the list of state machine that represents lock sequences for each threads. These state machines can be one of these sequences: 1) acquire -> acquired -> release 2) acquire -> contended -> acquired -> release 3) acquire (w/ try) -> release 4) acquire (w/ read) -> release The case of 4) is a little special. Double acquire of read lock is allowed, so the state machine counts read lock number, and permits double acquire and release. But, things are not so simple. Something in my model is still wrong. I counted the number of lock instances with bad sequence, and ratio is like this (case of tracing whoami): bad:233, total:2279 version 2: * threads are now identified with tid, not pid * prepared SEQ_STATE_READ_ACQUIRED for read lock. * bunch of struct lock_seq_stat is now linked list * debug information enhanced (this have to be removed someday) e.g. | === output for debug=== | | bad:233, total:2279 | bad rate:0.000000 | histogram of events caused bad sequence | acquire: 165 | acquired: 0 | contended: 0 | release: 68 Signed-off-by: Hitoshi Mitake <mitake@dcl.info.waseda.ac.jp> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Jens Axboe <jens.axboe@oracle.com> Cc: Jason Baron <jbaron@redhat.com> Cc: Xiao Guangrong <xiaoguangrong@cn.fujitsu.com> Cc: Ingo Molnar <mingo@elte.hu> LKML-Reference: <1271852634-9351-1-git-send-email-mitake@dcl.info.waseda.ac.jp> [rename SEQ_STATE_UNINITED to SEQ_STATE_UNINITIALIZED] Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>
2010-04-21 20:23:54 +08:00
if (ls->discard)
return 0;
ts = thread_stat_findnew(sample->tid);
if (!ts)
return -ENOMEM;
seq = get_seq(ts, addr);
if (!seq)
return -ENOMEM;
perf lock: Fix state machine to recognize lock sequence Previous state machine of perf lock was really broken. This patch improves it a little. This patch prepares the list of state machine that represents lock sequences for each threads. These state machines can be one of these sequences: 1) acquire -> acquired -> release 2) acquire -> contended -> acquired -> release 3) acquire (w/ try) -> release 4) acquire (w/ read) -> release The case of 4) is a little special. Double acquire of read lock is allowed, so the state machine counts read lock number, and permits double acquire and release. But, things are not so simple. Something in my model is still wrong. I counted the number of lock instances with bad sequence, and ratio is like this (case of tracing whoami): bad:233, total:2279 version 2: * threads are now identified with tid, not pid * prepared SEQ_STATE_READ_ACQUIRED for read lock. * bunch of struct lock_seq_stat is now linked list * debug information enhanced (this have to be removed someday) e.g. | === output for debug=== | | bad:233, total:2279 | bad rate:0.000000 | histogram of events caused bad sequence | acquire: 165 | acquired: 0 | contended: 0 | release: 68 Signed-off-by: Hitoshi Mitake <mitake@dcl.info.waseda.ac.jp> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Jens Axboe <jens.axboe@oracle.com> Cc: Jason Baron <jbaron@redhat.com> Cc: Xiao Guangrong <xiaoguangrong@cn.fujitsu.com> Cc: Ingo Molnar <mingo@elte.hu> LKML-Reference: <1271852634-9351-1-git-send-email-mitake@dcl.info.waseda.ac.jp> [rename SEQ_STATE_UNINITED to SEQ_STATE_UNINITIALIZED] Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>
2010-04-21 20:23:54 +08:00
switch (seq->state) {
case SEQ_STATE_UNINITIALIZED:
goto end;
case SEQ_STATE_ACQUIRED:
break;
case SEQ_STATE_READ_ACQUIRED:
seq->read_count--;
BUG_ON(seq->read_count < 0);
if (!seq->read_count) {
ls->nr_release++;
goto end;
}
perf lock: Fix state machine to recognize lock sequence Previous state machine of perf lock was really broken. This patch improves it a little. This patch prepares the list of state machine that represents lock sequences for each threads. These state machines can be one of these sequences: 1) acquire -> acquired -> release 2) acquire -> contended -> acquired -> release 3) acquire (w/ try) -> release 4) acquire (w/ read) -> release The case of 4) is a little special. Double acquire of read lock is allowed, so the state machine counts read lock number, and permits double acquire and release. But, things are not so simple. Something in my model is still wrong. I counted the number of lock instances with bad sequence, and ratio is like this (case of tracing whoami): bad:233, total:2279 version 2: * threads are now identified with tid, not pid * prepared SEQ_STATE_READ_ACQUIRED for read lock. * bunch of struct lock_seq_stat is now linked list * debug information enhanced (this have to be removed someday) e.g. | === output for debug=== | | bad:233, total:2279 | bad rate:0.000000 | histogram of events caused bad sequence | acquire: 165 | acquired: 0 | contended: 0 | release: 68 Signed-off-by: Hitoshi Mitake <mitake@dcl.info.waseda.ac.jp> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Jens Axboe <jens.axboe@oracle.com> Cc: Jason Baron <jbaron@redhat.com> Cc: Xiao Guangrong <xiaoguangrong@cn.fujitsu.com> Cc: Ingo Molnar <mingo@elte.hu> LKML-Reference: <1271852634-9351-1-git-send-email-mitake@dcl.info.waseda.ac.jp> [rename SEQ_STATE_UNINITED to SEQ_STATE_UNINITIALIZED] Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>
2010-04-21 20:23:54 +08:00
break;
case SEQ_STATE_ACQUIRING:
case SEQ_STATE_CONTENDED:
case SEQ_STATE_RELEASED:
/* broken lock sequence, discard it */
ls->discard = 1;
bad_hist[BROKEN_RELEASE]++;
perf lock: Fix state machine to recognize lock sequence Previous state machine of perf lock was really broken. This patch improves it a little. This patch prepares the list of state machine that represents lock sequences for each threads. These state machines can be one of these sequences: 1) acquire -> acquired -> release 2) acquire -> contended -> acquired -> release 3) acquire (w/ try) -> release 4) acquire (w/ read) -> release The case of 4) is a little special. Double acquire of read lock is allowed, so the state machine counts read lock number, and permits double acquire and release. But, things are not so simple. Something in my model is still wrong. I counted the number of lock instances with bad sequence, and ratio is like this (case of tracing whoami): bad:233, total:2279 version 2: * threads are now identified with tid, not pid * prepared SEQ_STATE_READ_ACQUIRED for read lock. * bunch of struct lock_seq_stat is now linked list * debug information enhanced (this have to be removed someday) e.g. | === output for debug=== | | bad:233, total:2279 | bad rate:0.000000 | histogram of events caused bad sequence | acquire: 165 | acquired: 0 | contended: 0 | release: 68 Signed-off-by: Hitoshi Mitake <mitake@dcl.info.waseda.ac.jp> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Jens Axboe <jens.axboe@oracle.com> Cc: Jason Baron <jbaron@redhat.com> Cc: Xiao Guangrong <xiaoguangrong@cn.fujitsu.com> Cc: Ingo Molnar <mingo@elte.hu> LKML-Reference: <1271852634-9351-1-git-send-email-mitake@dcl.info.waseda.ac.jp> [rename SEQ_STATE_UNINITED to SEQ_STATE_UNINITIALIZED] Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>
2010-04-21 20:23:54 +08:00
goto free_seq;
default:
perf lock: Fix state machine to recognize lock sequence Previous state machine of perf lock was really broken. This patch improves it a little. This patch prepares the list of state machine that represents lock sequences for each threads. These state machines can be one of these sequences: 1) acquire -> acquired -> release 2) acquire -> contended -> acquired -> release 3) acquire (w/ try) -> release 4) acquire (w/ read) -> release The case of 4) is a little special. Double acquire of read lock is allowed, so the state machine counts read lock number, and permits double acquire and release. But, things are not so simple. Something in my model is still wrong. I counted the number of lock instances with bad sequence, and ratio is like this (case of tracing whoami): bad:233, total:2279 version 2: * threads are now identified with tid, not pid * prepared SEQ_STATE_READ_ACQUIRED for read lock. * bunch of struct lock_seq_stat is now linked list * debug information enhanced (this have to be removed someday) e.g. | === output for debug=== | | bad:233, total:2279 | bad rate:0.000000 | histogram of events caused bad sequence | acquire: 165 | acquired: 0 | contended: 0 | release: 68 Signed-off-by: Hitoshi Mitake <mitake@dcl.info.waseda.ac.jp> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Jens Axboe <jens.axboe@oracle.com> Cc: Jason Baron <jbaron@redhat.com> Cc: Xiao Guangrong <xiaoguangrong@cn.fujitsu.com> Cc: Ingo Molnar <mingo@elte.hu> LKML-Reference: <1271852634-9351-1-git-send-email-mitake@dcl.info.waseda.ac.jp> [rename SEQ_STATE_UNINITED to SEQ_STATE_UNINITIALIZED] Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>
2010-04-21 20:23:54 +08:00
BUG_ON("Unknown state of lock sequence found!\n");
break;
}
