android_kernel_xiaomi_sm8350/tools/perf/builtin-sched.c

1503 lines
32 KiB
C
Raw Normal View History

#include "builtin.h"
#include "util/util.h"
#include "util/cache.h"
#include "util/symbol.h"
#include "util/thread.h"
#include "util/header.h"
#include "util/parse-options.h"
#include "perf.h"
#include "util/debug.h"
#include "util/trace-event.h"
#include <sys/types.h>
#define MAX_CPUS 4096
static char const *input_name = "perf.data";
static int input;
static unsigned long page_size;
static unsigned long mmap_window = 32;
static unsigned long total_comm = 0;
static struct rb_root threads;
static struct thread *last_match;
static struct perf_header *header;
static u64 sample_type;
static int replay_mode;
static int lat_mode;
/*
* Scheduler benchmarks
*/
#include <sys/resource.h>
#include <sys/types.h>
#include <sys/stat.h>
#include <sys/time.h>
#include <sys/prctl.h>
#include <linux/unistd.h>
#include <semaphore.h>
#include <pthread.h>
#include <signal.h>
#include <values.h>
#include <string.h>
#include <unistd.h>
#include <stdlib.h>
#include <assert.h>
#include <fcntl.h>
#include <time.h>
#include <math.h>
#include <stdio.h>
#define PR_SET_NAME 15 /* Set process name */
#define BUG_ON(x) assert(!(x))
perf sched: Implement the scheduling workload replay engine Integrate the schedbench.c bits with the raw trace events that we get from the perf machinery, and activate the workload replayer/simulator. Example of a captured 'make -j' workload: $ perf sched run measurement overhead: 90 nsecs sleep measurement overhead: 2724743 nsecs the run test took 1000081 nsecs the sleep test took 2981111 nsecs version = 0.5 ... nr_run_events: 70 nr_sleep_events: 66 nr_wakeup_events: 9 target-less wakeups: 71 multi-target wakeups: 47 run events optimized: 139 task 0 ( perf: 6607), nr_events: 2 task 1 ( perf: 6608), nr_events: 6 task 2 ( : 0), nr_events: 1 task 3 ( make: 6609), nr_events: 5 task 4 ( sh: 6610), nr_events: 4 task 5 ( make: 6611), nr_events: 6 task 6 ( sh: 6612), nr_events: 4 task 7 ( make: 6613), nr_events: 5 task 8 ( migration/11: 25), nr_events: 1 task 9 ( migration/13: 29), nr_events: 1 task 10 ( migration/15: 33), nr_events: 1 task 11 ( migration/9: 21), nr_events: 1 task 12 ( sh: 6614), nr_events: 4 task 13 ( make: 6615), nr_events: 5 task 14 ( sh: 6616), nr_events: 4 task 15 ( make: 6617), nr_events: 7 task 16 ( migration/3: 9), nr_events: 1 task 17 ( migration/5: 13), nr_events: 1 task 18 ( migration/7: 17), nr_events: 1 task 19 ( migration/1: 5), nr_events: 1 task 20 ( sh: 6618), nr_events: 4 task 21 ( make: 6619), nr_events: 5 task 22 ( sh: 6620), nr_events: 4 task 23 ( make: 6621), nr_events: 10 task 24 ( sh: 6623), nr_events: 3 task 25 ( gcc: 6624), nr_events: 4 task 26 ( gcc: 6625), nr_events: 4 task 27 ( gcc: 6626), nr_events: 5 task 28 ( collect2: 6627), nr_events: 5 task 29 ( sh: 6622), nr_events: 1 task 30 ( make: 6628), nr_events: 7 task 31 ( sh: 6630), nr_events: 4 task 32 ( gcc: 6631), nr_events: 4 task 33 ( sh: 6629), nr_events: 1 task 34 ( gcc: 6632), nr_events: 4 task 35 ( gcc: 6633), nr_events: 4 task 36 ( collect2: 6634), nr_events: 4 task 37 ( make: 6635), nr_events: 8 task 38 ( sh: 6637), nr_events: 4 task 39 ( sh: 6636), nr_events: 1 task 40 ( gcc: 6638), nr_events: 4 task 41 ( gcc: 6639), nr_events: 4 task 42 ( gcc: 6640), nr_events: 4 task 43 ( collect2: 6641), nr_events: 4 task 44 ( make: 6642), nr_events: 6 task 45 ( sh: 6643), nr_events: 5 task 46 ( sh: 6644), nr_events: 3 task 47 ( sh: 6645), nr_events: 4 task 48 ( make: 6646), nr_events: 6 task 49 ( sh: 6647), nr_events: 3 task 50 ( make: 6648), nr_events: 5 task 51 ( sh: 6649), nr_events: 5 task 52 ( sh: 6650), nr_events: 6 task 53 ( make: 6651), nr_events: 4 task 54 ( make: 6652), nr_events: 5 task 55 ( make: 6653), nr_events: 4 task 56 ( make: 6654), nr_events: 4 task 57 ( make: 6655), nr_events: 5 task 58 ( sh: 6656), nr_events: 4 task 59 ( gcc: 6657), nr_events: 9 task 60 ( ksoftirqd/3: 10), nr_events: 1 task 61 ( gcc: 6658), nr_events: 4 task 62 ( make: 6659), nr_events: 5 task 63 ( sh: 6660), nr_events: 3 task 64 ( gcc: 6661), nr_events: 5 task 65 ( collect2: 6662), nr_events: 4 ------------------------------------------------------------ #1 : 256.745, ravg: 256.74, cpu: 0.00 / 0.00 #2 : 439.372, ravg: 275.01, cpu: 0.00 / 0.00 #3 : 411.971, ravg: 288.70, cpu: 0.00 / 0.00 #4 : 385.500, ravg: 298.38, cpu: 0.00 / 0.00 #5 : 366.526, ravg: 305.20, cpu: 0.00 / 0.00 #6 : 381.281, ravg: 312.81, cpu: 0.00 / 0.00 #7 : 410.756, ravg: 322.60, cpu: 0.00 / 0.00 #8 : 368.009, ravg: 327.14, cpu: 0.00 / 0.00 #9 : 408.098, ravg: 335.24, cpu: 0.00 / 0.00 #10 : 368.582, ravg: 338.57, cpu: 0.00 / 0.00 I.e. we successfully analyzed the trace, replayed it via real threads and measured the replayed workload's scheduling properties. This is how it looked like in 'top' output: PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND 7164 mingo 20 0 1434m 8080 888 R 57.0 0.1 0:02.04 :perf 7165 mingo 20 0 1434m 8080 888 R 41.8 0.1 0:01.52 :perf 7228 mingo 20 0 1434m 8080 888 R 39.8 0.1 0:01.44 :gcc 7225 mingo 20 0 1434m 8080 888 R 33.8 0.1 0:01.26 :gcc 7202 mingo 20 0 1434m 8080 888 R 31.2 0.1 0:01.16 :sh 7222 mingo 20 0 1434m 8080 888 R 25.2 0.1 0:00.96 :sh 7211 mingo 20 0 1434m 8080 888 R 21.9 0.1 0:00.82 :sh 7213 mingo 20 0 1434m 8080 888 D 19.2 0.1 0:00.74 :sh 7194 mingo 20 0 1434m 8080 888 D 18.6 0.1 0:00.72 :make There's still various kinks in it - more patches to come. Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Mike Galbraith <efault@gmx.de> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> LKML-Reference: <new-submission> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-09-11 06:12:54 -04:00
#define DEBUG 0
typedef unsigned long long nsec_t;
static nsec_t run_measurement_overhead;
static nsec_t sleep_measurement_overhead;
static nsec_t get_nsecs(void)
{
struct timespec ts;
clock_gettime(CLOCK_MONOTONIC, &ts);
return ts.tv_sec * 1000000000ULL + ts.tv_nsec;
}
static void burn_nsecs(nsec_t nsecs)
{
nsec_t T0 = get_nsecs(), T1;
do {
T1 = get_nsecs();
} while (T1 + run_measurement_overhead < T0 + nsecs);
}
static void sleep_nsecs(nsec_t nsecs)
{
struct timespec ts;
ts.tv_nsec = nsecs % 999999999;
ts.tv_sec = nsecs / 999999999;
nanosleep(&ts, NULL);
}
static void calibrate_run_measurement_overhead(void)
{
nsec_t T0, T1, delta, min_delta = 1000000000ULL;
int i;
for (i = 0; i < 10; i++) {
T0 = get_nsecs();
burn_nsecs(0);
T1 = get_nsecs();
delta = T1-T0;
min_delta = min(min_delta, delta);
}
run_measurement_overhead = min_delta;
printf("run measurement overhead: %Ld nsecs\n", min_delta);
}
static void calibrate_sleep_measurement_overhead(void)
{
nsec_t T0, T1, delta, min_delta = 1000000000ULL;
int i;
for (i = 0; i < 10; i++) {
T0 = get_nsecs();
sleep_nsecs(10000);
T1 = get_nsecs();
delta = T1-T0;
min_delta = min(min_delta, delta);
}
min_delta -= 10000;
sleep_measurement_overhead = min_delta;
printf("sleep measurement overhead: %Ld nsecs\n", min_delta);
}
#define COMM_LEN 20
#define SYM_LEN 129
#define MAX_PID 65536
static unsigned long nr_tasks;
struct sched_event;
struct task_desc {
unsigned long nr;
unsigned long pid;
char comm[COMM_LEN];
unsigned long nr_events;
unsigned long curr_event;
struct sched_event **events;
pthread_t thread;
sem_t sleep_sem;
sem_t ready_for_work;
sem_t work_done_sem;
nsec_t cpu_usage;
};
enum sched_event_type {
SCHED_EVENT_RUN,
SCHED_EVENT_SLEEP,
SCHED_EVENT_WAKEUP,
};
struct sched_event {
enum sched_event_type type;
nsec_t timestamp;
nsec_t duration;
unsigned long nr;
int specific_wait;
sem_t *wait_sem;
struct task_desc *wakee;
};
static struct task_desc *pid_to_task[MAX_PID];
static struct task_desc **tasks;
static pthread_mutex_t start_work_mutex = PTHREAD_MUTEX_INITIALIZER;
static nsec_t start_time;
static pthread_mutex_t work_done_wait_mutex = PTHREAD_MUTEX_INITIALIZER;
static unsigned long nr_run_events;
static unsigned long nr_sleep_events;
static unsigned long nr_wakeup_events;
static unsigned long nr_sleep_corrections;
static unsigned long nr_run_events_optimized;
static struct sched_event *
get_new_event(struct task_desc *task, nsec_t timestamp)
{
struct sched_event *event = calloc(1, sizeof(*event));
unsigned long idx = task->nr_events;
size_t size;
event->timestamp = timestamp;
event->nr = idx;
task->nr_events++;
size = sizeof(struct sched_event *) * task->nr_events;
task->events = realloc(task->events, size);
BUG_ON(!task->events);
task->events[idx] = event;
return event;
}
static struct sched_event *last_event(struct task_desc *task)
{
if (!task->nr_events)
return NULL;
return task->events[task->nr_events - 1];
}
static void
perf sched: Implement the scheduling workload replay engine Integrate the schedbench.c bits with the raw trace events that we get from the perf machinery, and activate the workload replayer/simulator. Example of a captured 'make -j' workload: $ perf sched run measurement overhead: 90 nsecs sleep measurement overhead: 2724743 nsecs the run test took 1000081 nsecs the sleep test took 2981111 nsecs version = 0.5 ... nr_run_events: 70 nr_sleep_events: 66 nr_wakeup_events: 9 target-less wakeups: 71 multi-target wakeups: 47 run events optimized: 139 task 0 ( perf: 6607), nr_events: 2 task 1 ( perf: 6608), nr_events: 6 task 2 ( : 0), nr_events: 1 task 3 ( make: 6609), nr_events: 5 task 4 ( sh: 6610), nr_events: 4 task 5 ( make: 6611), nr_events: 6 task 6 ( sh: 6612), nr_events: 4 task 7 ( make: 6613), nr_events: 5 task 8 ( migration/11: 25), nr_events: 1 task 9 ( migration/13: 29), nr_events: 1 task 10 ( migration/15: 33), nr_events: 1 task 11 ( migration/9: 21), nr_events: 1 task 12 ( sh: 6614), nr_events: 4 task 13 ( make: 6615), nr_events: 5 task 14 ( sh: 6616), nr_events: 4 task 15 ( make: 6617), nr_events: 7 task 16 ( migration/3: 9), nr_events: 1 task 17 ( migration/5: 13), nr_events: 1 task 18 ( migration/7: 17), nr_events: 1 task 19 ( migration/1: 5), nr_events: 1 task 20 ( sh: 6618), nr_events: 4 task 21 ( make: 6619), nr_events: 5 task 22 ( sh: 6620), nr_events: 4 task 23 ( make: 6621), nr_events: 10 task 24 ( sh: 6623), nr_events: 3 task 25 ( gcc: 6624), nr_events: 4 task 26 ( gcc: 6625), nr_events: 4 task 27 ( gcc: 6626), nr_events: 5 task 28 ( collect2: 6627), nr_events: 5 task 29 ( sh: 6622), nr_events: 1 task 30 ( make: 6628), nr_events: 7 task 31 ( sh: 6630), nr_events: 4 task 32 ( gcc: 6631), nr_events: 4 task 33 ( sh: 6629), nr_events: 1 task 34 ( gcc: 6632), nr_events: 4 task 35 ( gcc: 6633), nr_events: 4 task 36 ( collect2: 6634), nr_events: 4 task 37 ( make: 6635), nr_events: 8 task 38 ( sh: 6637), nr_events: 4 task 39 ( sh: 6636), nr_events: 1 task 40 ( gcc: 6638), nr_events: 4 task 41 ( gcc: 6639), nr_events: 4 task 42 ( gcc: 6640), nr_events: 4 task 43 ( collect2: 6641), nr_events: 4 task 44 ( make: 6642), nr_events: 6 task 45 ( sh: 6643), nr_events: 5 task 46 ( sh: 6644), nr_events: 3 task 47 ( sh: 6645), nr_events: 4 task 48 ( make: 6646), nr_events: 6 task 49 ( sh: 6647), nr_events: 3 task 50 ( make: 6648), nr_events: 5 task 51 ( sh: 6649), nr_events: 5 task 52 ( sh: 6650), nr_events: 6 task 53 ( make: 6651), nr_events: 4 task 54 ( make: 