perf lock: Fix state machine to recognize lock sequence Previous state machine of perf lock was really broken. This patch improves it a little. This patch prepares the list of state machine that represents lock sequences for each threads. These state machines can be one of these sequences: 1) acquire -> acquired -> release 2) acquire -> contended -> acquired -> release 3) acquire (w/ try) -> release 4) acquire (w/ read) -> release The case of 4) is a little special. Double acquire of read lock is allowed, so the state machine counts read lock number, and permits double acquire and release. But, things are not so simple. Something in my model is still wrong. I counted the number of lock instances with bad sequence, and ratio is like this (case of tracing whoami): bad:233, total:2279 version 2: * threads are now identified with tid, not pid * prepared SEQ_STATE_READ_ACQUIRED for read lock. * bunch of struct lock_seq_stat is now linked list * debug information enhanced (this have to be removed someday) e.g. | === output for debug=== | | bad:233, total:2279 | bad rate:0.000000 | histogram of events caused bad sequence | acquire: 165 | acquired: 0 | contended: 0 | release: 68 Signed-off-by: Hitoshi Mitake <mitake@dcl.info.waseda.ac.jp> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Jens Axboe <jens.axboe@oracle.com> Cc: Jason Baron <jbaron@redhat.com> Cc: Xiao Guangrong <xiaoguangrong@cn.fujitsu.com> Cc: Ingo Molnar <mingo@elte.hu> LKML-Reference: <1271852634-9351-1-git-send-email-mitake@dcl.info.waseda.ac.jp> [rename SEQ_STATE_UNINITED to SEQ_STATE_UNINITIALIZED] Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>
2010-04-21 20:23:54 +08:00
ls->nr_release++;
free_seq:
list_del(&seq->list);
free(seq);
end:
return 0;
}
/* lock oriented handlers */
/* TODO: handlers for CPU oriented, thread oriented */
static struct trace_lock_handler report_lock_ops = {
.acquire_event = report_lock_acquire_event,
.acquired_event = report_lock_acquired_event,
.contended_event = report_lock_contended_event,
.release_event = report_lock_release_event,
};
static struct trace_lock_handler *trace_handler;
static int perf_evsel__process_lock_acquire(struct perf_evsel *evsel,
struct perf_sample *sample)
{
if (trace_handler->acquire_event)
return trace_handler->acquire_event(evsel, sample);
return 0;
}
static int perf_evsel__process_lock_acquired(struct perf_evsel *evsel,
struct perf_sample *sample)
{
if (trace_handler->acquired_event)
return trace_handler->acquired_event(evsel, sample);
return 0;
}
static int perf_evsel__process_lock_contended(struct perf_evsel *evsel,
struct perf_sample *sample)
{
if (trace_handler->contended_event)
return trace_handler->contended_event(evsel, sample);
return 0;
}
static int perf_evsel__process_lock_release(struct perf_evsel *evsel,
struct perf_sample *sample)
{
if (trace_handler->release_event)
return trace_handler->release_event(evsel, sample);
return 0;
}
static void print_bad_events(int bad, int total)
{
/* Output for debug, this have to be removed */
int i;
const char *name[4] =
{ "acquire", "acquired", "contended", "release" };
pr_info("\n=== output for debug===\n\n");
pr_info("bad: %d, total: %d\n", bad, total);
pr_info("bad rate: %.2f %%\n", (double)bad / (double)total * 100);
pr_info("histogram of events caused bad sequence\n");
for (i = 0; i < BROKEN_MAX; i++)
pr_info(" %10s: %d\n", name[i], bad_hist[i]);
}
/* TODO: various way to print, coloring, nano or milli sec */
static void print_result(void)
{
struct lock_stat *st;
char cut_name[20];
perf lock: Fix state machine to recognize lock sequence Previous state machine of perf lock was really broken. This patch improves it a little. This patch prepares the list of state machine that represents lock sequences for each threads. These state machines can be one of these sequences: 1) acquire -> acquired -> release 2) acquire -> contended -> acquired -> release 3) acquire (w/ try) -> release 4) acquire (w/ read) -> release The case of 4) is a little special. Double acquire of read lock is allowed, so the state machine counts read lock number, and permits double acquire and release. But, things are not so simple. Something in my model is still wrong. I counted the number of lock instances with bad sequence, and ratio is like this (case of tracing whoami): bad:233, total:2279 version 2: * threads are now identified with tid, not pid * prepared SEQ_STATE_READ_ACQUIRED for read lock. * bunch of struct lock_seq_stat is now linked list * debug information enhanced (this have to be removed someday) e.g. | === output for debug=== | | bad:233, total:2279 | bad rate:0.000000 | histogram of events caused bad sequence | acquire: 165 | acquired: 0 | contended: 0 | release: 68 Signed-off-by: Hitoshi Mitake <mitake@dcl.info.waseda.ac.jp> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Jens Axboe <jens.axboe@oracle.com> Cc: Jason Baron <jbaron@redhat.com> Cc: Xiao Guangrong <xiaoguangrong@cn.fujitsu.com> Cc: Ingo Molnar <mingo@elte.hu> LKML-Reference: <1271852634-9351-1-git-send-email-mitake@dcl.info.waseda.ac.jp> [rename SEQ_STATE_UNINITED to SEQ_STATE_UNINITIALIZED] Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>
2010-04-21 20:23:54 +08:00
int bad, total;
perf lock: Add "info" subcommand for dumping misc information This adds the "info" subcommand to perf lock which can be used to dump metadata like threads or addresses of lock instances. "map" was removed because info should do the work for it. This will be useful not only for debugging but also for ordinary analyzing. v2: adding example of usage % sudo ./perf lock info -t | Thread ID: comm | 0: swapper | 1: init | 18: migration/5 | 29: events/2 | 32: events/5 | 33: events/6 ... % sudo ./perf lock info -m | Address of instance: name of class | 0xffff8800b95adae0: &(&sighand->siglock)->rlock | 0xffff8800bbb41ae0: &(&sighand->siglock)->rlock | 0xffff8800bf165ae0: &(&sighand->siglock)->rlock | 0xffff8800b9576a98: &p->cred_guard_mutex | 0xffff8800bb890a08: &(&p->alloc_lock)->rlock | 0xffff8800b9522a08: &(&p->alloc_lock)->rlock | 0xffff8800bb8aaa08: &(&p->alloc_lock)->rlock | 0xffff8800bba72a08: &(&p->alloc_lock)->rlock | 0xffff8800bf18ea08: &(&p->alloc_lock)->rlock | 0xffff8800b8a0d8a0: &(&ip->i_lock)->mr_lock | 0xffff88009bf818a0: &(&ip->i_lock)->mr_lock | 0xffff88004c66b8a0: &(&ip->i_lock)->mr_lock | 0xffff8800bb6478a0: &(shost->host_lock)->rlock v3: fixed some problems Frederic pointed out * better rbtree tracking in dump_threads() * removed printf() and used pr_info() and pr_debug() Signed-off-by: Hitoshi Mitake <mitake@dcl.info.waseda.ac.jp> Cc: Ingo Molnar <mingo@elte.hu> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Jens Axboe <jens.axboe@oracle.com> Cc: Jason Baron <jbaron@redhat.com> Cc: Xiao Guangrong <xiaoguangrong@cn.fujitsu.com> LKML-Reference: <1272863520-16179-1-git-send-email-mitake@dcl.info.waseda.ac.jp> Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>
2010-05-03 13:12:00 +08:00
pr_info("%20s ", "Name");
pr_info("%10s ", "acquired");
pr_info("%10s ", "contended");
pr_info("%15s ", "avg wait (ns)");
perf lock: Add "info" subcommand for dumping misc information This adds the "info" subcommand to perf lock which can be used to dump metadata like threads or addresses of lock instances. "map" was removed because info should do the work for it. This will be useful not only for debugging but also for ordinary analyzing. v2: adding example of usage % sudo ./perf lock info -t | Thread ID: comm | 0: swapper | 1: init | 18: migration/5 | 29: events/2 | 32: events/5 | 33: events/6 ... % sudo ./perf lock info -m | Address of instance: name of class | 0xffff8800b95adae0: &(&sighand->siglock)->rlock | 0xffff8800bbb41ae0: &(&sighand->siglock)->rlock | 0xffff8800bf165ae0: &(&sighand->siglock)->rlock | 0xffff8800b9576a98: &p->cred_guard_mutex | 0xffff8800bb890a08: &(&p->alloc_lock)->rlock | 0xffff8800b9522a08: &(&p->alloc_lock)->rlock | 0xffff8800bb8aaa08: &(&p->alloc_lock)->rlock | 0xffff8800bba72a08: &(&p->alloc_lock)->rlock | 0xffff8800bf18ea08: &(&p->alloc_lock)->rlock | 0xffff8800b8a0d8a0: &(&ip->i_lock)->mr_lock | 0xffff88009bf818a0: &(&ip->i_lock)->mr_lock | 0xffff88004c66b8a0: &(&ip->i_lock)->mr_lock | 0xffff8800bb6478a0: &(shost->host_lock)->rlock v3: fixed some problems Frederic pointed out * better rbtree tracking in dump_threads() * removed printf() and used pr_info() and pr_debug() Signed-off-by: Hitoshi Mitake <mitake@dcl.info.waseda.ac.jp> Cc: Ingo Molnar <mingo@elte.hu> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Jens Axboe <jens.axboe@oracle.com> Cc: Jason Baron <jbaron@redhat.com> Cc: Xiao Guangrong <xiaoguangrong@cn.fujitsu.com> LKML-Reference: <1272863520-16179-1-git-send-email-mitake@dcl.info.waseda.ac.jp> Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>
2010-05-03 13:12:00 +08:00
pr_info("%15s ", "total wait (ns)");
pr_info("%15s ", "max wait (ns)");
pr_info("%15s ", "min wait (ns)");
perf lock: Add "info" subcommand for dumping misc information This adds the "info" subcommand to perf lock which can be used to dump metadata like threads or addresses of lock instances. "map" was removed because info should do the work for it. This will be useful not only for debugging but also for ordinary analyzing. v2: adding example of usage % sudo ./perf lock info -t | Thread ID: comm | 0: swapper | 1: init | 18: migration/5 | 29: events/2 | 32: events/5 | 33: events/6 ... % sudo ./perf lock info -m | Address of instance: name of class | 0xffff8800b95adae0: &(&sighand->siglock)->rlock | 0xffff8800bbb41ae0: &(&sighand->siglock)->rlock | 0xffff8800bf165ae0: &(&sighand->siglock)->rlock | 0xffff8800b9576a98: &p->cred_guard_mutex | 0xffff8800bb890a08: &(&p->alloc_lock)->rlock | 0xffff8800b9522a08: &(&p->alloc_lock)->rlock | 0xffff8800bb8aaa08: &(&p->alloc_lock)->rlock | 0xffff8800bba72a08: &(&p->alloc_lock)->rlock | 0xffff8800bf18ea08: &(&p->alloc_lock)->rlock | 0xffff8800b8a0d8a0: &(&ip->i_lock)->mr_lock | 0xffff88009bf818a0: &(&ip->i_lock)->mr_lock | 0xffff88004c66b8a0: &(&ip->i_lock)->mr_lock | 0xffff8800bb6478a0: &(shost->host_lock)->rlock v3: fixed some problems Frederic pointed out * better rbtree tracking in dump_threads() * removed printf() and used pr_info() and pr_debug() Signed-off-by: Hitoshi Mitake <mitake@dcl.info.waseda.ac.jp> Cc: Ingo Molnar <mingo@elte.hu> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Jens Axboe <jens.axboe@oracle.com> Cc: Jason Baron <jbaron@redhat.com> Cc: Xiao Guangrong <xiaoguangrong@cn.fujitsu.com> LKML-Reference: <1272863520-16179-1-git-send-email-mitake@dcl.info.waseda.ac.jp> Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>
2010-05-03 13:12:00 +08:00
pr_info("\n\n");
perf lock: Fix state machine to recognize lock sequence Previous state machine of perf lock was really broken. This patch improves it a little. This patch prepares the list of state machine that represents lock sequences for each threads. These state machines can be one of these sequences: 1) acquire -> acquired -> release 2) acquire -> contended -> acquired -> release 3) acquire (w/ try) -> release 4) acquire (w/ read) -> release The case of 4) is a little special. Double acquire of read lock is allowed, so the state machine counts read lock number, and permits double acquire and release. But, things are not so simple. Something in my model is still wrong. I counted the number of lock instances with bad sequence, and ratio is like this (case of tracing whoami): bad:233, total:2279 version 2: * threads are now identified with tid, not pid * prepared SEQ_STATE_READ_ACQUIRED for read lock. * bunch of struct lock_seq_stat is now linked list * debug information enhanced (this have to be removed someday) e.g. | === output for debug=== | | bad:233, total:2279 | bad rate:0.000000 | histogram of events caused bad sequence | acquire: 165 | acquired: 0 | contended: 0 | release: 68 Signed-off-by: Hitoshi Mitake <mitake@dcl.info.waseda.ac.jp> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Jens Axboe <jens.axboe@oracle.com> Cc: Jason Baron <jbaron@redhat.com> Cc: Xiao Guangrong <xiaoguangrong@cn.fujitsu.com> Cc: Ingo Molnar <mingo@elte.hu> LKML-Reference: <1271852634-9351-1-git-send-email-mitake@dcl.info.waseda.ac.jp> [rename SEQ_STATE_UNINITED to SEQ_STATE_UNINITIALIZED] Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>
2010-04-21 20:23:54 +08:00
bad = total = 0;
while ((st = pop_from_result())) {
perf lock: Fix state machine to recognize lock sequence Previous state machine of perf lock was really broken. This patch improves it a little. This patch prepares the list of state machine that represents lock sequences for each threads. These state machines can be one of these sequences: 1) acquire -> acquired -> release 2) acquire -> contended -> acquired -> release 3) acquire (w/ try) -> release 4) acquire (w/ read) -> release The case of 4) is a little special. Double acquire of read lock is allowed, so the state machine counts read lock number, and permits double acquire and release. But, things are not so simple. Something in my model is still wrong. I counted the number of lock instances with bad sequence, and ratio is like this (case of tracing whoami): bad:233, total:2279 version 2: * threads are now identified with tid, not pid * prepared SEQ_STATE_READ_ACQUIRED for read lock. * bunch of struct lock_seq_stat is now linked list * debug information enhanced (this have to be removed someday) e.g. | === output for debug=== | | bad:233, total:2279 | bad rate:0.000000 | histogram of events caused bad sequence | acquire: 165 | acquired: 0 | contended: 0 | release: 68 Signed-off-by: Hitoshi Mitake <mitake@dcl.info.waseda.ac.jp> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Jens Axboe <jens.axboe@oracle.com> Cc: Jason Baron <jbaron@redhat.com> Cc: Xiao Guangrong <xiaoguangrong@cn.fujitsu.com> Cc: Ingo Molnar <mingo@elte.hu> LKML-Reference: <1271852634-9351-1-git-send-email-mitake@dcl.info.waseda.ac.jp> [rename SEQ_STATE_UNINITED to SEQ_STATE_UNINITIALIZED] Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>
2010-04-21 20:23:54 +08:00
total++;
if (st->discard) {
bad++;
continue;
}
bzero(cut_name, 20);
if (strlen(st->name) < 16) {
/* output raw name */
perf lock: Add "info" subcommand for dumping misc information This adds the "info" subcommand to perf lock which can be used to dump metadata like threads or addresses of lock instances. "map" was removed because info should do the work for it. This will be useful not only for debugging but also for ordinary analyzing. v2: adding example of usage % sudo ./perf lock info -t | Thread ID: comm | 0: swapper | 1: init | 18: migration/5 | 29: events/2 | 32: events/5 | 33: events/6 ... % sudo ./perf lock info -m | Address of instance: name of class | 0xffff8800b95adae0: &(&sighand->siglock)->rlock | 0xffff8800bbb41ae0: &(&sighand->siglock)->rlock | 0xffff8800bf165ae0: &(&sighand->siglock)->rlock | 0xffff8800b9576a98: &p->cred_guard_mutex | 0xffff8800bb890a08: &(&p->alloc_lock)->rlock | 0xffff8800b9522a08: &(&p->alloc_lock)->rlock | 0xffff8800bb8aaa08: &(&p->alloc_lock)->rlock | 0xffff8800bba72a08: &(&p->alloc_lock)->rlock | 0xffff8800bf18ea08: &(&p->alloc_lock)->rlock | 0xffff8800b8a0d8a0: &(&ip->i_lock)->mr_lock | 0xffff88009bf818a0: &(&ip->i_lock)->mr_lock | 0xffff88004c66b8a0: &(&ip->i_lock)->mr_lock | 0xffff8800bb6478a0: &(shost->host_lock)->rlock v3: fixed some problems Frederic pointed out * better rbtree tracking in dump_threads() * removed printf() and used pr_info() and pr_debug() Signed-off-by: Hitoshi Mitake <mitake@dcl.info.waseda.ac.jp> Cc: Ingo Molnar <mingo@elte.hu> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Jens Axboe <jens.axboe@oracle.com> Cc: Jason Baron <jbaron@redhat.com> Cc: Xiao Guangrong <xiaoguangrong@cn.fujitsu.com> LKML-Reference: <1272863520-16179-1-git-send-email-mitake@dcl.info.waseda.ac.jp> Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>
2010-05-03 13:12:00 +08:00
pr_info("%20s ", st->name);
} else {
strncpy(cut_name, st->name, 16);
cut_name[16] = '.';
cut_name[17] = '.';
cut_name[18] = '.';
cut_name[19] = '\0';
/* cut off name for saving output style */
perf lock: Add "info" subcommand for dumping misc information This adds the "info" subcommand to perf lock which can be used to dump metadata like threads or addresses of lock instances. "map" was removed because info should do the work for it. This will be useful not only for debugging but also for ordinary analyzing. v2: adding example of usage % sudo ./perf lock info -t | Thread ID: comm | 0: swapper | 1: init | 18: migration/5 | 29: events/2 | 32: events/5 | 33: events/6 ... % sudo ./perf lock info -m | Address of instance: name of class | 0xffff8800b95adae0: &(&sighand->siglock)->rlock | 0xffff8800bbb41ae0: &(&sighand->siglock)->rlock | 0xffff8800bf165ae0: &(&sighand->siglock)->rlock | 0xffff8800b9576a98: &p->cred_guard_mutex | 0xffff8800bb890a08: &(&p->alloc_lock)->rlock | 0xffff8800b9522a08: &(&p->alloc_lock)->rlock | 0xffff8800bb8aaa08: &(&p->alloc_lock)->rlock | 0xffff8800bba72a08: &(&p->alloc_lock)->rlock | 0xffff8800bf18ea08: &(&p->alloc_lock)->rlock | 0xffff8800b8a0d8a0: &(&ip->i_lock)->mr_lock | 0xffff88009bf818a0: &(&ip->i_lock)->mr_lock | 0xffff88004c66b8a0: &(&ip->i_lock)->mr_lock | 0xffff8800bb6478a0: &(shost->host_lock)->rlock v3: fixed some problems Frederic pointed out * better rbtree tracking in dump_threads() * removed printf() and used pr_info() and pr_debug() Signed-off-by: Hitoshi Mitake <mitake@dcl.info.waseda.ac.jp> Cc: Ingo Molnar <mingo@elte.hu> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Jens Axboe <jens.axboe@oracle.com> Cc: Jason Baron <jbaron@redhat.com> Cc: Xiao Guangrong <xiaoguangrong@cn.fujitsu.com> LKML-Reference: <1272863520-16179-1-git-send-email-mitake@dcl.info.waseda.ac.jp> Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>
2010-05-03 13:12:00 +08:00
pr_info("%20s ", cut_name);
}
perf lock: Add "info" subcommand for dumping misc information This adds the "info" subcommand to perf lock which can be used to dump metadata like threads or addresses of lock instances. "map" was removed because info should do the work for it. This will be useful not only for debugging but also for ordinary analyzing. v2: adding example of usage % sudo ./perf lock info -t | Thread ID: comm | 0: swapper | 1: init | 18: migration/5 | 29: events/2 | 32: events/5 | 33: events/6 ... % sudo ./perf lock info -m | Address of instance: name of class | 0xffff8800b95adae0: &(&sighand->siglock)->rlock | 0xffff8800bbb41ae0: &(&sighand->siglock)->rlock | 0xffff8800bf165ae0: &(&sighand->siglock)->rlock | 0xffff8800b9576a98: &p->cred_guard_mutex | 0xffff8800bb890a08: &(&p->alloc_lock)->rlock | 0xffff8800b9522a08: &(&p->alloc_lock)->rlock | 0xffff8800bb8aaa08: &(&p->alloc_lock)->rlock | 0xffff8800bba72a08: &(&p->alloc_lock)->rlock | 0xffff8800bf18ea08: &(&p->alloc_lock)->rlock | 0xffff8800b8a0d8a0: &(&ip->i_lock)->mr_lock | 0xffff88009bf818a0: &(&ip->i_lock)->mr_lock | 0xffff88004c66b8a0: &(&ip->i_lock)->mr_lock | 0xffff8800bb6478a0: &(shost->host_lock)->rlock v3: fixed some problems Frederic pointed out * better rbtree tracking in dump_threads() * removed printf() and used pr_info() and pr_debug() Signed-off-by: Hitoshi Mitake <mitake@dcl.info.waseda.ac.jp> Cc: Ingo Molnar <mingo@elte.hu> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Jens Axboe <jens.axboe@oracle.com> Cc: Jason Baron <jbaron@redhat.com> Cc: Xiao Guangrong <xiaoguangrong@cn.fujitsu.com> LKML-Reference: <1272863520-16179-1-git-send-email-mitake@dcl.info.waseda.ac.jp> Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>
2010-05-03 13:12:00 +08:00
pr_info("%10u ", st->nr_acquired);
pr_info("%10u ", st->nr_contended);
pr_info("%15" PRIu64 " ", st->avg_wait_time);
pr_info("%15" PRIu64 " ", st->wait_time_total);
pr_info("%15" PRIu64 " ", st->wait_time_max);
pr_info("%15" PRIu64 " ", st->wait_time_min == ULLONG_MAX ?