6652), nr_events: 5 task 55 ( make: 6653), nr_events: 4 task 56 ( make: 6654), nr_events: 4 task 57 ( make: 6655), nr_events: 5 task 58 ( sh: 6656), nr_events: 4 task 59 ( gcc: 6657), nr_events: 9 task 60 ( ksoftirqd/3: 10), nr_events: 1 task 61 ( gcc: 6658), nr_events: 4 task 62 ( make: 6659), nr_events: 5 task 63 ( sh: 6660), nr_events: 3 task 64 ( gcc: 6661), nr_events: 5 task 65 ( collect2: 6662), nr_events: 4 ------------------------------------------------------------ #1 : 256.745, ravg: 256.74, cpu: 0.00 / 0.00 #2 : 439.372, ravg: 275.01, cpu: 0.00 / 0.00 #3 : 411.971, ravg: 288.70, cpu: 0.00 / 0.00 #4 : 385.500, ravg: 298.38, cpu: 0.00 / 0.00 #5 : 366.526, ravg: 305.20, cpu: 0.00 / 0.00 #6 : 381.281, ravg: 312.81, cpu: 0.00 / 0.00 #7 : 410.756, ravg: 322.60, cpu: 0.00 / 0.00 #8 : 368.009, ravg: 327.14, cpu: 0.00 / 0.00 #9 : 408.098, ravg: 335.24, cpu: 0.00 / 0.00 #10 : 368.582, ravg: 338.57, cpu: 0.00 / 0.00 I.e. we successfully analyzed the trace, replayed it via real threads and measured the replayed workload's scheduling properties. This is how it looked like in 'top' output: PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND 7164 mingo 20 0 1434m 8080 888 R 57.0 0.1 0:02.04 :perf 7165 mingo 20 0 1434m 8080 888 R 41.8 0.1 0:01.52 :perf 7228 mingo 20 0 1434m 8080 888 R 39.8 0.1 0:01.44 :gcc 7225 mingo 20 0 1434m 8080 888 R 33.8 0.1 0:01.26 :gcc 7202 mingo 20 0 1434m 8080 888 R 31.2 0.1 0:01.16 :sh 7222 mingo 20 0 1434m 8080 888 R 25.2 0.1 0:00.96 :sh 7211 mingo 20 0 1434m 8080 888 R 21.9 0.1 0:00.82 :sh 7213 mingo 20 0 1434m 8080 888 D 19.2 0.1 0:00.74 :sh 7194 mingo 20 0 1434m 8080 888 D 18.6 0.1 0:00.72 :make There's still various kinks in it - more patches to come. Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Mike Galbraith <efault@gmx.de> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> LKML-Reference: <new-submission> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-09-11 06:12:54 -04:00
add_sched_event_run(struct task_desc *task, nsec_t timestamp, u64 duration)
{
struct sched_event *event, *curr_event = last_event(task);
/*
perf sched: Implement the scheduling workload replay engine Integrate the schedbench.c bits with the raw trace events that we get from the perf machinery, and activate the workload replayer/simulator. Example of a captured 'make -j' workload: $ perf sched run measurement overhead: 90 nsecs sleep measurement overhead: 2724743 nsecs the run test took 1000081 nsecs the sleep test took 2981111 nsecs version = 0.5 ... nr_run_events: 70 nr_sleep_events: 66 nr_wakeup_events: 9 target-less wakeups: 71 multi-target wakeups: 47 run events optimized: 139 task 0 ( perf: 6607), nr_events: 2 task 1 ( perf: 6608), nr_events: 6 task 2 ( : 0), nr_events: 1 task 3 ( make: 6609), nr_events: 5 task 4 ( sh: 6610), nr_events: 4 task 5 ( make: 6611), nr_events: 6 task 6 ( sh: 6612), nr_events: 4 task 7 ( make: 6613), nr_events: 5 task 8 ( migration/11: 25), nr_events: 1 task 9 ( migration/13: 29), nr_events: 1 task 10 ( migration/15: 33), nr_events: 1 task 11 ( migration/9: 21), nr_events: 1 task 12 ( sh: 6614), nr_events: 4 task 13 ( make: 6615), nr_events: 5 task 14 ( sh: 6616), nr_events: 4 task 15 ( make: 6617), nr_events: 7 task 16 ( migration/3: 9), nr_events: 1 task 17 ( migration/5: 13), nr_events: 1 task 18 ( migration/7: 17), nr_events: 1 task 19 ( migration/1: 5), nr_events: 1 task 20 ( sh: 6618), nr_events: 4 task 21 ( make: 6619), nr_events: 5 task 22 ( sh: 6620), nr_events: 4 task 23 ( make: 6621), nr_events: 10 task 24 ( sh: 6623), nr_events: 3 task 25 ( gcc: 6624), nr_events: 4 task 26 ( gcc: 6625), nr_events: 4 task 27 ( gcc: 6626), nr_events: 5 task 28 ( collect2: 6627), nr_events: 5 task 29 ( sh: 6622), nr_events: 1 task 30 ( make: 6628), nr_events: 7 task 31 ( sh: 6630), nr_events: 4 task 32 ( gcc: 6631), nr_events: 4 task 33 ( sh: 6629), nr_events: 1 task 34 ( gcc: 6632), nr_events: 4 task 35 ( gcc: 6633), nr_events: 4 task 36 ( collect2: 6634), nr_events: 4 task 37 ( make: 6635), nr_events: 8 task 38 ( sh: 6637), nr_events: 4 task 39 ( sh: 6636), nr_events: 1 task 40 ( gcc: 6638), nr_events: 4 task 41 ( gcc: 6639), nr_events: 4 task 42 ( gcc: 6640), nr_events: 4 task 43 ( collect2: 6641), nr_events: 4 task 44 ( make: 6642), nr_events: 6 task 45 ( sh: 6643), nr_events: 5 task 46 ( sh: 6644), nr_events: 3 task 47 ( sh: 6645), nr_events: 4 task 48 ( make: 6646), nr_events: 6 task 49 ( sh: 6647), nr_events: 3 task 50 ( make: 6648), nr_events: 5 task 51 ( sh: 6649), nr_events: 5 task 52 ( sh: 6650), nr_events: 6 task 53 ( make: 6651), nr_events: 4 task 54 ( make: 6652), nr_events: 5 task 55 ( make: 6653), nr_events: 4 task 56 ( make: 6654), nr_events: 4 task 57 ( make: 6655), nr_events: 5 task 58 ( sh: 6656), nr_events: 4 task 59 ( gcc: 6657), nr_events: 9 task 60 ( ksoftirqd/3: 10), nr_events: 1 task 61 ( gcc: 6658), nr_events: 4 task 62 ( make: 6659), nr_events: 5 task 63 ( sh: 6660), nr_events: 3 task 64 ( gcc: 6661), nr_events: 5 task 65 ( collect2: 6662), nr_events: 4 ------------------------------------------------------------ #1 : 256.745, ravg: 256.74, cpu: 0.00 / 0.00 #2 : 439.372, ravg: 275.01, cpu: 0.00 / 0.00 #3 : 411.971, ravg: 288.70, cpu: 0.00 / 0.00 #4 : 385.500, ravg: 298.38, cpu: 0.00 / 0.00 #5 : 366.526, ravg: 305.20, cpu: 0.00 / 0.00 #6 : 381.281, ravg: 312.81, cpu: 0.00 / 0.00 #7 : 410.756, ravg: 322.60, cpu: 0.00 / 0.00 #8 : 368.009, ravg: 327.14, cpu: 0.00 / 0.00 #9 : 408.098, ravg: 335.24, cpu: 0.00 / 0.00 #10 : 368.582, ravg: 338.57, cpu: 0.00 / 0.00 I.e. we successfully analyzed the trace, replayed it via real threads and measured the replayed workload's scheduling properties. This is how it looked like in 'top' output: PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND 7164 mingo 20 0 1434m 8080 888 R 57.0 0.1 0:02.04 :perf 7165 mingo 20 0 1434m 8080 888 R 41.8 0.1 0:01.52 :perf 7228 mingo 20 0 1434m 8080 888 R 39.8 0.1 0:01.44 :gcc 7225 mingo 20 0 1434m 8080 888 R 33.8 0.1 0:01.26 :gcc 7202 mingo 20 0 1434m 8080 888 R 31.2 0.1 0:01.16 :sh 7222 mingo 20 0 1434m 8080 888 R 25.2 0.1 0:00.96 :sh 7211 mingo 20 0 1434m 8080 888 R 21.9 0.1 0:00.82 :sh 7213 mingo 20 0 1434m 8080 888 D 19.2 0.1 0:00.74 :sh 7194 mingo 20 0 1434m 8080 888 D 18.6 0.1 0:00.72 :make There's still various kinks in it - more patches to come. Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Mike Galbraith <efault@gmx.de> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> LKML-Reference: <new-submission> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-09-11 06:12:54 -04:00
* optimize an existing RUN event by merging this one
* to it:
*/
if (curr_event && curr_event->type == SCHED_EVENT_RUN) {
nr_run_events_optimized++;
curr_event->duration += duration;
return;
}
event = get_new_event(task, timestamp);
event->type = SCHED_EVENT_RUN;
event->duration = duration;
nr_run_events++;
}
static unsigned long targetless_wakeups;
static unsigned long multitarget_wakeups;
static void
add_sched_event_wakeup(struct task_desc *task, nsec_t timestamp,
struct task_desc *wakee)
{
struct sched_event *event, *wakee_event;
event = get_new_event(task, timestamp);
event->type = SCHED_EVENT_WAKEUP;
event->wakee = wakee;
wakee_event = last_event(wakee);
if (!wakee_event || wakee_event->type != SCHED_EVENT_SLEEP) {
targetless_wakeups++;
return;
}
if (wakee_event->wait_sem) {
multitarget_wakeups++;
return;
}
wakee_event->wait_sem = calloc(1, sizeof(*wakee_event->wait_sem));
sem_init(wakee_event->wait_sem, 0, 0);
wakee_event->specific_wait = 1;
event->wait_sem = wakee_event->wait_sem;
nr_wakeup_events++;
}
static void
add_sched_event_sleep(struct task_desc *task, nsec_t timestamp,
u64 task_state __used)
{
struct sched_event *event = get_new_event(task, timestamp);
event->type = SCHED_EVENT_SLEEP;
nr_sleep_events++;
}
static struct task_desc *register_pid(unsigned long pid, const char *comm)
{
struct task_desc *task;
BUG_ON(pid >= MAX_PID);
task = pid_to_task[pid];
if (task)
return task;
task = calloc(1, sizeof(*task));
task->pid = pid;
task->nr = nr_tasks;
strcpy(task->comm, comm);
/*
* every task starts in sleeping state - this gets ignored
* if there's no wakeup pointing to this sleep state:
*/
add_sched_event_sleep(task, 0, 0);
pid_to_task[pid] = task;
nr_tasks++;
tasks = realloc(tasks, nr_tasks*sizeof(struct task_task *));
BUG_ON(!tasks);
tasks[task->nr] = task;
if (verbose)
printf("registered task #%ld, PID %ld (%s)\n", nr_tasks, pid, comm);
return task;
}
static void print_task_traces(void)
{
struct task_desc *task;
unsigned long i;
for (i = 0; i < nr_tasks; i++) {
task = tasks[i];
printf("task %6ld (%20s:%10ld), nr_events: %ld\n",
task->nr, task->comm, task->pid, task->nr_events);
}
}
static void add_cross_task_wakeups(void)
{
struct task_desc *task1, *task2;
unsigned long i, j;
for (i = 0; i < nr_tasks; i++) {
task1 = tasks[i];
j = i + 1;
if (j == nr_tasks)
j = 0;
task2 = tasks[j];
add_sched_event_wakeup(task1, 0, task2);
}
}
static void
perf sched: Implement the scheduling workload replay engine Integrate the schedbench.c bits with the raw trace events that we get from the perf machinery, and activate the workload replayer/simulator. Example of a captured 'make -j' workload: $ perf sched run measurement overhead: 90 nsecs sleep measurement overhead: 2724743 nsecs the run test took 1000081 nsecs the sleep test took 2981111 nsecs version = 0.5 ... nr_run_events: 70 nr_sleep_events: 66 nr_wakeup_events: 9 target-less wakeups: 71 multi-target wakeups: 47 run events optimized: 139 task 0 ( perf: 6607), nr_events: 2 task 1 ( perf: 6608), nr_events: 6 task 2 ( : 0), nr_events: 1 task 3 ( make: 6609), nr_events: 5 task 4 ( sh: 6610), nr_events: 4 task 5 ( make: 6611), nr_events: 6 task 6 ( sh: 6612), nr_events: 4 task 7 ( make: 6613), nr_events: 5 task 8 ( migration/11: 25), nr_events: 1 task 9 ( migration/13: 29), nr_events: 1 task 10 ( migration/15: 33), nr_events: 1 task 11 ( migration/9: 21), nr_events: 1 task 12 ( sh: 6614), nr_events: 4 task 13 ( make: 6615), nr_events: 5 task 14 ( sh: 6616), nr_events: 4 task 15 ( make: 6617), nr_events: 7 task 16 ( migration/3: 9), nr_events: 1 task 17 ( migration/5: 13), nr_events: 1 task 18 ( migration/7: 17), nr_events: 1 task 19 ( migration/1: 5), nr_events: 1 task 20 ( sh: 6618), nr_events: 4 task 21 ( make: 6619), nr_events: 5 task 22 ( sh: 6620), nr_events: 4 task 23 ( make: 6621), nr_events: 10 task 24 ( sh: 6623), nr_events: 3 task 25 ( gcc: 6624), nr_events: 4 task 26 ( gcc: 6625), nr_events: 4 task 27 ( gcc: 6626), nr_events: 5 task 28 ( collect2: 6627), nr_events: 5 task 29 ( sh: 6622), nr_events: 1 task 30 ( make: 6628), nr_events: 7 task 31 ( sh: 6630), nr_events: 4 task 32 ( gcc: 6631), nr_events: 4 task 33 ( sh: 6629), nr_events: 1 task 34 ( gcc: 6632), nr_events: 4 task 35 ( gcc: 6633), nr_events: 4 task 36 ( collect2: 6634), nr_events: 4 task 37 ( make: 6635), nr_events: 8 task 38 ( sh: 6637), nr_events: 4 task 39 ( sh: 6636), nr_events: 1 task 40 ( gcc: 6638), nr_events: 4 task 41 ( gcc: 6639), nr_events: 4 task 42 ( gcc: 6640), nr_events: 4 task 43 ( collect2: 6641), nr_events: 4 task 44 ( make: 