0 : st->wait_time_min);
perf lock: Add "info" subcommand for dumping misc information This adds the "info" subcommand to perf lock which can be used to dump metadata like threads or addresses of lock instances. "map" was removed because info should do the work for it. This will be useful not only for debugging but also for ordinary analyzing. v2: adding example of usage % sudo ./perf lock info -t | Thread ID: comm | 0: swapper | 1: init | 18: migration/5 | 29: events/2 | 32: events/5 | 33: events/6 ... % sudo ./perf lock info -m | Address of instance: name of class | 0xffff8800b95adae0: &(&sighand->siglock)->rlock | 0xffff8800bbb41ae0: &(&sighand->siglock)->rlock | 0xffff8800bf165ae0: &(&sighand->siglock)->rlock | 0xffff8800b9576a98: &p->cred_guard_mutex | 0xffff8800bb890a08: &(&p->alloc_lock)->rlock | 0xffff8800b9522a08: &(&p->alloc_lock)->rlock | 0xffff8800bb8aaa08: &(&p->alloc_lock)->rlock | 0xffff8800bba72a08: &(&p->alloc_lock)->rlock | 0xffff8800bf18ea08: &(&p->alloc_lock)->rlock | 0xffff8800b8a0d8a0: &(&ip->i_lock)->mr_lock | 0xffff88009bf818a0: &(&ip->i_lock)->mr_lock | 0xffff88004c66b8a0: &(&ip->i_lock)->mr_lock | 0xffff8800bb6478a0: &(shost->host_lock)->rlock v3: fixed some problems Frederic pointed out * better rbtree tracking in dump_threads() * removed printf() and used pr_info() and pr_debug() Signed-off-by: Hitoshi Mitake <mitake@dcl.info.waseda.ac.jp> Cc: Ingo Molnar <mingo@elte.hu> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Jens Axboe <jens.axboe@oracle.com> Cc: Jason Baron <jbaron@redhat.com> Cc: Xiao Guangrong <xiaoguangrong@cn.fujitsu.com> LKML-Reference: <1272863520-16179-1-git-send-email-mitake@dcl.info.waseda.ac.jp> Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>
2010-05-03 13:12:00 +08:00
pr_info("\n");
}
perf lock: Fix state machine to recognize lock sequence Previous state machine of perf lock was really broken. This patch improves it a little. This patch prepares the list of state machine that represents lock sequences for each threads. These state machines can be one of these sequences: 1) acquire -> acquired -> release 2) acquire -> contended -> acquired -> release 3) acquire (w/ try) -> release 4) acquire (w/ read) -> release The case of 4) is a little special. Double acquire of read lock is allowed, so the state machine counts read lock number, and permits double acquire and release. But, things are not so simple. Something in my model is still wrong. I counted the number of lock instances with bad sequence, and ratio is like this (case of tracing whoami): bad:233, total:2279 version 2: * threads are now identified with tid, not pid * prepared SEQ_STATE_READ_ACQUIRED for read lock. * bunch of struct lock_seq_stat is now linked list * debug information enhanced (this have to be removed someday) e.g. | === output for debug=== | | bad:233, total:2279 | bad rate:0.000000 | histogram of events caused bad sequence | acquire: 165 | acquired: 0 | contended: 0 | release: 68 Signed-off-by: Hitoshi Mitake <mitake@dcl.info.waseda.ac.jp> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Jens Axboe <jens.axboe@oracle.com> Cc: Jason Baron <jbaron@redhat.com> Cc: Xiao Guangrong <xiaoguangrong@cn.fujitsu.com> Cc: Ingo Molnar <mingo@elte.hu> LKML-Reference: <1271852634-9351-1-git-send-email-mitake@dcl.info.waseda.ac.jp> [rename SEQ_STATE_UNINITED to SEQ_STATE_UNINITIALIZED] Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>
2010-04-21 20:23:54 +08:00
print_bad_events(bad, total);
}
static bool info_threads, info_map;
perf lock: Add "info" subcommand for dumping misc information This adds the "info" subcommand to perf lock which can be used to dump metadata like threads or addresses of lock instances. "map" was removed because info should do the work for it. This will be useful not only for debugging but also for ordinary analyzing. v2: adding example of usage % sudo ./perf lock info -t | Thread ID: comm | 0: swapper | 1: init | 18: migration/5 | 29: events/2 | 32: events/5 | 33: events/6 ... % sudo ./perf lock info -m | Address of instance: name of class | 0xffff8800b95adae0: &(&sighand->siglock)->rlock | 0xffff8800bbb41ae0: &(&sighand->siglock)->rlock | 0xffff8800bf165ae0: &(&sighand->siglock)->rlock | 0xffff8800b9576a98: &p->cred_guard_mutex | 0xffff8800bb890a08: &(&p->alloc_lock)->rlock | 0xffff8800b9522a08: &(&p->alloc_lock)->rlock | 0xffff8800bb8aaa08: &(&p->alloc_lock)->rlock | 0xffff8800bba72a08: &(&p->alloc_lock)->rlock | 0xffff8800bf18ea08: &(&p->alloc_lock)->rlock | 0xffff8800b8a0d8a0: &(&ip->i_lock)->mr_lock | 0xffff88009bf818a0: &(&ip->i_lock)->mr_lock | 0xffff88004c66b8a0: &(&ip->i_lock)->mr_lock | 0xffff8800bb6478a0: &(shost->host_lock)->rlock v3: fixed some problems Frederic pointed out * better rbtree tracking in dump_threads() * removed printf() and used pr_info() and pr_debug() Signed-off-by: Hitoshi Mitake <mitake@dcl.info.waseda.ac.jp> Cc: Ingo Molnar <mingo@elte.hu> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Jens Axboe <jens.axboe@oracle.com> Cc: Jason Baron <jbaron@redhat.com> Cc: Xiao Guangrong <xiaoguangrong@cn.fujitsu.com> LKML-Reference: <1272863520-16179-1-git-send-email-mitake@dcl.info.waseda.ac.jp> Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>
2010-05-03 13:12:00 +08:00
static void dump_threads(void)
{
struct thread_stat *st;
struct rb_node *node;
struct thread *t;
pr_info("%10s: comm\n", "Thread ID");
node = rb_first(&thread_stats);
while (node) {
st = container_of(node, struct thread_stat, rb);
t = perf_session__findnew(session, st->tid);
pr_info("%10d: %s\n", st->tid, thread__comm_str(t));
perf lock: Add "info" subcommand for dumping misc information This adds the "info" subcommand to perf lock which can be used to dump metadata like threads or addresses of lock instances. "map" was removed because info should do the work for it. This will be useful not only for debugging but also for ordinary analyzing. v2: adding example of usage % sudo ./perf lock info -t | Thread ID: comm | 0: swapper | 1: init | 18: migration/5 | 29: events/2 | 32: events/5 | 33: events/6 ... % sudo ./perf lock info -m | Address of instance: name of class | 0xffff8800b95adae0: &(&sighand->siglock)->rlock | 0xffff8800bbb41ae0: &(&sighand->siglock)->rlock | 0xffff8800bf165ae0: &(&sighand->siglock)->rlock | 0xffff8800b9576a98: &p->cred_guard_mutex | 0xffff8800bb890a08: &(&p->alloc_lock)->rlock | 0xffff8800b9522a08: &(&p->alloc_lock)->rlock | 0xffff8800bb8aaa08: &(&p->alloc_lock)->rlock | 0xffff8800bba72a08: &(&p->alloc_lock)->rlock | 0xffff8800bf18ea08: &(&p->alloc_lock)->rlock | 0xffff8800b8a0d8a0: &(&ip->i_lock)->mr_lock | 0xffff88009bf818a0: &(&ip->i_lock)->mr_lock | 0xffff88004c66b8a0: &(&ip->i_lock)->mr_lock | 0xffff8800bb6478a0: &(shost->host_lock)->rlock v3: fixed some problems Frederic pointed out * better rbtree tracking in dump_threads() * removed printf() and used pr_info() and pr_debug() Signed-off-by: Hitoshi Mitake <mitake@dcl.info.waseda.ac.jp> Cc: Ingo Molnar <mingo@elte.hu> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Jens Axboe <jens.axboe@oracle.com> Cc: Jason Baron <jbaron@redhat.com> Cc: Xiao Guangrong <xiaoguangrong@cn.fujitsu.com> LKML-Reference: <1272863520-16179-1-git-send-email-mitake@dcl.info.waseda.ac.jp> Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>
2010-05-03 13:12:00 +08:00
node = rb_next(node);
perf machine: Protect the machine->threads with a rwlock In addition to using refcounts for the struct thread lifetime management, we need to protect access to machine->threads from concurrent access. That happens in 'perf top', where a thread processes events, inserting and deleting entries from that rb_tree while another thread decays hist_entries, that end up dropping references and ultimately deleting threads from the rb_tree and releasing its resources when no further hist_entry (or other data structures, like in 'perf sched') references it. So the rule is the same for refcounts + protected trees in the kernel, get the tree lock, find object, bump the refcount, drop the tree lock, return, use object, drop the refcount if no more use of it is needed, keep it if storing it in some other data structure, drop when releasing that data structure. I.e. pair "t = machine__find(new)_thread()" with a "thread__put(t)", and "perf_event__preprocess_sample(&al)" with "addr_location__put(&al)". The addr_location__put() one is because as we return references to several data structures, we may end up adding more reference counting for the other data structures and then we'll drop it at addr_location__put() time. Acked-by: David Ahern <dsahern@gmail.com> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Borislav Petkov <bp@suse.de> Cc: Don Zickus <dzickus@redhat.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Jiri Olsa <jolsa@redhat.com> Cc: Namhyung Kim <namhyung@kernel.org> Cc: Stephane Eranian <eranian@google.com> Link: http://lkml.kernel.org/n/tip-bs9rt4n0jw3hi9f3zxyy3xln@git.kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2015-04-07 07:43:22 +08:00
thread__put(t);
perf lock: Add "info" subcommand for dumping misc information This adds the "info" subcommand to perf lock which can be used to dump metadata like threads or addresses of lock instances. "map" was removed because info should do the work for it. This will be useful not only for debugging but also for ordinary analyzing. v2: adding example of usage % sudo ./perf lock info -t | Thread ID: comm | 0: swapper | 1: init | 18: migration/5 | 29: events/2 | 32: events/5 | 33: events/6 ... % sudo ./perf lock info -m | Address of instance: name of class | 0xffff8800b95adae0: &(&sighand->siglock)->rlock | 0xffff8800bbb41ae0: &(&sighand->siglock)->rlock | 0xffff8800bf165ae0: &(&sighand->siglock)->rlock | 0xffff8800b9576a98: &p->cred_guard_mutex | 0xffff8800bb890a08: &(&p->alloc_lock)->rlock | 0xffff8800b9522a08: &(&p->alloc_lock)->rlock | 0xffff8800bb8aaa08: &(&p->alloc_lock)->rlock | 0xffff8800bba72a08: &(&p->alloc_lock)->rlock | 0xffff8800bf18ea08: &(&p->alloc_lock)->rlock | 0xffff8800b8a0d8a0: &(&ip->i_lock)->mr_lock | 0xffff88009bf818a0: &(&ip->i_lock)->mr_lock | 0xffff88004c66b8a0: &(&ip->i_lock)->mr_lock | 0xffff8800bb6478a0: &(shost->host_lock)->rlock v3: fixed some problems Frederic pointed out * better rbtree tracking in dump_threads() * removed printf() and used pr_info() and pr_debug() Signed-off-by: Hitoshi Mitake <mitake@dcl.info.waseda.ac.jp> Cc: Ingo Molnar <mingo@elte.hu> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Jens Axboe <jens.axboe@oracle.com> Cc: Jason Baron <jbaron@redhat.com> Cc: Xiao Guangrong <xiaoguangrong@cn.fujitsu.com> LKML-Reference: <1272863520-16179-1-git-send-email-mitake@dcl.info.waseda.ac.jp> Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>
2010-05-03 13:12:00 +08:00
};
}
static void dump_map(void)
{
unsigned int i;
struct lock_stat *st;
perf lock: Add "info" subcommand for dumping misc information This adds the "info" subcommand to perf lock which can be used to dump metadata like threads or addresses of lock instances. "map" was removed because info should do the work for it. This will be useful not only for debugging but also for ordinary analyzing. v2: adding example of usage % sudo ./perf lock info -t | Thread ID: comm | 0: swapper | 1: init | 18: migration/5 | 29: events/2 | 32: events/5 | 33: events/6 ... % sudo ./perf lock info -m | Address of instance: name of class | 0xffff8800b95adae0: &(&sighand->siglock)->rlock | 0xffff8800bbb41ae0: &(&sighand->siglock)->rlock | 0xffff8800bf165ae0: &(&sighand->siglock)->rlock | 0xffff8800b9576a98: &p->cred_guard_mutex | 0xffff8800bb890a08: &(&p->alloc_lock)->rlock | 0xffff8800b9522a08: &(&p->alloc_lock)->rlock | 0xffff8800bb8aaa08: &(&p->alloc_lock)->rlock | 0xffff8800bba72a08: &(&p->alloc_lock)->rlock | 0xffff8800bf18ea08: &(&p->alloc_lock)->rlock | 0xffff8800b8a0d8a0: &(&ip->i_lock)->mr_lock | 0xffff88009bf818a0: &(&ip->i_lock)->mr_lock | 0xffff88004c66b8a0: &(&ip->i_lock)->mr_lock | 0xffff8800bb6478a0: &(shost->host_lock)->rlock v3: fixed some problems Frederic pointed out * better rbtree tracking in dump_threads() * removed printf() and used pr_info() and pr_debug() Signed-off-by: Hitoshi Mitake <mitake@dcl.info.waseda.ac.jp> Cc: Ingo Molnar <mingo@elte.hu> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Jens Axboe <jens.axboe@oracle.com> Cc: Jason Baron <jbaron@redhat.com> Cc: Xiao Guangrong <xiaoguangrong@cn.fujitsu.com> LKML-Reference: <1272863520-16179-1-git-send-email-mitake@dcl.info.waseda.ac.jp> Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>
2010-05-03 13:12:00 +08:00
pr_info("Address of instance: name of class\n");
for (i = 0; i < LOCKHASH_SIZE; i++) {
list_for_each_entry(st, &lockhash_table[i], hash_entry) {
perf lock: Add "info" subcommand for dumping misc information This adds the "info" subcommand to perf lock which can be used to dump metadata like threads or addresses of lock instances. "map" was removed because info should do the work for it. This will be useful not only for debugging but also for ordinary analyzing. v2: adding example of usage % sudo ./perf lock info -t | Thread ID: comm | 0: swapper | 1: init | 18: migration/5 | 29: events/2 | 32: events/5 | 33: events/6 ... % sudo ./perf lock info -m | Address of instance: name of class | 0xffff8800b95adae0: &(&sighand->siglock)->rlock | 0xffff8800bbb41ae0: &(&sighand->siglock)->rlock | 0xffff8800bf165ae0: &(&sighand->siglock)->rlock | 0xffff8800b9576a98: &p->cred_guard_mutex | 0xffff8800bb890a08: &(&p->alloc_lock)->rlock | 0xffff8800b9522a08: &(&p->alloc_lock)->rlock | 0xffff8800bb8aaa08: &(&p->alloc_lock)->rlock | 0xffff8800bba72a08: &(&p->alloc_lock)->rlock | 0xffff8800bf18ea08: &(&p->alloc_lock)->rlock | 0xffff8800b8a0d8a0: &(&ip->i_lock)->mr_lock | 0xffff88009bf818a0: &(&ip->i_lock)->mr_lock | 0xffff88004c66b8a0: &(&ip->i_lock)->mr_lock | 0xffff8800bb6478a0: &(shost->host_lock)->rlock v3: fixed some problems Frederic pointed out * better rbtree tracking in dump_threads() * removed printf() and used pr_info() and pr_debug() Signed-off-by: Hitoshi Mitake <mitake@dcl.info.waseda.ac.jp> Cc: Ingo Molnar <mingo@elte.hu> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Jens Axboe <jens.axboe@oracle.com> Cc: Jason Baron <jbaron@redhat.com> Cc: Xiao Guangrong <xiaoguangrong@cn.fujitsu.com> LKML-Reference: <1272863520-16179-1-git-send-email-mitake@dcl.info.waseda.ac.jp> Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>
2010-05-03 13:12:00 +08:00
pr_info(" %p: %s\n", st->addr, st->name);
}
}
}
static int dump_info(void)
perf lock: Add "info" subcommand for dumping misc information This adds the "info" subcommand to perf lock which can be used to dump metadata like threads or addresses of lock instances. "map" was removed because info should do the work for it. This will be useful not only for debugging but also for ordinary analyzing. v2: adding example of usage % sudo ./perf lock info -t | Thread ID: comm | 0: swapper | 1: init | 18: migration/5 | 29: events/2 | 32: events/5 | 33: events/6 ... % sudo ./perf lock info -m | Address of instance: name of class | 0xffff8800b95adae0: &(&sighand->siglock)->rlock | 0xffff8800bbb41ae0: &(&sighand->siglock)->rlock | 0xffff8800bf165ae0: &(&sighand->siglock)->rlock | 0xffff8800b9576a98: &p->cred_guard_mutex | 0xffff8800bb890a08: &(&p->alloc_lock)->rlock | 0xffff8800b9522a08: &(&p->alloc_lock)->rlock | 0xffff8800bb8aaa08: &(&p->alloc_lock)->rlock | 0xffff8800bba72a08: &(&p->alloc_lock)->rlock | 0xffff8800bf18ea08: &(&p->alloc_lock)->rlock | 0xffff8800b8a0d8a0: &(&ip->i_lock)->mr_lock | 0xffff88009bf818a0: &(&ip->i_lock)->mr_lock | 0xffff88004c66b8a0: &(&ip->i_lock)->mr_lock | 0xffff8800bb6478a0: &(shost->host_lock)->rlock v3: fixed some problems Frederic pointed out * better rbtree tracking in dump_threads() * removed printf() and used pr_info() and pr_debug() Signed-off-by: Hitoshi Mitake <mitake@dcl.info.waseda.ac.jp> Cc: Ingo Molnar <mingo@elte.hu> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Jens Axboe <jens.axboe@oracle.com> Cc: Jason Baron <jbaron@redhat.com> Cc: Xiao Guangrong <xiaoguangrong@cn.fujitsu.com> LKML-Reference: <1272863520-16179-1-git-send-email-mitake@dcl.info.waseda.ac.jp> Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>
2010-05-03 13:12:00 +08:00
{
int rc = 0;
perf lock: Add "info" subcommand for dumping misc information This adds the "info" subcommand to perf lock which can be used to dump metadata like threads or addresses of lock instances. "map" was removed because info should do the work for it. This will be useful not only for debugging but also for ordinary analyzing. v2: adding example of usage % sudo ./perf lock info -t | Thread ID: comm | 0: swapper | 1: init | 18: migration/5 | 29: events/2 | 32: events/5 | 33: events/6 ... % sudo ./perf lock info -m | Address of instance: name of class | 0xffff8800b95adae0: &(&sighand->siglock)->rlock | 0xffff8800bbb41ae0: &(&sighand->siglock)->rlock | 0xffff8800bf165ae0: &(&sighand->siglock)->rlock | 0xffff8800b9576a98: &p->cred_guard_mutex | 0xffff8800bb890a08: &(&p->alloc_lock)->rlock | 0xffff8800b9522a08: &(&p->alloc_lock)->rlock | 0xffff8800bb8aaa08: &(&p->alloc_lock)->rlock | 0xffff8800bba72a08: &(&p->alloc_lock)->rlock | 0xffff8800bf18ea08: &(&p->alloc_lock)->rlock | 0xffff8800b8a0d8a0: &(&ip->i_lock)->mr_lock | 0xffff88009bf818a0: &(&ip->i_lock)->mr_lock | 0xffff88004c66b8a0: &(&ip->i_lock)->mr_lock | 0xffff8800bb6478a0: &(shost->host_lock)->rlock v3: fixed some problems Frederic pointed out * better rbtree tracking in dump_threads() * removed printf() and used pr_info() and pr_debug() Signed-off-by: Hitoshi Mitake <mitake@dcl.info.waseda.ac.jp> Cc: Ingo Molnar <mingo@elte.hu> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Jens Axboe <jens.axboe@oracle.com> Cc: Jason Baron <jbaron@redhat.com> Cc: Xiao Guangrong <xiaoguangrong@cn.fujitsu.com> LKML-Reference: <1272863520-16179-1-git-send-email-mitake@dcl.info.waseda.ac.jp> Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>
2010-05-03 13:12:00 +08:00
if (info_threads)
dump_threads();
else if (info_map)
dump_map();
else {
rc = -1;
pr_err("Unknown type of information\n");
}
return rc;
perf lock: Add "info" subcommand for dumping misc information This adds the "info" subcommand to perf lock which can be used to dump metadata like threads or addresses of lock instances. "map" was removed because info should do the work for it. This will be useful not only for debugging but also for ordinary analyzing. v2: adding example of usage % sudo ./perf lock info -t | Thread ID: comm | 0: swapper | 1: init | 18: migration/5 | 29: events/2 | 32: events/5 | 33: events/6 ... % sudo ./perf lock info -m | Address of instance: name of class | 0xffff8800b95adae0: &(&sighand->siglock)->rlock | 0xffff8800bbb41ae0: &(&sighand->siglock)->rlock | 0xffff8800bf165ae0: &(&sighand->siglock)->rlock | 0xffff8800b9576a98: &p->cred_guard_mutex | 0xffff8800bb890a08: &(&p->alloc_lock)->rlock | 0xffff8800b9522a08: &(&p->alloc_lock)->rlock | 0xffff8800bb8aaa08: &(&p->alloc_lock)->rlock | 0xffff8800bba72a08: &(&p->alloc_lock)->rlock | 0xffff8800bf18ea08: &(&p->alloc_lock)->rlock | 0xffff8800b8a0d8a0: &(&ip->i_lock)->mr_lock | 0xffff88009bf818a0: &(&ip->i_lock)->mr_lock | 0xffff88004c66b8a0: &(&ip->i_lock)->mr_lock | 0xffff8800bb6478a0: &(shost->host_lock)->rlock v3: fixed some problems Frederic pointed out * better rbtree tracking in dump_threads() * removed printf() and used pr_info() and pr_debug() Signed-off-by: Hitoshi Mitake <mitake@dcl.info.waseda.ac.jp> Cc: Ingo Molnar <mingo@elte.hu> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Jens Axboe <jens.axboe@oracle.com> Cc: Jason Baron <jbaron@redhat.com> Cc: Xiao Guangrong <xiaoguangrong@cn.fujitsu.com> LKML-Reference: <1272863520-16179-1-git-send-email-mitake@dcl.info.waseda.ac.jp> Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>