6642), nr_events: 6 task 45 ( sh: 6643), nr_events: 5 task 46 ( sh: 6644), nr_events: 3 task 47 ( sh: 6645), nr_events: 4 task 48 ( make: 6646), nr_events: 6 task 49 ( sh: 6647), nr_events: 3 task 50 ( make: 6648), nr_events: 5 task 51 ( sh: 6649), nr_events: 5 task 52 ( sh: 6650), nr_events: 6 task 53 ( make: 6651), nr_events: 4 task 54 ( make: 6652), nr_events: 5 task 55 ( make: 6653), nr_events: 4 task 56 ( make: 6654), nr_events: 4 task 57 ( make: 6655), nr_events: 5 task 58 ( sh: 6656), nr_events: 4 task 59 ( gcc: 6657), nr_events: 9 task 60 ( ksoftirqd/3: 10), nr_events: 1 task 61 ( gcc: 6658), nr_events: 4 task 62 ( make: 6659), nr_events: 5 task 63 ( sh: 6660), nr_events: 3 task 64 ( gcc: 6661), nr_events: 5 task 65 ( collect2: 6662), nr_events: 4 ------------------------------------------------------------ #1 : 256.745, ravg: 256.74, cpu: 0.00 / 0.00 #2 : 439.372, ravg: 275.01, cpu: 0.00 / 0.00 #3 : 411.971, ravg: 288.70, cpu: 0.00 / 0.00 #4 : 385.500, ravg: 298.38, cpu: 0.00 / 0.00 #5 : 366.526, ravg: 305.20, cpu: 0.00 / 0.00 #6 : 381.281, ravg: 312.81, cpu: 0.00 / 0.00 #7 : 410.756, ravg: 322.60, cpu: 0.00 / 0.00 #8 : 368.009, ravg: 327.14, cpu: 0.00 / 0.00 #9 : 408.098, ravg: 335.24, cpu: 0.00 / 0.00 #10 : 368.582, ravg: 338.57, cpu: 0.00 / 0.00 I.e. we successfully analyzed the trace, replayed it via real threads and measured the replayed workload's scheduling properties. This is how it looked like in 'top' output: PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND 7164 mingo 20 0 1434m 8080 888 R 57.0 0.1 0:02.04 :perf 7165 mingo 20 0 1434m 8080 888 R 41.8 0.1 0:01.52 :perf 7228 mingo 20 0 1434m 8080 888 R 39.8 0.1 0:01.44 :gcc 7225 mingo 20 0 1434m 8080 888 R 33.8 0.1 0:01.26 :gcc 7202 mingo 20 0 1434m 8080 888 R 31.2 0.1 0:01.16 :sh 7222 mingo 20 0 1434m 8080 888 R 25.2 0.1 0:00.96 :sh 7211 mingo 20 0 1434m 8080 888 R 21.9 0.1 0:00.82 :sh 7213 mingo 20 0 1434m 8080 888 D 19.2 0.1 0:00.74 :sh 7194 mingo 20 0 1434m 8080 888 D 18.6 0.1 0:00.72 :make There's still various kinks in it - more patches to come. Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Mike Galbraith <efault@gmx.de> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> LKML-Reference: <new-submission> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-09-11 06:12:54 -04:00
process_sched_event(struct task_desc *this_task __used, struct sched_event *event)
{
int ret = 0;
nsec_t now;
long long delta;
now = get_nsecs();
delta = start_time + event->timestamp - now;
switch (event->type) {
case SCHED_EVENT_RUN:
burn_nsecs(event->duration);
break;
case SCHED_EVENT_SLEEP:
if (event->wait_sem)
ret = sem_wait(event->wait_sem);
BUG_ON(ret);
break;
case SCHED_EVENT_WAKEUP:
if (event->wait_sem)
ret = sem_post(event->wait_sem);
BUG_ON(ret);
break;
default:
BUG_ON(1);
}
}
static nsec_t get_cpu_usage_nsec_parent(void)
{
struct rusage ru;
nsec_t sum;
int err;
err = getrusage(RUSAGE_SELF, &ru);
BUG_ON(err);
sum = ru.ru_utime.tv_sec*1e9 + ru.ru_utime.tv_usec*1e3;
sum += ru.ru_stime.tv_sec*1e9 + ru.ru_stime.tv_usec*1e3;
return sum;
}
static nsec_t get_cpu_usage_nsec_self(void)
{
char filename [] = "/proc/1234567890/sched";
unsigned long msecs, nsecs;
char *line = NULL;
nsec_t total = 0;
size_t len = 0;
ssize_t chars;
FILE *file;
int ret;
sprintf(filename, "/proc/%d/sched", getpid());
file = fopen(filename, "r");
BUG_ON(!file);
while ((chars = getline(&line, &len, file)) != -1) {
ret = sscanf(line, "se.sum_exec_runtime : %ld.%06ld\n",
&msecs, &nsecs);
if (ret == 2) {
total = msecs*1e6 + nsecs;
break;
}
}
if (line)
free(line);
fclose(file);
return total;
}
static void *thread_func(void *ctx)
{
struct task_desc *this_task = ctx;
nsec_t cpu_usage_0, cpu_usage_1;
unsigned long i, ret;
char comm2[22];
sprintf(comm2, ":%s", this_task->comm);
prctl(PR_SET_NAME, comm2);
again:
ret = sem_post(&this_task->ready_for_work);
BUG_ON(ret);
ret = pthread_mutex_lock(&start_work_mutex);
BUG_ON(ret);
ret = pthread_mutex_unlock(&start_work_mutex);
BUG_ON(ret);
cpu_usage_0 = get_cpu_usage_nsec_self();
for (i = 0; i < this_task->nr_events; i++) {
this_task->curr_event = i;
process_sched_event(this_task, this_task->events[i]);
}
cpu_usage_1 = get_cpu_usage_nsec_self();
this_task->cpu_usage = cpu_usage_1 - cpu_usage_0;
ret = sem_post(&this_task->work_done_sem);
BUG_ON(ret);
ret = pthread_mutex_lock(&work_done_wait_mutex);
BUG_ON(ret);
ret = pthread_mutex_unlock(&work_done_wait_mutex);
BUG_ON(ret);
goto again;
}
static void create_tasks(void)
{
struct task_desc *task;
pthread_attr_t attr;
unsigned long i;
int err;
err = pthread_attr_init(&attr);
BUG_ON(err);
err = pthread_attr_setstacksize(&attr, (size_t)(16*1024));
BUG_ON(err);
err = pthread_mutex_lock(&start_work_mutex);
BUG_ON(err);
err = pthread_mutex_lock(&work_done_wait_mutex);
BUG_ON(err);
for (i = 0; i < nr_tasks; i++) {
task = tasks[i];
sem_init(&task->sleep_sem, 0, 0);
sem_init(&task->ready_for_work, 0, 0);
sem_init(&task->work_done_sem, 0, 0);
task->curr_event = 0;
err = pthread_create(&task->thread, &attr, thread_func, task);
BUG_ON(err);
}
}
static nsec_t cpu_usage;
static nsec_t runavg_cpu_usage;
static nsec_t parent_cpu_usage;
static nsec_t runavg_parent_cpu_usage;
static void wait_for_tasks(void)
{
nsec_t cpu_usage_0, cpu_usage_1;
struct task_desc *task;
unsigned long i, ret;
start_time = get_nsecs();
cpu_usage = 0;
pthread_mutex_unlock(&work_done_wait_mutex);
for (i = 0; i < nr_tasks; i++) {
task = tasks[i];
ret = sem_wait(&task->ready_for_work);
BUG_ON(ret);
sem_init(&task->ready_for_work, 0, 0);
}
ret = pthread_mutex_lock(&work_done_wait_mutex);
BUG_ON(ret);
cpu_usage_0 = get_cpu_usage_nsec_parent();
pthread_mutex_unlock(&start_work_mutex);
for (i = 0; i < nr_tasks; i++) {
task = tasks[i];
ret = sem_wait(&task->work_done_sem);
BUG_ON(ret);
sem_init(&task->work_done_sem, 0, 0);
cpu_usage += task->cpu_usage;
task->cpu_usage = 0;
}
cpu_usage_1 = get_cpu_usage_nsec_parent();
if (!runavg_cpu_usage)
runavg_cpu_usage = cpu_usage;
runavg_cpu_usage = (runavg_cpu_usage*9 + cpu_usage)/10;
parent_cpu_usage = cpu_usage_1 - cpu_usage_0;
if (!runavg_parent_cpu_usage)
runavg_parent_cpu_usage = parent_cpu_usage;
runavg_parent_cpu_usage = (runavg_parent_cpu_usage*9 +
parent_cpu_usage)/10;
ret = pthread_mutex_lock(&start_work_mutex);
BUG_ON(ret);
for (i = 0; i < nr_tasks; i++) {
task = tasks[i];
sem_init(&task->sleep_sem, 0, 0);
task->curr_event = 0;
}
}
static int read_events(void);
static unsigned long nr_runs;
static nsec_t sum_runtime;
static nsec_t sum_fluct;
static nsec_t run_avg;
static void run_one_test(void)
{
nsec_t T0, T1, delta, avg_delta, fluct, std_dev;
T0 = get_nsecs();
wait_for_tasks();
T1 = get_nsecs();
delta = T1 - T0;
sum_runtime += delta;
nr_runs++;
avg_delta = sum_runtime / nr_runs;
if (delta < avg_delta)
fluct = avg_delta - delta;
else
fluct = delta - avg_delta;
sum_fluct += fluct;
std_dev = sum_fluct / nr_runs / sqrt(nr_runs);
if (!run_avg)
run_avg = delta;
run_avg = (run_avg*9 + delta)/10;
printf("#%-3ld: %0.3f, ",
nr_runs, (double)delta/1000000.0);
#if 0
printf("%0.2f +- %0.2f, ",
(double)avg_delta/1e6, (double)std_dev/1e6);
#endif
printf("ravg: %0.2f, ",
(double)run_avg/1e6);
printf("cpu: %0.2f / %0.2f",
(double)cpu_usage/1e6, (double)runavg_cpu_usage/1e6);
#if 0
/*
perf sched: Implement the scheduling workload replay engine Integrate the schedbench.c bits with the raw trace events that we get from the perf machinery, and activate the workload replayer/simulator. Example of a captured 'make -j' workload: $ perf sched run measurement overhead: 90 nsecs sleep measurement overhead: 2724743 nsecs the run test took 1000081 nsecs the sleep test took 2981111 nsecs version = 0.5 ... nr_run_events: 70 nr_sleep_events: 66 nr_wakeup_events: 9 target-less wakeups: 71 multi-target wakeups: 47 run events optimized: 139 task 0 ( perf: 6607), nr_events: 2 task 1 ( perf: 6608), nr_events: 6 task 2 ( : 0), nr_events: 1 task 3 ( make: 6609), nr_events: 5 task 4 ( sh: 6610), nr_events: 4 task 5 ( make: 6611), nr_events: 6 task 6 ( sh: 6612), nr_events: 4 task 7 ( make: 6613), nr_events: 5 task 8 ( migration/11: 25), nr_events: 1 task 9 ( migration/13: 29), nr_events: 1 task 10 ( migration/15: 33), nr_events: 1 task 11 ( migration/9: 21), nr_events: 1 task 12 ( sh: 6614), nr_events: 4 task 13 ( make: 6615), nr_events: 5 task 14 ( sh: 6616), nr_events: 4 task 15 ( make: 6617), nr_events: 7 task 16 ( migration/3: 9), nr_events: 1 task 17 ( migration/5: 13), nr_events: 1 task 18 ( migration/7: 17), nr_events: 1 task 19 ( migration/1: 5), nr_events: 1 task 20 ( sh: 6618), nr_events: 4 task 21 ( make: 6619), nr_events: 5 task 22 ( sh: 6620), nr_events: 4 task 23 ( make: 6621), nr_events: 10 task 24 ( sh: 6623), nr_events: 3 task 25 ( gcc: 6624), nr_events: 4 task 26 ( gcc: 6625), nr_events: 4 task 27 ( gcc: 6626), nr_events: 5 task 28 ( collect2: 6627), nr_events: 5 task 29 ( sh: 6622), nr_events: 1 task 30 ( make: 6628), nr_events: 7 task 31 ( sh: 6630), nr_events: 4 task 32 ( gcc: 6631), nr_events: 4 task 33 ( sh: 6629), nr_events: 1 task 34 ( gcc: 6632), nr_events: 4 task 35 ( gcc: 6633), nr_events: 4 task 36 ( collect2: 6634), nr_events: 4 task 37 ( make: 6635), nr_events: 8 task 38 ( sh: 6637), nr_events: 4 task 39 ( sh: 6636), nr_events: 1 task 40 ( gcc: 6638), nr_events: 4 task 41 ( gcc: 6639), nr_events: 4 task 42 ( gcc: 6640), nr_events: 4 task 43 ( collect2: 6641), nr_events: 4 task 44 ( make: 6642), nr_events: 6 task 45 ( sh: 6643), nr_events: 5 task 46 ( sh: 6644), nr_events: 3 task 47 ( sh: 6645), nr_events: 4 task 48 ( make: 6646), nr_events: 6 task 49 ( sh: 6647), nr_events: 3 task 50 ( make: 6648), nr_events: 5 task 51 ( sh: 6649), nr_events: 5 task 52 ( sh: 6650), nr_events: 6 task 53 ( make: 6651), nr_events: 4 task 54 ( make: 6652), nr_events: 5 task 55 ( make: 6653), nr_events: 4 task 56 ( make: 6654), nr_events: 4 task 57 ( make: 6655), nr_events: 5 task 58 ( sh: 6656), nr_events: 4 task 59 ( gcc: 6657), nr_events: 9 task 60 ( ksoftirqd/3: 10), nr_events: 1 task 61 ( gcc: 6658), nr_events: 4 task 62 ( make: 6659), nr_events: 5 task 63 ( sh: 6660), nr_events: 3 task 64 ( gcc: 6661), nr_events: 5 task 65 ( collect2: 6662), nr_events: 4 ------------------------------------------------------------ #1 : 256.745, ravg: 256.74, cpu: 0.00 / 0.00 #2 : 439.372, ravg: 275.01, cpu: 0.00 / 0.00 #3 : 411.971, ravg: 288.70, cpu: 0.00 / 0.00 #4 : 385.500, ravg: 298.38, cpu: 0.00 / 0.00 #5 : 366.526, ravg: 305.20, cpu: 0.00 / 0.00 #6 : 381.281, ravg: 312.81, cpu: 0.00 / 0.00 #7 : 410.756, ravg: 322.60, cpu: 0.00 / 0.00 #8 : 368.009, ravg: 327.14, cpu: 0.00 / 0.00 #9 : 408.098, ravg: 335.24, cpu: 0.00 / 0.00 #10 : 368.582, ravg: 338.57, cpu: 0.00 / 0.00 I.e. we successfully analyzed the trace, replayed it via real threads and measured the replayed workload's scheduling properties. This is how it looked like in 'top' output: PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND 7164 mingo 20 0 1434m 8080 888 R 57.0 0.1 0:02.04 :perf 7165 mingo 20 0 1434m 8080 888 R 41.8 0.1 0:01.52 :perf 7228 mingo 20 0 1434m 8080 888 R 39.8 0.1 0:01.44 :gcc 7225 mingo 20 0 1434m 8080 888 R 33.8 0.1 0:01.26 :gcc 7202 mingo 20 0 1434m 8080 888 R 31.2 0.1 0:01.16 :sh 7222 mingo 20 0 1434m 8080 888 R 25.2 0.1 0:00.96 :sh 7211 mingo 20 0 1434m 8080 888 R 21.9 0.1 0:00.82 :sh 7213 mingo 20 0 1434m 8080 888 D 19.2 0.1 0:00.74 :sh 7194 mingo 20 0 1434m 8080 888 D 18.6 0.1 0:00.72 :make There's still various kinks in it - more patches to come. Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Mike Galbraith <efault@gmx.de> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> LKML-Reference: <new-submission> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-09-11 06:12:54 -04:00