2010-05-03 13:12:00 +08:00
}
typedef int (*tracepoint_handler)(struct perf_evsel *evsel,
struct perf_sample *sample);
perf tools: Use __maybe_used for unused variables perf defines both __used and __unused variables to use for marking unused variables. The variable __used is defined to __attribute__((__unused__)), which contradicts the kernel definition to __attribute__((__used__)) for new gcc versions. On Android, __used is also defined in system headers and this leads to warnings like: warning: '__used__' attribute ignored __unused is not defined in the kernel and is not a standard definition. If __unused is included everywhere instead of __used, this leads to conflicts with glibc headers, since glibc has a variables with this name in its headers. The best approach is to use __maybe_unused, the definition used in the kernel for __attribute__((unused)). In this way there is only one definition in perf sources (instead of 2 definitions that point to the same thing: __used and __unused) and it works on both Linux and Android. This patch simply replaces all instances of __used and __unused with __maybe_unused. Signed-off-by: Irina Tirdea <irina.tirdea@intel.com> Acked-by: Pekka Enberg <penberg@kernel.org> Cc: David Ahern <dsahern@gmail.com> Cc: Ingo Molnar <mingo@redhat.com> Cc: Namhyung Kim <namhyung.kim@lge.com> Cc: Paul Mackerras <paulus@samba.org> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Steven Rostedt <rostedt@goodmis.org> Link: http://lkml.kernel.org/r/1347315303-29906-7-git-send-email-irina.tirdea@intel.com [ committer note: fixed up conflict with a116e05 in builtin-sched.c ] Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2012-09-11 06:15:03 +08:00
static int process_sample_event(struct perf_tool *tool __maybe_unused,
union perf_event *event,
struct perf_sample *sample,
struct perf_evsel *evsel,
struct machine *machine)
perf: Generalize perf lock's sample event reordering to the session layer The sample events recorded by perf record are not time ordered because we have one buffer per cpu for each event (even demultiplexed per task/per cpu for task bound events). But when we read trace events we want them to be ordered by time because many state machines are involved. There are currently two ways perf tools deal with that: - use -M to multiplex every buffers (perf sched, perf kmem) But this creates a lot of contention in SMP machines on record time. - use a post-processing time reordering (perf timechart, perf lock) The reordering used by timechart is simple but doesn't scale well with huge flow of events, in terms of performance and memory use (unusable with perf lock for example). Perf lock has its own samples reordering that flushes its memory use in a regular basis and that uses a sorting based on the previous event queued (a new event to be queued is close to the previous one most of the time). This patch proposes to export perf lock's samples reordering facility to the session layer that reads the events. So if a tool wants to get ordered sample events, it needs to set its struct perf_event_ops::ordered_samples to true and that's it. This prepares tracing based perf tools to get rid of the need to use buffers multiplexing (-M) or to implement their own reordering. Also lower the flush period to 2 as it's sufficient already. Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Paul Mackerras <paulus@samba.org> Cc: Hitoshi Mitake <mitake@dcl.info.waseda.ac.jp> Cc: Ingo Molnar <mingo@elte.hu> Cc: Masami Hiramatsu <mhiramat@redhat.com> Cc: Tom Zanussi <tzanussi@gmail.com>
2010-04-24 06:04:12 +08:00
{
perf machine: Protect the machine->threads with a rwlock In addition to using refcounts for the struct thread lifetime management, we need to protect access to machine->threads from concurrent access. That happens in 'perf top', where a thread processes events, inserting and deleting entries from that rb_tree while another thread decays hist_entries, that end up dropping references and ultimately deleting threads from the rb_tree and releasing its resources when no further hist_entry (or other data structures, like in 'perf sched') references it. So the rule is the same for refcounts + protected trees in the kernel, get the tree lock, find object, bump the refcount, drop the tree lock, return, use object, drop the refcount if no more use of it is needed, keep it if storing it in some other data structure, drop when releasing that data structure. I.e. pair "t = machine__find(new)_thread()" with a "thread__put(t)", and "perf_event__preprocess_sample(&al)" with "addr_location__put(&al)". The addr_location__put() one is because as we return references to several data structures, we may end up adding more reference counting for the other data structures and then we'll drop it at addr_location__put() time. Acked-by: David Ahern <dsahern@gmail.com> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Borislav Petkov <bp@suse.de> Cc: Don Zickus <dzickus@redhat.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Jiri Olsa <jolsa@redhat.com> Cc: Namhyung Kim <namhyung@kernel.org> Cc: Stephane Eranian <eranian@google.com> Link: http://lkml.kernel.org/n/tip-bs9rt4n0jw3hi9f3zxyy3xln@git.kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2015-04-07 07:43:22 +08:00
int err = 0;
struct thread *thread = machine__findnew_thread(machine, sample->pid,
sample->tid);
perf: Generalize perf lock's sample event reordering to the session layer The sample events recorded by perf record are not time ordered because we have one buffer per cpu for each event (even demultiplexed per task/per cpu for task bound events). But when we read trace events we want them to be ordered by time because many state machines are involved. There are currently two ways perf tools deal with that: - use -M to multiplex every buffers (perf sched, perf kmem) But this creates a lot of contention in SMP machines on record time. - use a post-processing time reordering (perf timechart, perf lock) The reordering used by timechart is simple but doesn't scale well with huge flow of events, in terms of performance and memory use (unusable with perf lock for example). Perf lock has its own samples reordering that flushes its memory use in a regular basis and that uses a sorting based on the previous event queued (a new event to be queued is close to the previous one most of the time). This patch proposes to export perf lock's samples reordering facility to the session layer that reads the events. So if a tool wants to get ordered sample events, it needs to set its struct perf_event_ops::ordered_samples to true and that's it. This prepares tracing based perf tools to get rid of the need to use buffers multiplexing (-M) or to implement their own reordering. Also lower the flush period to 2 as it's sufficient already. Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Paul Mackerras <paulus@samba.org> Cc: Hitoshi Mitake <mitake@dcl.info.waseda.ac.jp> Cc: Ingo Molnar <mingo@elte.hu> Cc: Masami Hiramatsu <mhiramat@redhat.com> Cc: Tom Zanussi <tzanussi@gmail.com>
2010-04-24 06:04:12 +08:00
if (thread == NULL) {
pr_debug("problem processing %d event, skipping it.\n",
event->header.type);
perf: Generalize perf lock's sample event reordering to the session layer The sample events recorded by perf record are not time ordered because we have one buffer per cpu for each event (even demultiplexed per task/per cpu for task bound events). But when we read trace events we want them to be ordered by time because many state machines are involved. There are currently two ways perf tools deal with that: - use -M to multiplex every buffers (perf sched, perf kmem) But this creates a lot of contention in SMP machines on record time. - use a post-processing time reordering (perf timechart, perf lock) The reordering used by timechart is simple but doesn't scale well with huge flow of events, in terms of performance and memory use (unusable with perf lock for example). Perf lock has its own samples reordering that flushes its memory use in a regular basis and that uses a sorting based on the previous event queued (a new event to be queued is close to the previous one most of the time). This patch proposes to export perf lock's samples reordering facility to the session layer that reads the events. So if a tool wants to get ordered sample events, it needs to set its struct perf_event_ops::ordered_samples to true and that's it. This prepares tracing based perf tools to get rid of the need to use buffers multiplexing (-M) or to implement their own reordering. Also lower the flush period to 2 as it's sufficient already. Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Paul Mackerras <paulus@samba.org> Cc: Hitoshi Mitake <mitake@dcl.info.waseda.ac.jp> Cc: Ingo Molnar <mingo@elte.hu> Cc: Masami Hiramatsu <mhiramat@redhat.com> Cc: Tom Zanussi <tzanussi@gmail.com>
2010-04-24 06:04:12 +08:00
return -1;
}
if (evsel->handler != NULL) {
tracepoint_handler f = evsel->handler;
perf machine: Protect the machine->threads with a rwlock In addition to using refcounts for the struct thread lifetime management, we need to protect access to machine->threads from concurrent access. That happens in 'perf top', where a thread processes events, inserting and deleting entries from that rb_tree while another thread decays hist_entries, that end up dropping references and ultimately deleting threads from the rb_tree and releasing its resources when no further hist_entry (or other data structures, like in 'perf sched') references it. So the rule is the same for refcounts + protected trees in the kernel, get the tree lock, find object, bump the refcount, drop the tree lock, return, use object, drop the refcount if no more use of it is needed, keep it if storing it in some other data structure, drop when releasing that data structure. I.e. pair "t = machine__find(new)_thread()" with a "thread__put(t)", and "perf_event__preprocess_sample(&al)" with "addr_location__put(&al)". The addr_location__put() one is because as we return references to several data structures, we may end up adding more reference counting for the other data structures and then we'll drop it at addr_location__put() time. Acked-by: David Ahern <dsahern@gmail.com> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Borislav Petkov <bp@suse.de> Cc: Don Zickus <dzickus@redhat.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Jiri Olsa <jolsa@redhat.com> Cc: Namhyung Kim <namhyung@kernel.org> Cc: Stephane Eranian <eranian@google.com> Link: http://lkml.kernel.org/n/tip-bs9rt4n0jw3hi9f3zxyy3xln@git.kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2015-04-07 07:43:22 +08:00
err = f(evsel, sample);
}
perf machine: Protect the machine->threads with a rwlock In addition to using refcounts for the struct thread lifetime management, we need to protect access to machine->threads from concurrent access. That happens in 'perf top', where a thread processes events, inserting and deleting entries from that rb_tree while another thread decays hist_entries, that end up dropping references and ultimately deleting threads from the rb_tree and releasing its resources when no further hist_entry (or other data structures, like in 'perf sched') references it. So the rule is the same for refcounts + protected trees in the kernel, get the tree lock, find object, bump the refcount, drop the tree lock, return, use object, drop the refcount if no more use of it is needed, keep it if storing it in some other data structure, drop when releasing that data structure. I.e. pair "t = machine__find(new)_thread()" with a "thread__put(t)", and "perf_event__preprocess_sample(&al)" with "addr_location__put(&al)". The addr_location__put() one is because as we return references to several data structures, we may end up adding more reference counting for the other data structures and then we'll drop it at addr_location__put() time. Acked-by: David Ahern <dsahern@gmail.com> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Borislav Petkov <bp@suse.de> Cc: Don Zickus <dzickus@redhat.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Jiri Olsa <jolsa@redhat.com> Cc: Namhyung Kim <namhyung@kernel.org> Cc: Stephane Eranian <eranian@google.com> Link: http://lkml.kernel.org/n/tip-bs9rt4n0jw3hi9f3zxyy3xln@git.kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2015-04-07 07:43:22 +08:00
thread__put(thread);
return err;
perf: Generalize perf lock's sample event reordering to the session layer The sample events recorded by perf record are not time ordered because we have one buffer per cpu for each event (even demultiplexed per task/per cpu for task bound events). But when we read trace events we want them to be ordered by time because many state machines are involved. There are currently two ways perf tools deal with that: - use -M to multiplex every buffers (perf sched, perf kmem) But this creates a lot of contention in SMP machines on record time. - use a post-processing time reordering (perf timechart, perf lock) The reordering used by timechart is simple but doesn't scale well with huge flow of events, in terms of performance and memory use (unusable with perf lock for example). Perf lock has its own samples reordering that flushes its memory use in a regular basis and that uses a sorting based on the previous event queued (a new event to be queued is close to the previous one most of the time). This patch proposes to export perf lock's samples reordering facility to the session layer that reads the events. So if a tool wants to get ordered sample events, it needs to set its struct perf_event_ops::ordered_samples to true and that's it. This prepares tracing based perf tools to get rid of the need to use buffers multiplexing (-M) or to implement their own reordering. Also lower the flush period to 2 as it's sufficient already. Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Paul Mackerras <paulus@samba.org> Cc: Hitoshi Mitake <mitake@dcl.info.waseda.ac.jp> Cc: Ingo Molnar <mingo@elte.hu> Cc: Masami Hiramatsu <mhiramat@redhat.com> Cc: Tom Zanussi <tzanussi@gmail.com>
2010-04-24 06:04:12 +08:00
}
static void sort_result(void)
{
unsigned int i;
struct lock_stat *st;
for (i = 0; i < LOCKHASH_SIZE; i++) {
list_for_each_entry(st, &lockhash_table[i], hash_entry) {
insert_to_result(st, compare);
}
}
}
static const struct perf_evsel_str_handler lock_tracepoints[] = {
{ "lock:lock_acquire", perf_evsel__process_lock_acquire, }, /* CONFIG_LOCKDEP */
{ "lock:lock_acquired", perf_evsel__process_lock_acquired, }, /* CONFIG_LOCKDEP, CONFIG_LOCK_STAT */
{ "lock:lock_contended", perf_evsel__process_lock_contended, }, /* CONFIG_LOCKDEP, CONFIG_LOCK_STAT */
{ "lock:lock_release", perf_evsel__process_lock_release, }, /* CONFIG_LOCKDEP */
};
perf lock: Support using -f to override perf.data file ownership Enable perf lock to use perf.data when it is not owned by current user or root. Example: # perf lock record ls # chown Yunlong.Song:Yunlong.Song perf.data # ls -al perf.data -rw------- 1 Yunlong.Song Yunlong.Song 4880686 Apr 2 14:14 perf.data # id uid=0(root) gid=0(root) groups=0(root),64(pkcs11) Before this patch: # perf lock report File perf.data not owned by current user or root (use -f to override) Initializing perf session failed # perf lock report -f Error: unknown switch `f' usage: perf lock report [<options>] -k, --key <acquired> key for sorting (acquired / contended / avg_wait / wait_total / wait_max / wait_min) As shown above, the -f option does not work at all. After this patch: # perf lock report File perf.data not owned by current user or root (use -f to override) Initializing perf session failed # perf lock report -f Name acquired contended avg wait (ns) total wait (ns) ... &ldata->output_l... 128 0 0 0 ... &ctx->lock 114 0 0 0 ... &p->pi_lock 112 0 0 0 ... &(&pool->lock)->... 112 0 0 0 ... &(&dentry->d_loc... 70 0 0 0 ... &(&newf->file_lo... 62 0 0 0 ... &(&fs->lock)->rl... 43 0 0 0 ... ... As shown above, the -f option really works now. Signed-off-by: Yunlong Song <yunlong.song@huawei.com> Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Paul Mackerras <paulus@samba.org> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Wang Nan <wangnan0@huawei.com> Link: http://lkml.kernel.org/r/1427982439-27388-6-git-send-email-yunlong.song@huawei.com Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2015-04-02 21:47:14 +08:00
static bool force;
static int __cmd_report(bool display_info)
{
int err = -EINVAL;
struct perf_tool eops = {
.sample = process_sample_event,
.comm = perf_event__process_comm,
perf tools: Add PERF_RECORD_NAMESPACES to include namespaces related info Introduce a new option to record PERF_RECORD_NAMESPACES events emitted by the kernel when fork, clone, setns or unshare are invoked. And update perf-record documentation with the new option to record namespace events. Committer notes: Combined it with a later patch to allow printing it via 'perf report -D' and be able to test the feature introduced in this patch. Had to move here also perf_ns__name(), that was introduced in another later patch. Also used PRIu64 and PRIx64 to fix the build in some enfironments wrt: util/event.c:1129:39: error: format '%lx' expects argument of type 'long unsigned int', but argument 6 has type 'long long unsigned int' [-Werror=format=] ret += fprintf(fp, "%u/%s: %lu/0x%lx%s", idx ^ Testing it: # perf record --namespaces -a ^C[ perf record: Woken up 1 times to write data ] [ perf record: Captured and wrote 1.083 MB perf.data (423 samples) ] # # perf report -D <SNIP> 3 2028902078892 0x115140 [0xa0]: PERF_RECORD_NAMESPACES 14783/14783 - nr_namespaces: 7 [0/net: 3/0xf0000081, 1/uts: 3/0xeffffffe, 2/ipc: 3/0xefffffff, 3/pid: 3/0xeffffffc, 4/user: 3/0xeffffffd, 5/mnt: 3/0xf0000000, 6/cgroup: 3/0xeffffffb] 0x1151e0 [0x30]: event: 9 . . ... raw event: size 48 bytes . 0000: 09 00 00 00 02 00 30 00 c4 71 82 68 0c 7f 00 00 ......0..q.h.... . 0010: a9 39 00 00 a9 39 00 00 94 28 fe 63 d8 01 00 00 .9...9...(.c.... . 0020: 03 00 00 00 00 00 00 00 ce c4 02 00 00 00 00 00 ................ <SNIP> NAMESPACES events: 1 <SNIP> # Signed-off-by: Hari Bathini <hbathini@linux.vnet.ibm.com> Acked-by: Jiri Olsa <jolsa@kernel.org> Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com> Cc: Alexei Starovoitov <ast@fb.com> Cc: Ananth N Mavinakayanahalli <ananth@linux.vnet.ibm.com> Cc: Aravinda Prasad <aravinda@linux.vnet.ibm.com> Cc: Brendan Gregg <brendan.d.gregg@gmail.com> Cc: Daniel Borkmann <daniel@iogearbox.net> Cc: Eric Biederman <ebiederm@xmission.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Sargun Dhillon <sargun@sargun.me> Cc: Steven Rostedt <rostedt@goodmis.org> Link: http://lkml.kernel.org/r/148891930386.25309.18412039920746995488.stgit@hbathini.in.ibm.com Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2017-03-08 04:41:43 +08:00