* rusage statistics done by the parent, these are less
* accurate than the sum_exec_runtime based statistics:
*/
printf(" [%0.2f / %0.2f]",
(double)parent_cpu_usage/1e6,
(double)runavg_parent_cpu_usage/1e6);
#endif
printf("\n");
if (nr_sleep_corrections)
printf(" (%ld sleep corrections)\n", nr_sleep_corrections);
nr_sleep_corrections = 0;
}
static void test_calibrations(void)
{
nsec_t T0, T1;
T0 = get_nsecs();
burn_nsecs(1e6);
T1 = get_nsecs();
printf("the run test took %Ld nsecs\n", T1-T0);
T0 = get_nsecs();
sleep_nsecs(1e6);
T1 = get_nsecs();
printf("the sleep test took %Ld nsecs\n", T1-T0);
}
static void __cmd_replay(void)
{
long nr_iterations = 10, i;
calibrate_run_measurement_overhead();
calibrate_sleep_measurement_overhead();
test_calibrations();
read_events();
printf("nr_run_events: %ld\n", nr_run_events);
printf("nr_sleep_events: %ld\n", nr_sleep_events);
printf("nr_wakeup_events: %ld\n", nr_wakeup_events);
if (targetless_wakeups)
printf("target-less wakeups: %ld\n", targetless_wakeups);
if (multitarget_wakeups)
printf("multi-target wakeups: %ld\n", multitarget_wakeups);
if (nr_run_events_optimized)
printf("run events optimized: %ld\n",
nr_run_events_optimized);
print_task_traces();
add_cross_task_wakeups();
create_tasks();
printf("------------------------------------------------------------\n");
for (i = 0; i < nr_iterations; i++)
run_one_test();
}
static int
process_comm_event(event_t *event, unsigned long offset, unsigned long head)
{
struct thread *thread;
thread = threads__findnew(event->comm.pid, &threads, &last_match);
dump_printf("%p [%p]: PERF_EVENT_COMM: %s:%d\n",
(void *)(offset + head),
(void *)(long)(event->header.size),
event->comm.comm, event->comm.pid);
if (thread == NULL ||
thread__set_comm(thread, event->comm.comm)) {
dump_printf("problem processing PERF_EVENT_COMM, skipping event.\n");
return -1;
}
total_comm++;
return 0;
}
perf sched: Fix bad event alignment perf sched raises the following error when it meets a sched switch event: perf: builtin-sched.c:286: register_pid: Assertion `!(pid >= 65536)' failed. Abandon Currently in x86-64, the sched switch events have a hole in the middle of the structure: u16 common_type; u8 common_flags; u8 common_preempt_count; u32 common_pid; u32 common_tgid; char prev_comm[16]; u32 prev_pid; u32 prev_prio; <--- there u64 prev_state; char next_comm[16]; u32 next_pid; u32 next_prio; Gcc inserts a 4 bytes hole there for prev_state to be u64 aligned. And the events are exported to userspace with this hole. But in userspace, from perf sched, we fetch it using a structure that has a new field in the beginning: u32 size. This is because our trace is exported with its size as a field. But now that we have this new field, the hole in the middle disappears because it makes prev_state becoming well aligned. And since we are using a pointer to the raw trace using this struct, instead of reading prev_state, we are reading the hole. We could fix it by keeping the size seperate from the struct but actually there a lot of other potential problems: some fields may be saved as long in a 64 bits system and later read as long in a 32 bits system. Also this direct cast doesn't care about the endianness differences between the host traced machine and the machine in which we do the post processing. So instead of using such dangerous direct casts, fetch the values using the trace parsing API that already takes care of all these problems. Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Mike Galbraith <efault@gmx.de> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-09-11 20:43:45 -04:00
struct raw_event_sample {
u32 size;
char data[0];
};
#define FILL_FIELD(ptr, field, event, data) \
ptr.field = (typeof(ptr.field)) raw_field_value(event, #field, data)
#define FILL_ARRAY(ptr, array, event, data) \
do { \
void *__array = raw_field_ptr(event, #array, data); \
memcpy(ptr.array, __array, sizeof(ptr.array)); \
} while(0)
#define FILL_COMMON_FIELDS(ptr, event, data) \
do { \
FILL_FIELD(ptr, common_type, event, data); \
FILL_FIELD(ptr, common_flags, event, data); \
FILL_FIELD(ptr, common_preempt_count, event, data); \
FILL_FIELD(ptr, common_pid, event, data); \
FILL_FIELD(ptr, common_tgid, event, data); \
} while (0)
struct trace_switch_event {
u32 size;
u16 common_type;
u8 common_flags;
u8 common_preempt_count;
u32 common_pid;
u32 common_tgid;
char prev_comm[16];
u32 prev_pid;
u32 prev_prio;
u64 prev_state;
char next_comm[16];
u32 next_pid;
u32 next_prio;
};
perf sched: Implement the scheduling workload replay engine Integrate the schedbench.c bits with the raw trace events that we get from the perf machinery, and activate the workload replayer/simulator. Example of a captured 'make -j' workload: $ perf sched run measurement overhead: 90 nsecs sleep measurement overhead: 2724743 nsecs the run test took 1000081 nsecs the sleep test took 2981111 nsecs version = 0.5 ... nr_run_events: 70 nr_sleep_events: 66 nr_wakeup_events: 9 target-less wakeups: 71 multi-target wakeups: 47 run events optimized: 139 task 0 ( perf: 6607), nr_events: 2 task 1 ( perf: 6608), nr_events: 6 task 2 ( : 0), nr_events: 1 task 3 ( make: 6609), nr_events: 5 task 4 ( sh: 6610), nr_events: 4 task 5 ( make: 6611), nr_events: 6 task 6 ( sh: 6612), nr_events: 4 task 7 ( make: 6613), nr_events: 5 task 8 ( migration/11: 25), nr_events: 1 task 9 ( migration/13: 29), nr_events: 1 task 10 ( migration/15: 33), nr_events: 1 task 11 ( migration/9: 21), nr_events: 1 task 12 ( sh: 6614), nr_events: 4 task 13 ( make: 6615), nr_events: 5 task 14 ( sh: 6616), nr_events: 4 task 15 ( make: 6617), nr_events: 7 task 16 ( migration/3: 9), nr_events: 1 task 17 ( migration/5: 13), nr_events: 1 task 18 ( migration/7: 17), nr_events: 1 task 19 ( migration/1: 5), nr_events: 1 task 20 ( sh: 6618), nr_events: 4 task 21 ( make: 6619), nr_events: 5 task 22 ( sh: 6620), nr_events: 4 task 23 ( make: 6621), nr_events: 10 task 24 ( sh: 6623), nr_events: 3 task 25 ( gcc: 6624), nr_events: 4 task 26 ( gcc: 6625), nr_events: 4 task 27 ( gcc: 6626), nr_events: 5 task 28 ( collect2: 6627), nr_events: 5 task 29 ( sh: 6622), nr_events: 1 task 30 ( make: 6628), nr_events: 7 task 31 ( sh: 6630), nr_events: 4 task 32 ( gcc: 6631), nr_events: 4 task 33 ( sh: 6629), nr_events: 1 task 34 ( gcc: 6632), nr_events: 4 task 35 ( gcc: 6633), nr_events: 4 task 36 ( collect2: 6634), nr_events: 4 task 37 ( make: 6635), nr_events: 8 task 38 ( sh: 6637), nr_events: 4 task 39 ( sh: 6636), nr_events: 1 task 40 ( gcc: 6638), nr_events: 4 task 41 ( gcc: 6639), nr_events: 4 task 42 ( gcc: 6640), nr_events: 4 task 43 ( collect2: 6641), nr_events: 4 task 44 ( make: 6642), nr_events: 6 task 45 ( sh: 6643), nr_events: 5 task 46 ( sh: 6644), nr_events: 3 task 47 ( sh: 6645), nr_events: 4 task 48 ( make: 6646), nr_events: 6 task 49 ( sh: 6647), nr_events: 3 task 50 ( make: 6648), nr_events: 5 task 51 ( sh: 6649), nr_events: 5 task 52 ( sh: 6650), nr_events: 6 task 53 ( make: 6651), nr_events: 4 task 54 ( make: 6652), nr_events: 5 task 55 ( make: 6653), nr_events: 4 task 56 ( make: 6654), nr_events: 4 task 57 ( make: 6655), nr_events: 5 task 58 ( sh: 6656), nr_events: 4 task 59 ( gcc: 6657), nr_events: 9 task 60 ( ksoftirqd/3: 10), nr_events: 1 task 61 ( gcc: 6658), nr_events: 4 task 62 ( make: 6659), nr_events: 5 task 63 ( sh: 6660), nr_events: 3 task 64 ( gcc: 6661), nr_events: 5 task 65 ( collect2: 6662), nr_events: 4 ------------------------------------------------------------ #1 : 256.745, ravg: 256.74, cpu: 0.00 / 0.00 #2 : 439.372, ravg: 275.01, cpu: 0.00 / 0.00 #3 : 411.971, ravg: 288.70, cpu: 0.00 / 0.00 #4 : 385.500, ravg: 298.38, cpu: 0.00 / 0.00 #5 : 366.526, ravg: 305.20, cpu: 0.00 / 0.00 #6 : 381.281, ravg: 312.81, cpu: 0.00 / 0.00 #7 : 410.756, ravg: 322.60, cpu: 0.00 / 0.00 #8 : 368.009, ravg: 327.14, cpu: 0.00 / 0.00 #9 : 408.098, ravg: 335.24, cpu: 0.00 / 0.00 #10 : 368.582, ravg: 338.57, cpu: 0.00 / 0.00 I.e. we successfully analyzed the trace, replayed it via real threads and measured the replayed workload's scheduling properties. This is how it looked like in 'top' output: PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND 7164 mingo 20 0 1434m 8080 888 R 57.0 0.1 0:02.04 :perf 7165 mingo 20 0 1434m 8080 888 R 41.8 0.1 0:01.52 :perf 7228 mingo 20 0 1434m 8080 888 R 39.8 0.1 0:01.44 :gcc 7225 mingo 20 0 1434m 8080 888 R 33.8 0.1 0:01.26 :gcc 7202 mingo 20 0 1434m 8080 888 R 31.2 0.1 0:01.16 :sh 7222 mingo 20 0 1434m 8080 888 R 25.2 0.1 0:00.96 :sh 7211 mingo 20 0 1434m 8080 888 R 21.9 0.1 0:00.82 :sh 7213 mingo 20 0 1434m 8080 888 D 19.2 0.1 0:00.74 :sh 7194 mingo 20 0 1434m 8080 888 D 18.6 0.1 0:00.72 :make There's still various kinks in it - more patches to come. Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Mike Galbraith <efault@gmx.de> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> LKML-Reference: <new-submission> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-09-11 06:12:54 -04:00
struct trace_wakeup_event {
u32 size;
u16 common_type;
u8 common_flags;
u8 common_preempt_count;
u32 common_pid;
u32 common_tgid;
char comm[16];
u32 pid;
u32 prio;
u32 success;
u32 cpu;
};
struct trace_fork_event {
u32 size;
perf sched: Fix bad event alignment perf sched raises the following error when it meets a sched switch event: perf: builtin-sched.c:286: register_pid: Assertion `!(pid >= 65536)' failed. Abandon Currently in x86-64, the sched switch events have a hole in the middle of the structure: u16 common_type; u8 common_flags; u8 common_preempt_count; u32 common_pid; u32 common_tgid; char prev_comm[16]; u32 prev_pid; u32 prev_prio; <--- there u64 prev_state; char next_comm[16]; u32 next_pid; u32 next_prio; Gcc inserts a 4 bytes hole there for prev_state to be u64 aligned. And the events are exported to userspace with this hole. But in userspace, from perf sched, we fetch it using a structure that has a new field in the beginning: u32 size. This is because our trace is exported with its size as a field. But now that we have this new field, the hole in the middle disappears because it makes prev_state becoming well aligned. And since we are using a pointer to the raw trace using this struct, instead of reading prev_state, we are reading the hole. We could fix it by keeping the size seperate from the struct but actually there a lot of other potential problems: some fields may be saved as long in a 64 bits system and later read as long in a 32 bits system. Also this direct cast doesn't care about the endianness differences between the host traced machine and the machine in which we do the post processing. So instead of using such dangerous direct casts, fetch the values using the trace parsing API that already takes care of all these problems. Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Mike Galbraith <efault@gmx.de> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-09-11 20:43:45 -04:00