.namespaces = perf_event__process_namespaces,
.ordered_events = true,
};
struct perf_data_file file = {
.path = input_name,
.mode = PERF_DATA_MODE_READ,
perf lock: Support using -f to override perf.data file ownership Enable perf lock to use perf.data when it is not owned by current user or root. Example: # perf lock record ls # chown Yunlong.Song:Yunlong.Song perf.data # ls -al perf.data -rw------- 1 Yunlong.Song Yunlong.Song 4880686 Apr 2 14:14 perf.data # id uid=0(root) gid=0(root) groups=0(root),64(pkcs11) Before this patch: # perf lock report File perf.data not owned by current user or root (use -f to override) Initializing perf session failed # perf lock report -f Error: unknown switch `f' usage: perf lock report [<options>] -k, --key <acquired> key for sorting (acquired / contended / avg_wait / wait_total / wait_max / wait_min) As shown above, the -f option does not work at all. After this patch: # perf lock report File perf.data not owned by current user or root (use -f to override) Initializing perf session failed # perf lock report -f Name acquired contended avg wait (ns) total wait (ns) ... &ldata->output_l... 128 0 0 0 ... &ctx->lock 114 0 0 0 ... &p->pi_lock 112 0 0 0 ... &(&pool->lock)->... 112 0 0 0 ... &(&dentry->d_loc... 70 0 0 0 ... &(&newf->file_lo... 62 0 0 0 ... &(&fs->lock)->rl... 43 0 0 0 ... ... As shown above, the -f option really works now. Signed-off-by: Yunlong Song <yunlong.song@huawei.com> Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Paul Mackerras <paulus@samba.org> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Wang Nan <wangnan0@huawei.com> Link: http://lkml.kernel.org/r/1427982439-27388-6-git-send-email-yunlong.song@huawei.com Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2015-04-02 21:47:14 +08:00
.force = force,
};
session = perf_session__new(&file, false, &eops);
if (!session) {
pr_err("Initializing perf session failed\n");
return -1;
}
perf tools: Check recorded kernel version when finding vmlinux Currently vmlinux_path__init() only tries to find vmlinux file from current directory, /boot and some canonical directories with version number of the running kernel. This can be a problem when reporting old data recorded on a kernel version not running currently. We can use --symfs option for this but it's annoying for user to do it always. As we already have the info in the perf.data file, it can be changed to use it for the search automatically. Before: $ perf report ... # Samples: 4K of event 'cpu-clock' # Event count (approx.): 1067250000 # # Overhead Command Shared Object Symbol # ........ .......... ................. .............................. 71.87% swapper [kernel.kallsyms] [k] recover_probed_instruction After: # Overhead Command Shared Object Symbol # ........ .......... ................. .................... 71.87% swapper [kernel.kallsyms] [k] native_safe_halt This requires to change signature of symbol__init() to receive struct perf_session_env *. Reported-by: Minchan Kim <minchan@kernel.org> Signed-off-by: Namhyung Kim <namhyung@kernel.org> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: David Ahern <dsahern@gmail.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@redhat.com> Cc: Masami Hiramatsu <masami.hiramatsu.pt@hitachi.com> Cc: Minchan Kim <minchan@kernel.org> Cc: Namhyung Kim <namhyung.kim@lge.com> Cc: Paul Mackerras <paulus@samba.org> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Stephane Eranian <eranian@google.com> Link: http://lkml.kernel.org/r/1407825645-24586-14-git-send-email-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2014-08-12 14:40:45 +08:00
symbol__init(&session->header.env);
if (!perf_session__has_traces(session, "lock record"))
goto out_delete;
if (perf_session__set_tracepoints_handlers(session, lock_tracepoints)) {
pr_err("Initializing perf session tracepoint handlers failed\n");
goto out_delete;
}
if (select_key())
goto out_delete;
err = perf_session__process_events(session);
if (err)
goto out_delete;
setup_pager();
if (display_info) /* used for info subcommand */
err = dump_info();
else {
sort_result();
print_result();
}
out_delete:
perf_session__delete(session);
return err;
}
static int __cmd_record(int argc, const char **argv)
{
const char *record_args[] = {
"record", "-R", "-m", "1024", "-c", "1",
};
unsigned int rec_argc, i, j, ret;
const char **rec_argv;
for (i = 0; i < ARRAY_SIZE(lock_tracepoints); i++) {
if (!is_valid_tracepoint(lock_tracepoints[i].name)) {
pr_err("tracepoint %s is not enabled. "
"Are CONFIG_LOCKDEP and CONFIG_LOCK_STAT enabled?\n",
lock_tracepoints[i].name);
return 1;
}
}
rec_argc = ARRAY_SIZE(record_args) + argc - 1;
/* factor of 2 is for -e in front of each tracepoint */
rec_argc += 2 * ARRAY_SIZE(lock_tracepoints);
rec_argv = calloc(rec_argc + 1, sizeof(char *));
if (!rec_argv)
return -ENOMEM;
for (i = 0; i < ARRAY_SIZE(record_args); i++)
rec_argv[i] = strdup(record_args[i]);
for (j = 0; j < ARRAY_SIZE(lock_tracepoints); j++) {
rec_argv[i++] = "-e";
rec_argv[i++] = strdup(lock_tracepoints[j].name);
}
for (j = 1; j < (unsigned int)argc; j++, i++)
rec_argv[i] = argv[j];
BUG_ON(i != rec_argc);
ret = cmd_record(i, rec_argv);
free(rec_argv);
return ret;
}
int cmd_lock(int argc, const char **argv)
{
const struct option lock_options[] = {
OPT_STRING('i', "input", &input_name, "file", "input file name"),
OPT_INCR('v', "verbose", &verbose, "be more verbose (show symbol address, etc)"),
OPT_BOOLEAN('D', "dump-raw-trace", &dump_trace, "dump raw trace in ASCII"),
OPT_BOOLEAN('f', "force", &force, "don't complain, do it"),
OPT_END()
};
const struct option info_options[] = {
OPT_BOOLEAN('t', "threads", &info_threads,
"dump thread list in perf.data"),
OPT_BOOLEAN('m', "map", &info_map,
"map of lock instances (address:name table)"),
OPT_PARENT(lock_options)
};
const struct option report_options[] = {
OPT_STRING('k', "key", &sort_key, "acquired",
"key for sorting (acquired / contended / avg_wait / wait_total / wait_max / wait_min)"),
/* TODO: type */
OPT_PARENT(lock_options)
};
const char * const info_usage[] = {
"perf lock info [<options>]",
NULL
};
const char *const lock_subcommands[] = { "record", "report", "script",
"info", NULL };
const char *lock_usage[] = {
NULL,
NULL
};
const char * const report_usage[] = {
"perf lock report [<options>]",
NULL
};
unsigned int i;
int rc = 0;
for (i = 0; i < LOCKHASH_SIZE; i++)
INIT_LIST_HEAD(lockhash_table + i);
argc = parse_options_subcommand(argc, argv, lock_options, lock_subcommands,
lock_usage, PARSE_OPT_STOP_AT_NON_OPTION);
if (!argc)
usage_with_options(lock_usage, lock_options);
if (!strncmp(argv[0], "rec", 3)) {
return __cmd_record(argc, argv);
} else if (!strncmp(argv[0], "report", 6)) {
trace_handler = &report_lock_ops;
if (argc) {
argc = parse_options(argc, argv,
report_options, report_usage, 0);
if (argc)
usage_with_options(report_usage, report_options);
}
rc = __cmd_report(false);
} else if (!strcmp(argv[0], "script")) {
/* Aliased to 'perf script' */
return cmd_script(argc, argv);
perf lock: Add "info" subcommand for dumping misc information This adds the "info" subcommand to perf lock which can be used to dump metadata like threads or addresses of lock instances. "map" was removed because info should do the work for it. This will be useful not only for debugging but also for ordinary analyzing. v2: adding example of usage % sudo ./perf lock info -t | Thread ID: comm | 0: swapper | 1: init | 18: migration/5 | 29: events/2 | 32: events/5 | 33: events/6 ... % sudo ./perf lock info -m | Address of instance: name of class | 0xffff8800b95adae0: &(&sighand->siglock)->rlock | 0xffff8800bbb41ae0: &(&sighand->siglock)->rlock | 0xffff8800bf165ae0: &(&sighand->siglock)->rlock | 0xffff8800b9576a98: &p->cred_guard_mutex | 0xffff8800bb890a08: &(&p->alloc_lock)->rlock | 0xffff8800b9522a08: &(&p->alloc_lock)->rlock | 0xffff8800bb8aaa08: &(&p->alloc_lock)->rlock | 0xffff8800bba72a08: &(&p->alloc_lock)->rlock | 0xffff8800bf18ea08: &(&p->alloc_lock)->rlock | 0xffff8800b8a0d8a0: &(&ip->i_lock)->mr_lock | 0xffff88009bf818a0: &(&ip->i_lock)->mr_lock | 0xffff88004c66b8a0: &(&ip->i_lock)->mr_lock | 0xffff8800bb6478a0: &(shost->host_lock)->rlock v3: fixed some problems Frederic pointed out * better rbtree tracking in dump_threads() * removed printf() and used pr_info() and pr_debug() Signed-off-by: Hitoshi Mitake <mitake@dcl.info.waseda.ac.jp> Cc: Ingo Molnar <mingo@elte.hu> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Jens Axboe <jens.axboe@oracle.com> Cc: Jason Baron <jbaron@redhat.com> Cc: Xiao Guangrong <xiaoguangrong@cn.fujitsu.com> LKML-Reference: <1272863520-16179-1-git-send-email-mitake@dcl.info.waseda.ac.jp> Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>
2010-05-03 13:12:00 +08:00
} else if (!strcmp(argv[0], "info")) {
if (argc) {
argc = parse_options(argc, argv,
info_options, info_usage, 0);
if (argc)
usage_with_options(info_usage, info_options);
}
/* recycling report_lock_ops */
trace_handler = &report_lock_ops;
rc = __cmd_report(true);
} else {
usage_with_options(lock_usage, lock_options);
}
return rc;
}