u16 common_type;
u8 common_flags;
u8 common_preempt_count;
u32 common_pid;
u32 common_tgid;
char parent_comm[16];
u32 parent_pid;
char child_comm[16];
u32 child_pid;
};
struct trace_sched_handler {
void (*switch_event)(struct trace_switch_event *,
struct event *,
int cpu,
u64 timestamp,
struct thread *thread);
void (*wakeup_event)(struct trace_wakeup_event *,
struct event *,
int cpu,
u64 timestamp,
struct thread *thread);
void (*fork_event)(struct trace_fork_event *,
struct event *,
int cpu,
u64 timestamp,
struct thread *thread);
};
perf sched: Fix bad event alignment perf sched raises the following error when it meets a sched switch event: perf: builtin-sched.c:286: register_pid: Assertion `!(pid >= 65536)' failed. Abandon Currently in x86-64, the sched switch events have a hole in the middle of the structure: u16 common_type; u8 common_flags; u8 common_preempt_count; u32 common_pid; u32 common_tgid; char prev_comm[16]; u32 prev_pid; u32 prev_prio; <--- there u64 prev_state; char next_comm[16]; u32 next_pid; u32 next_prio; Gcc inserts a 4 bytes hole there for prev_state to be u64 aligned. And the events are exported to userspace with this hole. But in userspace, from perf sched, we fetch it using a structure that has a new field in the beginning: u32 size. This is because our trace is exported with its size as a field. But now that we have this new field, the hole in the middle disappears because it makes prev_state becoming well aligned. And since we are using a pointer to the raw trace using this struct, instead of reading prev_state, we are reading the hole. We could fix it by keeping the size seperate from the struct but actually there a lot of other potential problems: some fields may be saved as long in a 64 bits system and later read as long in a 32 bits system. Also this direct cast doesn't care about the endianness differences between the host traced machine and the machine in which we do the post processing. So instead of using such dangerous direct casts, fetch the values using the trace parsing API that already takes care of all these problems. Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Mike Galbraith <efault@gmx.de> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-09-11 20:43:45 -04:00
static void
replay_wakeup_event(struct trace_wakeup_event *wakeup_event,
struct event *event,
int cpu __used,
u64 timestamp __used,
struct thread *thread __used)
{
struct task_desc *waker, *wakee;
perf sched: Implement the scheduling workload replay engine Integrate the schedbench.c bits with the raw trace events that we get from the perf machinery, and activate the workload replayer/simulator. Example of a captured 'make -j' workload: $ perf sched run measurement overhead: 90 nsecs sleep measurement overhead: 2724743 nsecs the run test took 1000081 nsecs the sleep test took 2981111 nsecs version = 0.5 ... nr_run_events: 70 nr_sleep_events: 66 nr_wakeup_events: 9 target-less wakeups: 71 multi-target wakeups: 47 run events optimized: 139 task 0 ( perf: 6607), nr_events: 2 task 1 ( perf: 6608), nr_events: 6 task 2 ( : 0), nr_events: 1 task 3 ( make: 6609), nr_events: 5 task 4 ( sh: 6610), nr_events: 4 task 5 ( make: 6611), nr_events: 6 task 6 ( sh: 6612), nr_events: 4 task 7 ( make: 6613), nr_events: 5 task 8 ( migration/11: 25), nr_events: 1 task 9 ( migration/13: 29), nr_events: 1 task 10 ( migration/15: 33), nr_events: 1 task 11 ( migration/9: 21), nr_events: 1 task 12 ( sh: 6614), nr_events: 4 task 13 ( make: 6615), nr_events: 5 task 14 ( sh: 6616), nr_events: 4 task 15 ( make: 6617), nr_events: 7 task 16 ( migration/3: 9), nr_events: 1 task 17 ( migration/5: 13), nr_events: 1 task 18 ( migration/7: 17), nr_events: 1 task 19 ( migration/1: 5), nr_events: 1 task 20 ( sh: 6618), nr_events: 4 task 21 ( make: 6619), nr_events: 5 task 22 ( sh: 6620), nr_events: 4 task 23 ( make: 6621), nr_events: 10 task 24 ( sh: 6623), nr_events: 3 task 25 ( gcc: 6624), nr_events: 4 task 26 ( gcc: 6625), nr_events: 4 task 27 ( gcc: 6626), nr_events: 5 task 28 ( collect2: 6627), nr_events: 5 task 29 ( sh: 6622), nr_events: 1 task 30 ( make: 6628), nr_events: 7 task 31 ( sh: 6630), nr_events: 4 task 32 ( gcc: 6631), nr_events: 4 task 33 ( sh: 6629), nr_events: 1 task 34 ( gcc: 6632), nr_events: 4 task 35 ( gcc: 6633), nr_events: 4 task 36 ( collect2: 6634), nr_events: 4 task 37 ( make: 6635), nr_events: 8 task 38 ( sh: 6637), nr_events: 4 task 39 ( sh: 6636), nr_events: 1 task 40 ( gcc: 6638), nr_events: 4 task 41 ( gcc: 6639), nr_events: 4 task 42 ( gcc: 6640), nr_events: 4 task 43 ( collect2: 6641), nr_events: 4 task 44 ( make: 6642), nr_events: 6 task 45 ( sh: 6643), nr_events: 5 task 46 ( sh: 6644), nr_events: 3 task 47 ( sh: 6645), nr_events: 4 task 48 ( make: 6646), nr_events: 6 task 49 ( sh: 6647), nr_events: 3 task 50 ( make: 6648), nr_events: 5 task 51 ( sh: 6649), nr_events: 5 task 52 ( sh: 6650), nr_events: 6 task 53 ( make: 6651), nr_events: 4 task 54 ( make: 6652), nr_events: 5 task 55 ( make: 6653), nr_events: 4 task 56 ( make: 6654), nr_events: 4 task 57 ( make: 6655), nr_events: 5 task 58 ( sh: 6656), nr_events: 4 task 59 ( gcc: 6657), nr_events: 9 task 60 ( ksoftirqd/3: 10), nr_events: 1 task 61 ( gcc: 6658), nr_events: 4 task 62 ( make: 6659), nr_events: 5 task 63 ( sh: 6660), nr_events: 3 task 64 ( gcc: 6661), nr_events: 5 task 65 ( collect2: 6662), nr_events: 4 ------------------------------------------------------------ #1 : 256.745, ravg: 256.74, cpu: 0.00 / 0.00 #2 : 439.372, ravg: 275.01, cpu: 0.00 / 0.00 #3 : 411.971, ravg: 288.70, cpu: 0.00 / 0.00 #4 : 385.500, ravg: 298.38, cpu: 0.00 / 0.00 #5 : 366.526, ravg: 305.20, cpu: 0.00 / 0.00 #6 : 381.281, ravg: 312.81, cpu: 0.00 / 0.00 #7 : 410.756, ravg: 322.60, cpu: 0.00 / 0.00 #8 : 368.009, ravg: 327.14, cpu: 0.00 / 0.00 #9 : 408.098, ravg: 335.24, cpu: 0.00 / 0.00 #10 : 368.582, ravg: 338.57, cpu: 0.00 / 0.00 I.e. we successfully analyzed the trace, replayed it via real threads and measured the replayed workload's scheduling properties. This is how it looked like in 'top' output: PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND 7164 mingo 20 0 1434m 8080 888 R 57.0 0.1 0:02.04 :perf 7165 mingo 20 0 1434m 8080 888 R 41.8 0.1 0:01.52 :perf 7228 mingo 20 0 1434m 8080 888 R 39.8 0.1 0:01.44 :gcc 7225 mingo 20 0 1434m 8080 888 R 33.8 0.1 0:01.26 :gcc 7202 mingo 20 0 1434m 8080 888 R 31.2 0.1 0:01.16 :sh 7222 mingo 20 0 1434m 8080 888 R 25.2 0.1 0:00.96 :sh 7211 mingo 20 0 1434m 8080 888 R 21.9 0.1 0:00.82 :sh 7213 mingo 20 0 1434m 8080 888 D 19.2 0.1 0:00.74 :sh 7194 mingo 20 0 1434m 8080 888 D 18.6 0.1 0:00.72 :make There's still various kinks in it - more patches to come. Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Mike Galbraith <efault@gmx.de> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> LKML-Reference: <new-submission> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-09-11 06:12:54 -04:00
if (verbose) {
printf("sched_wakeup event %p\n", event);
perf sched: Implement the scheduling workload replay engine Integrate the schedbench.c bits with the raw trace events that we get from the perf machinery, and activate the workload replayer/simulator. Example of a captured 'make -j' workload: $ perf sched run measurement overhead: 90 nsecs sleep measurement overhead: 2724743 nsecs the run test took 1000081 nsecs the sleep test took 2981111 nsecs version = 0.5 ... nr_run_events: 70 nr_sleep_events: 66 nr_wakeup_events: 9 target-less wakeups: 71 multi-target wakeups: 47 run events optimized: 139 task 0 ( perf: 6607), nr_events: 2 task 1 ( perf: 6608), nr_events: 6 task 2 ( : 0), nr_events: 1 task 3 ( make: 6609), nr_events: 5 task 4 ( sh: 6610), nr_events: 4 task 5 ( make: 6611), nr_events: 6 task 6 ( sh: 6612), nr_events: 4 task 7 ( make: 6613), nr_events: 5 task 8 ( migration/11: 25), nr_events: 1 task 9 ( migration/13: 29), nr_events: 1 task 10 ( migration/15: 33), nr_events: 1 task 11 ( migration/9: 21), nr_events: 1 task 12 ( sh: 6614), nr_events: 4 task 13 ( make: 6615), nr_events: 5 task 14 ( sh: 6616), nr_events: 4 task 15 ( make: 6617), nr_events: 7 task 16 ( migration/3: 9), nr_events: 1 task 17 ( migration/5: 13), nr_events: 1 task 18 ( migration/7: 17), nr_events: 1 task 19 ( migration/1: 5), nr_events: 1 task 20 ( sh: 6618), nr_events: 4 task 21 ( make: 6619), nr_events: 5 task 22 ( sh: 6620), nr_events: 4 task 23 ( make: 6621), nr_events: 10 task 24 ( sh: 6623), nr_events: 3 task 25 ( gcc: 6624), nr_events: 4 task 26 ( gcc: 6625), nr_events: 4 task 27 ( gcc: 6626), nr_events: 5 task 28 ( collect2: 6627), nr_events: 5 task 29 ( sh: 6622), nr_events: 1 task 30 ( make: 6628), nr_events: 7 task 31 ( sh: 6630), nr_events: 4 task 32 ( gcc: 6631), nr_events: 4 task 33 ( sh: 6629), nr_events: 1 task 34 ( gcc: 6632), nr_events: 4 task 35 ( gcc: 6633), nr_events: 4 task 36 ( collect2: 6634), nr_events: 4 task 37 ( make: 6635), nr_events: 8 task 38 ( sh: 6637), nr_events: 4 task 39 ( sh: 6636), nr_events: 1 task 40 ( gcc: 6638), nr_events: 4 task 41 ( gcc: 6639), nr_events: 4 task 42 ( gcc: 6640), nr_events: 4 task 43 ( collect2: 6641), nr_events: 4 task 44 ( make: 6642), nr_events: 6 task 45 ( sh: 6643), nr_events: 5 task 46 ( sh: 6644), nr_events: 3 task 47 ( sh: 6645), nr_events: 4 task 48 ( make: 6646), nr_events: 6 task 49 ( sh: 6647), nr_events: 3 task 50 ( make: 6648), nr_events: 5 task 51 ( sh: 6649), nr_events: 5 task 52 ( sh: 6650), nr_events: 6 task 53 ( make: 6651), nr_events: 4 task 54 ( make: 6652), nr_events: 5 task 55 ( make: 6653), nr_events: 4 task 56 ( make: 6654), nr_events: 4 task 57 ( make: 6655), nr_events: 5 task 58 ( sh: 6656), nr_events: 4 task 59 ( gcc: 6657), nr_events: 9 task 60 ( ksoftirqd/3: 10), nr_events: 1 task 61 ( gcc: 6658), nr_events: 4 task 62 ( make: 6659), nr_events: 5 task 63 ( sh: 6660), nr_events: 3 task 64 ( gcc: 6661), nr_events: 5 task 65 ( collect2: 6662), nr_events: 4 ------------------------------------------------------------ #1 : 256.745, ravg: 256.74, cpu: 0.00 / 0.00 #2 : 439.372, ravg: 275.01, cpu: 0.00 / 0.00 #3 : 411.971, ravg: 288.70, cpu: 0.00 / 0.00 #4 : 385.500, ravg: 298.38, cpu: 0.00 / 0.00 #5 : 366.526, ravg: 305.20, cpu: 0.00 / 0.00 #6 : 381.281, ravg: 312.81, cpu: 0.00 / 0.00 #7 : 410.756, ravg: 322.60, cpu: 0.00 / 0.00 #8 : 368.009, ravg: 327.14, cpu: 0.00 / 0.00 #9 : 408.098, ravg: 335.24, cpu: 0.00 / 0.00 #10 : 368.582, ravg: 338.57, cpu: 0.00 / 0.00 I.e. we successfully analyzed the trace, replayed it via real threads and measured the replayed workload's scheduling properties. This is how it looked like in 'top' output: PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND 7164 mingo 20 0 1434m 8080 888 R 57.0 0.1 0:02.04 :perf 7165 mingo 20 0 1434m 8080 888 R 41.8 0.1 0:01.52 :perf 7228 mingo 20 0 1434m 8080 888 R 39.8 0.1 0:01.44 :gcc 7225 mingo 20 0 1434m 8080 888 R 33.8 0.1 0:01.26 :gcc 7202 mingo 20 0 1434m 8080 888 R 31.2 0.1 0:01.16 :sh 7222 mingo 20 0 1434m 8080 888 R 25.2 0.1 0:00.96 :sh 7211 mingo 20 0 1434m 8080 888 R 21.9 0.1 0:00.82 :sh 7213 mingo 20 0 1434m 8080 888 D 19.2 0.1 0:00.74 :sh 7194 mingo 20 0 1434m 8080 888 D 18.6 0.1 0:00.72 :make There's still various kinks in it - more patches to come. Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Mike Galbraith <efault@gmx.de> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> LKML-Reference: <new-submission> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-09-11 06:12:54 -04:00
printf(" ... pid %d woke up %s/%d\n",
wakeup_event->common_pid,
wakeup_event->comm,
wakeup_event->pid);
}
perf sched: Implement the scheduling workload replay engine Integrate the schedbench.c bits with the raw trace events that we get from the perf machinery, and activate the workload replayer/simulator. Example of a captured 'make -j' workload: $ perf sched run measurement overhead: 90 nsecs sleep measurement overhead: 2724743 nsecs the run test took 1000081 nsecs the sleep test took 2981111 nsecs version = 0.5 ... nr_run_events: 70 nr_sleep_events: 66 nr_wakeup_events: 9 target-less wakeups: 71 multi-target wakeups: 47 run events optimized: 139 task 0 ( perf: 6607), nr_events: 2 task 1 ( perf: 6608), nr_events: 6 task 2 ( : 0), nr_events: 1 task 3 ( make: 6609), nr_events: 5 task 4 ( sh: 6610), nr_events: 4 task 5 ( make: 6611), nr_events: 6 task 6 ( sh: 6612), nr_events: 4 task 7 ( make: 6613), nr_events: 5 task 8 ( migration/11: 25), nr_events: 1 task 9 ( migration/13: 29), nr_events: 1 task 10 ( migration/15: 33), nr_events: 1 task 11 ( migration/9: 21), nr_events: 1 task 12 ( sh: 6614), nr_events: 4 task 13 ( make: 6615), nr_events: 5 task 14 ( sh: 6616), nr_events: 4 task 15 ( make: 6617), nr_events: 7 task 16 ( migration/3: 9), nr_events: 1 task 17 ( migration/5: 13), nr_events: 1 task 18 ( migration/7: 17), nr_events: 1 task 19 ( migration/1: 5), nr_events: 1 task 20 ( sh: 6618), nr_events: 4 task 21 ( make: 6619), nr_events: 5 task 22 ( sh: 6620), nr_events: 4 task 23 ( make: 6621), nr_events: 10 task 24 ( sh: 6623), nr_events: 3 task 25 ( gcc: 6624), nr_events: 4 task 26 ( gcc: 6625), nr_events: 4 task 27 ( gcc: 6626), nr_events: 5 task 28 ( collect2: 6627), nr_events: 5 task 29 ( sh: 6622), nr_events: 1 task 30 ( make: 6628), nr_events: 7 task 31 ( sh: 6630), nr_events: 4 task 32 ( gcc: 6631), nr_events: 4 task 33 ( sh: 6629), nr_events: 1 task 34 ( gcc: 6632), nr_events: 4 task 35 ( gcc: 6633), nr_events: 4 task 36 ( collect2: 6634), nr_events: 4 task 37 ( make: 6635), nr_events: 8 task 38 ( sh: 6637), nr_events: 4 task 39 ( sh: 6636), nr_events: 1 task 40 ( gcc: 6638), nr_events: 4 task 41 ( gcc: 6639), nr_events: 4 task 42 ( gcc: 6640), nr_events: 4 task 43 ( collect2: 6641), nr_events: 4 task 44 ( make: 6642), nr_events: 6 task 45 ( sh: 6643), nr_events: 5 task 46 ( sh: 6644), nr_events: 3 task 47 ( sh: 6645), nr_events: 4 task 48 ( make: 6646), nr_events: 6 task 49 ( sh: 6647), nr_events: 3 task 50 ( make: 6648), nr_events: 5 task 51 ( sh: 6649), nr_events: 5 task 52 ( sh: 6650), nr_events: 6 task 53 ( make: 6651), nr_events: 4 task 54 ( make: 6652), nr_events: 5 task 55 ( make: 6653), nr_events: 4 task 56 ( make: 6654), nr_events: 4 task 57 ( make: 6655), nr_events: 5 task 58 ( sh: 6656), nr_events: 4 task 59 ( gcc: 6657), nr_events: 9 task 60 ( ksoftirqd/3: 10), nr_events: 1 task 61 ( gcc: 6658), nr_events: 4 task 62 ( make: 6659), nr_events: 5 task 63 ( sh: 6660), nr_events: 3 task 64 ( gcc: 6661), nr_events: 5 task 65 ( collect2: 6662), nr_events: 4 ------------------------------------------------------------ #1 : 256.745, ravg: 256.74, cpu: 0.00 / 0.00 #2 : 439.372, ravg: 275.01, cpu: 0.00 / 0.00 #3 : 411.971, ravg: 288.70, cpu: 0.00 / 0.00 #4 : 385.500, ravg: 298.38, cpu: 0.00 / 0.00 #5 : 366.526, ravg: 305.20, cpu: 0.00 / 0.00 #6 : 381.281, ravg: 312.81, cpu: 0.00 / 0.00 #7 : 410.756, ravg: 322.60, cpu: 0.00 / 0.00 #8 : 368.009, ravg: 327.14, cpu: 0.00 / 0.00 #9 : 408.098, ravg: 335.24, cpu: 0.00 / 0.00 #10 : 368.582, ravg: 338.57, cpu: 0.00 / 0.00 I.e. we successfully analyzed the trace, replayed it via real threads and measured the replayed workload's scheduling properties. This is how it looked like in 'top' output: PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND 7164 mingo 20 0 1434m 8080 888 R 57.0 0.1 0:02.04 :perf 7165 mingo 20 0 1434m 8080 888 R 41.8 0.1 0:01.52 :perf 7228 mingo 20 0 1434m 8080 888 R 39.8 0.1 0:01.44 :gcc 7225 mingo 20 0 1434m 8080 888 R 33.8 0.1 0:01.26 :gcc 7202 mingo 20 0 1434m 8080 888 R 31.2 0.1 0:01.16 :sh 7222 mingo 20 0 1434m 8080 888 R 25.2 0.1 0:00.96 :sh 7211 mingo 20 0 1434m 8080 888 R 21.9 0.1 0:00.82 :sh 7213 mingo 20 0 1434m 8080 888 D 19.2 0.1 0:00.74 :sh 7194 mingo 20 0 1434m 8080 888 D 18.6 0.1 0:00.72 :make There's still various kinks in it - more patches to come. Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Mike Galbraith <efault@gmx.de> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> LKML-Reference: <new-submission> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-09-11 06:12:54 -04:00
waker = register_pid(wakeup_event->common_pid, "<unknown>");
wakee = register_pid(wakeup_event->pid, wakeup_event->comm);
perf sched: Implement the scheduling workload replay engine Integrate the schedbench.c bits with the raw trace events that we get from the perf machinery, and activate the workload replayer/simulator. Example of a captured 'make -j' workload: $ perf sched run measurement overhead: 90 nsecs sleep measurement overhead: 2724743 nsecs the run test took 1000081 nsecs the sleep test took 2981111 nsecs version = 0.5 ... nr_run_events: 70 nr_sleep_events: 66 nr_wakeup_events: 9 target-less wakeups: 71 multi-target wakeups: 47 run events optimized: 139 task 0 ( perf: 6607), nr_events: 2 task 1 ( perf: 6608), nr_events: 6 task 2 ( : 0), nr_events: 1 task 3 ( make: 6609), nr_events: 5 task 4 ( sh: 6610), nr_events: 4 task 5 ( make: 6611), nr_events: 6 task 6 ( sh: 6612), nr_events: 4 task 7 ( make: 6613), nr_events: 5 task 8 ( migration/11: 25), nr_events: 1 task 9 ( migration/13: 29), nr_events: 1 task 10 ( migration/15: 33), nr_events: 1 task 11 ( migration/9: 21), nr_events: 1 task 12 ( sh: 6614), nr_events: 4 task 13 ( make: 6615), nr_events: 5 task 14 ( sh: 6616), nr_events: 4 task 15 ( make: 6617), nr_events: 7 task 16 ( migration/3: 9), nr_events: 1 task 17 ( migration/5: 13), nr_events: 1 task 18 ( migration/7: 17), nr_events: 1 task 19 ( migration/1: 5), nr_events: 1 task 20 ( sh: 6618), nr_events: 4 task 21 ( make: 6619), nr_events: 5 task 22 ( sh: 6620), nr_events: 4 task 23 ( make: 6621), nr_events: 10 task 24 ( sh: 6623), nr_events: 3 task 25 ( gcc: 6624), nr_events: 4 task 26 ( gcc: 6625), nr_events: 4 task 27 ( gcc: 6626), nr_events: 5 task 28 ( collect2: 6627), nr_events: 5 task 29 ( sh: 6622), nr_events: 1 task 30 ( make: 6628), nr_events: 7 task 31 ( sh: 6630), nr_events: 4 task 32 ( gcc: 6631), nr_events: 4 task 33 ( sh: 6629), nr_events: 1 task 34 ( gcc: 6632), nr_events: 4 task 35 ( gcc: 6633), nr_events: 4 task 36 ( collect2: 6634), nr_events: 4 task 37 ( make: 6635), nr_events: 8 task 38 ( sh: 6637), nr_events: 4 task 39 ( sh: 6636), nr_events: 1 task 40 ( gcc: 6638), nr_events: 4 task 41 ( gcc: 6639), nr_events: 4 task 42 ( gcc: 6640), nr_events: 4 task 43 ( collect2: 6641), nr_events: 4 task 44 ( make: 6642), nr_events: 6 task 45 ( sh: 6643), nr_events: 5 task 46 ( sh: 6644), nr_events: 3 task 47 ( sh: 6645), nr_events: 4 task 48 ( make: 6646), nr_events: 6 task 49 ( sh: 6647), nr_events: 3 task 50 ( make: 6648), nr_events: 5 task 51 ( sh: 6649), nr_events: 5 task 52 ( sh: 6650), nr_events: 6 task 53 ( make: 6651), nr_events: 4 task 54 ( make: 6652), nr_events: 5 task 55 ( make: 6653), nr_events: 4 task 56 ( make: 6654), nr_events: 4 task 57 ( make: 6655), nr_events: 5 task 58 ( sh: 6656), nr_events: 4 task 59 ( gcc: 6657), nr_events: 9 task 60 ( ksoftirqd/3: 10), nr_events: 1 task 61 ( gcc: 6658), nr_events: 4 task 62 ( make: 6659), nr_events: 5 task 63 ( sh: 6660), nr_events: 3 task 64 ( gcc: 6661), nr_events: 5 task 65 ( collect2: 6662), nr_events: 4 ------------------------------------------------------------ #1 : 256.745, ravg: 256.74, cpu: 0.00 / 0.00 #2 : 439.372, ravg: 275.01, cpu: 0.00 / 0.00 #3 : 411.971, ravg: 288.70, cpu: 0.00 / 0.00 #4 : 385.500, ravg: 298.38, cpu: 0.00 / 0.00 #5 : 366.526, ravg: 305.20, cpu: 0.00 / 0.00 #6 : 381.281, ravg: 312.81, cpu: 0.00 / 0.00 #7 : 410.756, ravg: 322.60, cpu: 0.00 / 0.00 #8 : 368.009, ravg: 327.14, cpu: 0.00 / 0.00 #9 : 408.098, ravg: 335.24, cpu: 0.00 / 0.00 #10 : 368.582, ravg: 338.57, cpu: 0.00 / 0.00 I.e. we successfully analyzed the trace, replayed it via real threads and measured the replayed workload's scheduling properties. This is how it looked like in 'top' output: PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND 7164 mingo 20 0 1434m 8080 888 R 57.0 0.1 0:02.04 :perf 7165 mingo 20 0 1434m 8080 888 R 41.8 0.1 0:01.52 :perf 7228 mingo 20 0 1434m 8080 888 R 39.8 0.1 0:01.44 :gcc 7225 mingo 20 0 1434m 8080 888 R 33.8 0.1 0:01.26 :gcc 7202 mingo 20 0 1434m 8080 888 R 31.2 0.1 0:01.16 :sh 7222 mingo 20 0 1434m 8080 888 R 25.2 0.1 0:00.96 :sh 7211 mingo 20 0 1434m 8080 888 R 21.9 0.1 0:00.82 :sh 7213 mingo 20 0 1434m 8080 888 D 19.2 0.1 0:00.74 :sh 7194 mingo 20 0 1434m 8080 888 D 18.6 0.1 0:00.72 :make There's still various kinks in it - more patches to come. Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Mike Galbraith <efault@gmx.de> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> LKML-Reference: <new-submission> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-09-11 06:12:54 -04:00
add_sched_event_wakeup(waker, timestamp, wakee);
}
static unsigned long cpu_last_switched[MAX_CPUS];
perf sched: Implement the scheduling workload replay engine Integrate the schedbench.c bits with the raw trace events that we get from the perf machinery, and activate the workload replayer/simulator. Example of a captured 'make -j' workload: $ perf sched run measurement overhead: 90 nsecs sleep measurement overhead: 2724743 nsecs the run test took 1000081 nsecs the sleep test took 2981111 nsecs version = 0.5 ... nr_run_events: 70 nr_sleep_events: 66 nr_wakeup_events: 9 target-less wakeups: 71 multi-target wakeups: 47 run events optimized: 139 task 0 ( perf: 6607), nr_events: 2 task 1 ( perf: 6608), nr_events: 6 task 2 ( : 0), nr_events: 1 task 3 ( make: 6609), nr_events: 5 task 4 ( sh: 6610), nr_events: 4 task 5 ( make: 6611), nr_events: 6 task 6 ( sh: 6612), nr_events: 4 task 7 ( make: 6613), nr_events: 5 task 8 ( migration/11: 25), nr_events: 1 task 9 ( migration/13: 29), nr_events: 1 task 10 ( migration/15: 33), nr_events: 1 task 11 ( migration/9: 21), nr_events: 1 task 12 ( sh: 6614), nr_events: 4 task 13 ( make: 6615), nr_events: 5 task 14 ( sh: 6616), nr_events: 4 task 15 ( make: 6617), nr_events: 7 task 16 ( migration/3: 9), nr_events: 1 task 17 ( migration/5: 13), nr_events: 1 task 18 ( migration/7: 17), nr_events: 1 task 19 ( migration/1: 5), nr_events: 1 task 20 ( sh: 6618), nr_events: 4 task 21 ( make: 6619), nr_events: 5 task 22 ( sh: 6620), nr_events: 4 task 23 ( make: 6621), nr_events: 10 task 24 ( sh: 6623), nr_events: 3 task 25 ( gcc: 6624), nr_events: 4 task 26 ( gcc: 6625), nr_events: 4 task 27 ( gcc: 6626), nr_events: 5 task 28 ( collect2: 6627), nr_events: 5 task 29 ( sh: 6622), nr_events: 1 task 30 ( make: 6628), nr_events: 7 task 31 ( sh: 6630), nr_events: 4 task 32 ( gcc: 6631), nr_events: 4 task 33 ( sh: 6629), nr_events: 1 task 34 ( gcc: 6632), nr_events: 4 task 35 ( gcc: 6633), nr_events: 4 task 36 ( collect2: 6634), nr_events: 4 task 37 ( make: 6635), nr_events: 8 task 38 ( sh: 6637), nr_events: 4 task 39 ( sh: 6636), nr_events: 1 task 40 ( gcc: 6638), nr_events: 4 task 41 ( gcc: 6639), nr_events: 4 task 42 ( gcc: 6640), nr_events: 4 task 43 ( collect2: 6641), nr_events: 4 task 44 ( make: 6642), nr_events: 6 task 45 ( sh: 6643), nr_events: 5 task 46 ( sh: 6644), nr_events: 3 task 47 ( sh: 6645), nr_events: 4 task 48 ( make: 6646), nr_events: 6 task 49 ( sh: 6647), nr_events: 3 task 50 ( make: 6648), nr_events: 5 task 51 ( sh: 6649), nr_events: 5 task 52 ( sh: 6650), nr_events: 6 task 53 ( make: 6651), nr_events: 4 task 54 ( make: 6652), nr_events: 5 task 55 ( make: 6653), nr_events: 4 task 56 ( make: 6654), nr_events: 4 task 57 ( make: 6655), nr_events: 5 task 58 ( sh: 6656), nr_events: 4 task 59 ( gcc: 6657), nr_events: 9 task 60 ( ksoftirqd/3: 10), nr_events: 1 task 61 ( gcc: 6658), nr_events: 4 task 62 ( make: 6659), nr_events: 5 task 63 ( sh: 6660), nr_events: 3 task 64 ( gcc: 6661), nr_events: 5 task 65 ( collect2: 6662), nr_events: 4 ------------------------------------------------------------ #1 : 256.745, ravg: 256.74, cpu: 0.00 / 0.00 #2 : 439.372, ravg: 275.01, cpu: 0.00 / 0.00 #3 : 411.971, ravg: 288.70, cpu: 0.00 / 0.00 #4 : 385.500, ravg: 298.38, cpu: 0.00 / 0.00 #5 : 366.526, ravg: 305.20, cpu: 0.00 / 0.00 #6 : 381.281, ravg: 312.81, cpu: 0.00 / 0.00 #7 : 410.756, ravg: 322.60, cpu: 0.00 / 0.00 #8 : 368.009, ravg: 327.14, cpu: 0.00 / 0.00 #9 : 408.098, ravg: 335.24, cpu: 0.00 / 0.00 #10 : 368.582, ravg: 338.57, cpu: 0.00 / 0.00 I.e. we successfully analyzed the trace, replayed it via real threads and measured the replayed workload's scheduling properties. This is how it looked like in 'top' output: PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND 7164 mingo 20 0 1434m 8080 888 R 57.0 0.1 0:02.04 :perf 7165 mingo 20 0 1434m 8080 888 R 41.8 0.1 0:01.52 :perf 7228 mingo 20 0 1434m 8080 888 R 39.8 0.1 0:01.44 :gcc 7225 mingo 20 0 1434m 8080 888 R 33.8 0.1 0:01.26 :gcc 7202 mingo 20 0 1434m 8080 888 R 31.2 0.1 0:01.16 :sh 7222 mingo 20 0 1434m 8080 888 R 25.2 0.1 0:00.96 :sh 7211 mingo 20 0 1434m 8080 888 R 21.9 0.1 0:00.82 :sh 7213 mingo 20 0 1434m 8080 888 D 19.2 0.1 0:00.74 :sh 7194 mingo 20 0 1434m 8080 888 D 18.6 0.1 0:00.72 :make There's still various kinks in it - more patches to come. Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Mike Galbraith <efault@gmx.de> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> LKML-Reference: <new-submission> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-09-11 06:12:54 -04:00
static void
replay_switch_event(struct trace_switch_event *switch_event,
struct event *event,
int cpu,
u64 timestamp,
struct thread *thread __used)
{
perf sched: Implement the scheduling workload replay engine Integrate the schedbench.c bits with the raw trace events that we get from the perf machinery, and activate the workload replayer/simulator. Example of a captured 'make -j' workload: $ perf sched run measurement overhead: 90 nsecs sleep measurement overhead: 2724743 nsecs the run test took 1000081 nsecs the sleep test took 2981111 nsecs version = 0.5 ... nr_run_events: 70 nr_sleep_events: 66 nr_wakeup_events: 9 target-less wakeups: 71 multi-target wakeups: 47 run events optimized: 139 task 0 ( perf: 6607), nr_events: 2 task 1 ( perf: 6608), nr_events: 6 task 2 ( : 0), nr_events: 1 task 3 ( make: 6609), nr_events: 5 task 4 ( sh: 6610), nr_events: 4 task 5 ( make: 6611), nr_events: 6 task 6 ( sh: 6612), nr_events: 4 task 7 ( make: 6613), nr_events: 5 task 8 ( migration/11: 25), nr_events: 1 task 9 ( migration/13: 29), nr_events: 1 task 10 ( migration/15: 33), nr_events: 1 task 11 ( migration/9: 21), nr_events: 1 task 12 ( sh: 6614), nr_events: 4 task 13 ( make: 6615), nr_events: 5 task 14 ( sh: 6616), nr_events: 4 task 15 ( make: 6617), nr_events: 7 task 16 ( migration/3: 9), nr_events: 1 task 17 ( migration/5: 13), nr_events: 1 task 18 ( migration/7: 17), nr_events: 1 task 19 ( migration/1: 5), nr_events: 1 task 20 ( sh: 6618), nr_events: 4 task 21 ( make: 6619), nr_events: 5 task 22 ( sh: 6620), nr_events: 4 task 23 ( make: 6621), nr_events: 10 task 24 ( sh: 6623), nr_events: 3 task 25 ( gcc: 6624), nr_events: 4 task 26 ( gcc: 6625), nr_events: 4 task 27 ( gcc: 6626), nr_events: 5 task 28 ( collect2: 6627), nr_events: 5 task 29 ( sh: 6622), nr_events: 1 task 30 ( make: 6628), nr_events: 7 task 31 ( sh: 6630), nr_events: 4 task 32 ( gcc: 6631), nr_events: 4 task 33 ( sh: 6629), nr_events: 1 task 34 ( gcc: 6632), nr_events: 4 task 35 ( gcc: 6633), nr_events: 4 task 36 ( collect2: 6634), nr_events: 4 task 37 ( make: 6635), nr_events: 8 task 38 ( sh: 6637), nr_events: 4 task 39 ( sh: 6636), nr_events: 1 task 40 ( gcc: 6638), nr_events: 4 task 41 ( gcc: 6639), nr_events: 4 task 42 ( gcc: 6640), nr_events: 4 task 43 ( collect2: 6641), nr_events: 4 task 44 ( make: 6642), nr_events: 6 task 45 ( sh: 6643), nr_events: 5 task 46 ( sh: 6644), nr_events: 3 task 47 ( sh: 6645), nr_events: 4 task 48 ( make: 6646), nr_events: 6 task 49 ( sh: 6647), nr_events: 3 task 50 ( make: 6648), nr_events: 5 task 51 ( sh: 6649), nr_events: 5 task 52 ( sh: 6650), nr_events: 6 task 53 ( make: 6651), nr_events: 4 task 54 ( make: 6652), nr_events: 5 task 55 ( make: 6653), nr_events: 4 task 56 ( make: 6654), nr_events: 4 task 57 ( make: 6655), nr_events: 5 task 58 ( sh: 6656), nr_events: 4 task 59 ( gcc: 6657), nr_events: 9 task 60 ( ksoftirqd/3: 10), nr_events: 1 task 61 ( gcc: 6658), nr_events: 4 task 62 ( make: 6659), nr_events: 5 task 63 ( sh: 6660), nr_events: 3 task 64 ( gcc: 6661), nr_events: 5 task 65 ( collect2: 6662), nr_events: 4 ------------------------------------------------------------ #1 : 256.745, ravg: 256.74, cpu: 0.00 / 0.00 #2 : 439.372, ravg: 275.01, cpu: 0.00 / 0.00 #3 : 411.971, ravg: 288.70, cpu: 0.00 / 0.00 #4 : 385.500, ravg: 298.38, cpu: 0.00 / 0.00 #5 : 366.526, ravg: 305.20, cpu: 0.00 / 0.00 #6 : 381.281, ravg: 312.81, cpu: 0.00 / 0.00 #7 : 410.756, ravg: 322.60, cpu: 0.00 / 0.00 #8 : 368.009, ravg: 327.14, cpu: 0.00 / 0.00 #9 : 408.098, ravg: 335.24, cpu: 0.00 / 0.00 #10 : 368.582, ravg: 338.57, cpu: 0.00 / 0.00 I.e. we successfully analyzed the trace, replayed it via real threads and measured the replayed workload's scheduling properties. This is how it looked like in 'top' output: PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND 7164 mingo 20 0 1434m 8080 888 R 57.0 0.1 0:02.04 :perf 7165 mingo 20 0 1434m 8080 888 R 41.8 0.1 0:01.52 :perf 7228 mingo 20 0 1434m 8080 888 R 39.8 0.1 0:01.44 :gcc 7225 mingo 20 0 1434m 8080 888 R 33.8 0.1 0:01.26 :gcc 7202 mingo 20 0 1434m 8080 888 R 31.2 0.1 0:01.16 :sh 7222 mingo 20 0 1434m 8080 888 R 25.2 0.1 0:00.96 :sh 7211 mingo 20 0 1434m 8080 888 R 21.9 0.1 0:00.82 :sh 7213 mingo 20 0 1434m 8080 888 D 19.2 0.1 0:00.74 :sh 7194 mingo 20 0 1434m 8080 888 D 18.6 0.1 0:00.72 :make There's still various kinks in it - more patches to come. Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Mike Galbraith <efault@gmx.de> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> LKML-Reference: <new-submission> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-09-11 06:12:54 -04:00
struct task_desc *prev, *next;
u64 timestamp0;
s64 delta;
if (verbose)
printf("sched_switch event %p\n", event);
perf sched: Implement the scheduling workload replay engine Integrate the schedbench.c bits with the raw trace events that we get from the perf machinery, and activate the workload replayer/simulator. Example of a captured 'make -j' workload: $ perf sched run measurement overhead: 90 nsecs sleep measurement overhead: 2724743 nsecs the run test took 1000081 nsecs the sleep test took 2981111 nsecs version = 0.5 ... nr_run_events: 70 nr_sleep_events: 66 nr_wakeup_events: 9 target-less wakeups: 71 multi-target wakeups: 47 run events optimized: 139 task 0 ( perf: 6607), nr_events: 2 task 1 ( perf: 6608), nr_events: 6 task 2 ( : 0), nr_events: 1 task 3 ( make: 6609), nr_events: 5 task 4 ( sh: 6610), nr_events: 4 task 5 ( make: 6611), nr_events: 6 task 6 ( sh: 6612), nr_events: 4 task 7 ( make: 6613), nr_events: 5 task 8 ( migration/11: 25), nr_events: 1 task 9 ( migration/13: 29), nr_events: 1 task 10 ( migration/15: 33), nr_events: 1 task 11 ( migration/9: 21), nr_events: 1 task 12 ( sh: 6614), nr_events: 4 task 13 ( make: 6615), nr_events: 5 task 14 ( sh: 6616), nr_events: 4 task 15 ( make: 6617), nr_events: 7 task 16 ( migration/3: 9), nr_events: 1 task 17 ( migration/5: 13), nr_events: 1 task 18 ( migration/7: 17), nr_events: 1 task 19 ( migration/1: 5), nr_events: 1 task 20 ( sh: 6618), nr_events: 4 task 21 ( make: 6619), nr_events: 5 task 22 ( sh: 6620), nr_events: 4 task 23 ( make: 6621), nr_events: 10 task 24 ( sh: 6623), nr_events: 3 task 25 ( gcc: 6624), nr_events: 4 task 26 ( gcc: 6625), nr_events: 4 task 27 ( gcc: 6626), nr_events: 5 task 28 ( collect2: 6627), nr_events: 5 task 29 ( sh: 6622), nr_events: 1 task 30 ( make: 6628), nr_events: 7 task 31 ( sh: 6630), nr_events: 4 task 32 ( gcc: 6631), nr_events: 4 task 33 ( sh: 6629), nr_events: 1 task 34 ( gcc: 6632), nr_events: 4 task 35 ( gcc: 6633), nr_events: 4 task 36 ( collect2: 6634), nr_events: 4 task 37 ( make: 6635), nr_events: 8 task 38 ( sh: 6637), nr_events: 4 task 39 ( sh: 6636), nr_events: 1 task 40 ( gcc: 6638), nr_events: 4 task 41 ( gcc: 6639), nr_events: 4 task 42 ( gcc: 6640), nr_events: 4 task 43 ( collect2: 6641), nr_events: 4 task 44 ( make: 6642), nr_events: 6 task 45 ( sh: 6643), nr_events: 5 task 46 ( sh: 6644), nr_events: 3 task 47 ( sh: 6645), nr_events: 4 task 48 ( make: 6646), nr_events: 6 task 49 ( sh: 6647), nr_events: 3 task 50 ( make: 6648), nr_events: 5 task 51 ( sh: 6649), nr_events: 5 task 52 ( sh: 6650), nr_events: 6 task 53 ( make: 6651), nr_events: 4 task 54 ( make: 6652), nr_events: 5 task 55 ( make: 6653), nr_events: 4 task 56 ( make: 6654), nr_events: 4 task 57 ( make: 6655), nr_events: 5 task 58 ( sh: 6656), nr_events: 4 task 59 ( gcc: 6657), nr_events: 9 task 60 ( ksoftirqd/3: 10), nr_events: 1 task 61 ( gcc: 6658), nr_events: 4 task 62 ( make: 6659), nr_events: 5 task 63 ( sh: 6660), nr_events: 3 task 64 ( gcc: 6661), nr_events: 5 task 65 ( collect2: 6662), nr_events: 4 ------------------------------------------------------------ #1 : 256.745, ravg: 256.74, cpu: 0.00 / 0.00 #2 : 439.372, ravg: 275.01, cpu: 0.00 / 0.00 #3 : 411.971, ravg: 288.70, cpu: 0.00 / 0.00 #4 : 385.500, ravg: 298.38, cpu: 0.00 / 0.00 #5 : 366.526, ravg: 305.20, cpu: 0.00 / 0.00 #6 : 381.281, ravg: 312.81, cpu: 0.00 / 0.00 #7 : 410.756, ravg: 322.60, cpu: 0.00 / 0.00 #8 : 368.009, ravg: 327.14, cpu: 0.00 / 0.00 #9 : 408.098, ravg: 335.24, cpu: 0.00 / 0.00 #10 : 368.582, ravg: 338.57, cpu: 0.00 / 0.00 I.e. we successfully analyzed the trace, replayed it via real threads and measured the replayed workload's scheduling properties. This is how it looked like in 'top' output: PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND 7164 mingo 20 0 1434m 8080 888 R 57.0 0.1 0:02.04 :perf 7165 mingo 20 0 1434m 8080 888 R 41.8 0.1 0:01.52 :perf 7228 mingo 20 0 1434m 8080 888 R 39.8 0.1 0:01.44 :gcc 7225 mingo 20 0 1434m 8080 888 R 33.8 0.1 0:01.26 :gcc 7202 mingo 20 0 1434m 8080 888 R 31.2 0.1 0:01.16 :sh 7222 mingo 20 0 1434m 8080 888 R 25.2 0.1 0:00.96 :sh 7211 mingo 20 0 1434m 8080 888 R 21.9 0.1 0:00.82 :sh 7213 mingo 20 0 1434m 8080 888 D 19.2 0.1 0:00.74 :sh 7194 mingo 20 0 1434m 8080 888 D 18.6 0.1 0:00.72 :make There's still various kinks in it - more patches to come. Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Mike Galbraith <efault@gmx.de> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> LKML-Reference: <new-submission> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-09-11 06:12:54 -04:00
if (cpu >= MAX_CPUS || cpu < 0)
return;
timestamp0 = cpu_last_switched[cpu];
if (timestamp0)
delta = timestamp - timestamp0;
else
delta = 0;
if (delta < 0)
die("hm, delta: %Ld < 0 ?\n", delta);
if (verbose) {
printf(" ... switch from %s/%d to %s/%d [ran %Ld nsecs]\n",
switch_event->prev_comm, switch_event->prev_pid,
switch_event->next_comm, switch_event->next_pid,
delta);
}
perf sched: Implement the scheduling workload replay engine Integrate the schedbench.c bits with the raw trace events that we get from the perf machinery, and activate the workload replayer/simulator. Example of a captured 'make -j' workload: $ perf sched run measurement overhead: 90 nsecs sleep measurement overhead: 2724743 nsecs the run test took 1000081 nsecs the sleep test took 2981111 nsecs version = 0.5 ... nr_run_events: 70 nr_sleep_events: 66 nr_wakeup_events: 9 target-less wakeups: 71 multi-target wakeups: 47 run events optimized: 139 task 0 ( perf: 6607), nr_events: 2 task 1 ( perf: 6608), nr_events: 6 task 2 ( : 0), nr_events: 1 task 3 ( make: 6609), nr_events: 5 task 4 ( sh: 6610), nr_events: 4 task 5 ( make: 6611), nr_events: 6 task 6 ( sh: 6612), nr_events: 4 task 7 ( make: 6613), nr_events: 5 task 8 ( migration/11: 25), nr_events: 1 task 9 ( migration/13: 29), nr_events: 1 task 10 ( migration/15: 33), nr_events: 1 task 11 ( migration/9: 21), nr_events: 1 task 12 ( sh: 6614), nr_events: 4 task 13 ( make: 6615), nr_events: 5 task 14 ( sh: 6616), nr_events: 4 task 15 ( make: 6617), nr_events: 7 task 16 ( migration/3: 9), nr_events: 1 task 17 ( migration/5: 13), nr_events: 1 task 18 ( migration/7: 17), nr_events: 1 task 19 ( migration/1: 5), nr_events: 1 task 20 ( sh: 6618), nr_events: 4 task 21 ( make: 6619), nr_events: 5 task 22 ( sh: 6620), nr_events: 4 task 23 ( make: 6621), nr_events: 10 task 24 ( sh: 6623), nr_events: 3 task 25 ( gcc: 6624), nr_events: 4 task 26 ( gcc: 6625), nr_events: 4 task 27 ( gcc: 6626), nr_events: 5 task 28 ( collect2: 6627), nr_events: 5 task 29 ( sh: 6622), nr_events: 1 task 30 ( make: 6628), nr_events: 7 task 31 ( sh: 6630), nr_events: 4 task 32 ( gcc: 6631), nr_events: 4 task 33 ( sh: 6629), nr_events: 1 task 34 ( gcc: 6632), nr_events: 4 task 35 ( gcc: 6633), nr_events: 4 task 36 ( collect2: 6634), nr_events: 4 task 37 ( make: 6635), nr_events: 8 task 38 ( sh: 6637), nr_events: 4 task 39 ( sh: 6636), nr_events: 1 task 40 ( gcc: 6638), nr_events: 4 task 41 ( gcc: 6639), nr_events: 4 task 42 ( gcc: 6640), nr_events: 4 task 43 ( collect2: 6641), nr_events: 4 task 44 ( make: 6642), nr_events: 6 task 45 ( sh: 6643), nr_events: 5 task 46 ( sh: 6644), nr_events: 3 task 47 ( sh: 6645), nr_events: 4 task 48 ( make: 6646), nr_events: 6 task 49 ( sh: 6647), nr_events: 3 task 50 ( make: 6648), nr_events: 5 task 51 ( sh: 6649), nr_events: 5 task 52 ( sh: 6650), nr_events: 6 task 53 ( make: 6651), nr_events: 4 task 54 ( make: 6652), nr_events: 5 task 55 ( make: 6653), nr_events: 4 task 56 ( make: 6654), nr_events: 4 task 57 ( make: 6655), nr_events: 5 task 58 ( sh: 6656), nr_events: 4 task 59 ( gcc: 6657), nr_events: 9 task 60 ( ksoftirqd/3: 10), nr_events: 1 task 61 ( gcc: 6658), nr_events: 4 task 62 ( make: 6659), nr_events: 5 task 63 ( sh: 6660), nr_events: 3 task 64 ( gcc: 6661), nr_events: 5 task 65 ( collect2: 6662), nr_events: 4 ------------------------------------------------------------ #1 : 256.745, ravg: 256.74, cpu: 0.00 / 0.00 #2 : 439.372, ravg: 275.01, cpu: 0.00 / 0.00 #3 : 411.971, ravg: 288.70, cpu: 0.00 / 0.00 #4 : 385.500, ravg: 298.38, cpu: 0.00 / 0.00 #5 : 366.526, ravg: 305.20, cpu: 0.00 / 0.00 #6 : 381.281, ravg: 312.81, cpu: 0.00 / 0.00 #7 : 410.756, ravg: 322.60, cpu: 0.00 / 0.00 #8 : 368.009, ravg: 327.14, cpu: 0.00 / 0.00 #9 : 408.098, ravg: 335.24, cpu: 0.00 / 0.00 #10 : 368.582, ravg: 338.57, cpu: 0.00 / 0.00 I.e. we successfully analyzed the trace, replayed it via real threads and measured the replayed workload's scheduling properties. This is how it looked like in 'top' output: PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND 7164 mingo 20 0 1434m 8080 888 R 57.0 0.1 0:02.04 :perf 7165 mingo 20 0 1434m 8080 888 R 41.8 0.1 0:01.52 :perf 7228 mingo 20 0 1434m 8080 888 R 39.8 0.1 0:01.44 :gcc 7225 mingo 20 0 1434m 8080 888 R 33.8 0.1 0:01.26 :gcc 7202 mingo 20 0 1434m 8080 888 R 31.2 0.1 0:01.16 :sh 7222 mingo 20 0 1434m 8080 888 R 25.2 0.1 0:00.96 :sh 7211 mingo 20 0 1434m 8080 888 R 21.9 0.1 0:00.82 :sh 7213 mingo 20 0 1434m 8080 888 D 19.2 0.1 0:00.74 :sh 7194 mingo 20 0 1434m 8080 888 D 18.6 0.1 0:00.72 :make There's still various kinks in it - more patches to come. Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Mike Galbraith <efault@gmx.de> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> LKML-Reference: <new-submission> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-09-11 06:12:54 -04:00
prev = register_pid(switch_event->prev_pid, switch_event->prev_comm);
next = register_pid(switch_event->next_pid, switch_event->next_comm);
perf sched: Implement the scheduling workload replay engine Integrate the schedbench.c bits with the raw trace events that we get from the perf machinery, and activate the workload replayer/simulator. Example of a captured 'make -j' workload: $ perf sched run measurement overhead: 90 nsecs sleep measurement overhead: 2724743 nsecs the run test took 1000081 nsecs the sleep test took 2981111 nsecs version = 0.5 ... nr_run_events: 70 nr_sleep_events: 66 nr_wakeup_events: 9 target-less wakeups: 71 multi-target wakeups: 47 run events optimized: 139 task 0 ( perf: 6607), nr_events: 2 task 1 ( perf: 6608), nr_events: 6 task 2 ( : 0), nr_events: 1 task 3 ( make: 6609), nr_events: 5 task 4 ( sh: 6610), nr_events: 4 task 5 ( make: 6611), nr_events: 6 task 6 ( sh: 6612), nr_events: 4 task 7 ( make: 6613), nr_events: 5 task 8 ( migration/11: 25), nr_events: 1 task 9 ( migration/13: 29), nr_events: 1 task 10 ( migration/15: 33), nr_events: 1 task 11 ( migration/9: 21), nr_events: 1 task 12 ( sh: 6614), nr_events: 4 task 13 ( make: 6615), nr_events: 5 task 14 ( sh: 6616), nr_events: 4 task 15 ( make: 6617), nr_events: 7 task 16 ( migration/3: 9), nr_events: 1 task 17 ( migration/5: 13), nr_events: 1 task 18 ( migration/7: 17), nr_events: 1 task 19 ( migration/1: 5), nr_events: 1 task 20 ( sh: 6618), nr_events: 4 task 21 ( make: 6619), nr_events: 5 task 22 ( sh: 6620), nr_events: 4 task 23 ( make: 6621), nr_events: 10 task 24 ( sh: 6623), nr_events: 3 task 25 ( gcc: 6624), nr_events: 4 task 26 ( gcc: 6625), nr_events: 4 task 27 ( gcc: 6626), nr_events: 5 task 28 ( collect2: 6627), nr_events: 5 task 29 ( sh: 6622), nr_events: 1 task 30 ( make: 6628), nr_events: 7 task 31 ( sh: 6630), nr_events: 4 task 32 ( gcc: 6631), nr_events: 4 task 33 ( sh: 6629), nr_events: 1 task 34 ( gcc: 6632), nr_events: 4 task 35 ( gcc: 6633), nr_events: 4 task 36 ( collect2: 6634), nr_events: 4 task 37 ( make: 6635), nr_events: 8 task 38 ( sh: 6637), nr_events: 4 task 39 ( sh: 6636), nr_events: 1 task 40 ( gcc: 6638), nr_events: 4 task 41 ( gcc: 6639), nr_events: 4 task 42 ( gcc: 6640), nr_events: 4 task 43 ( collect2: 6641), nr_events: 4 task 44 ( make: 6642), nr_events: 6 task 45 ( sh: 6643), nr_events: 5 task 46 ( sh: 6644), nr_events: 3 task 47 ( sh: 6645), nr_events: 4 task 48 ( make: 6646), nr_events: 6 task 49 ( sh: 6647), nr_events: 3 task 50 ( make: 6648), nr_events: 5 task 51 ( sh: 6649), nr_events: 5 task 52 ( sh: 6650), nr_events: 6 task 53 ( make: 6651), nr_events: 4 task 54 ( make: 6652), nr_events: 5 task 55 ( make: 6653), nr_events: 4 task 56 ( make: 6654), nr_events: 4 task 57 ( make: 6655), nr_events: 5 task 58 ( sh: 6656), nr_events: 4 task 59 ( gcc: 6657), nr_events: 9 task 60 ( ksoftirqd/3: 10), nr_events: 1 task 61 ( gcc: 6658), nr_events: 4 task 62 ( make: 6659), nr_events: 5 task 63 ( sh: 6660), nr_events: 3 task 64 ( gcc: 6661), nr_events: 5 task 65 ( collect2: 6662), nr_events: 4 ------------------------------------------------------------ #1 : 256.745, ravg: 256.74, cpu: 0.00 / 0.00 #2 : 439.372, ravg: 275.01, cpu: 0.00 / 0.00 #3 : 411.971, ravg: 288.70, cpu: 0.00 / 0.00 #4 : 385.500, ravg: 298.38, cpu: 0.00 / 0.00 #5 : 366.526, ravg: 305.20, cpu: 0.00 / 0.00 #6 : 381.281, ravg: 312.81, cpu: 0.00 / 0.00 #7 : 410.756, ravg: 322.60, cpu: 0.00 / 0.00 #8 : 368.009, ravg: 327.14, cpu: 0.00 / 0.00 #9 : 408.098, ravg: 335.24, cpu: 0.00 / 0.00 #10 : 368.582, ravg: 338.57, cpu: 0.00 / 0.00 I.e. we successfully analyzed the trace, replayed it via real threads and measured the replayed workload's scheduling properties. This is how it looked like in 'top' output: PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND 7164 mingo 20 0 1434m 8080 888 R 57.0 0.1 0:02.04 :perf 7165 mingo 20 0 1434m 8080 888 R 41.8 0.1 0:01.52 :perf 7228 mingo 20 0 1434m 8080 888 R 39.8 0.1 0:01.44 :gcc 7225 mingo 20 0 1434m 8080 888 R 33.8 0.1 0:01.26 :gcc 7202 mingo 20 0 1434m 8080 888 R 31.2 0.1 0:01.16 :sh 7222 mingo 20 0 1434m 8080 888 R 25.2 0.1 0:00.96 :sh 7211 mingo 20 0 1434m 8080 888 R 21.9 0.1 0:00.82 :sh 7213 mingo 20 0 1434m 8080 888 D 19.2 0.1 0:00.74 :sh 7194 mingo 20 0 1434m 8080 888 D 18.6 0.1 0:00.72 :make There's still various kinks in it - more patches to come. Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Mike Galbraith <efault@gmx.de> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> LKML-Reference: <new-submission> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-09-11 06:12:54 -04:00
cpu_last_switched[cpu] = timestamp;
add_sched_event_run(prev, timestamp, delta);
add_sched_event_sleep(prev, timestamp, switch_event->prev_state);
perf sched: Implement the scheduling workload replay engine Integrate the schedbench.c bits with the raw trace events that we get from the perf machinery, and activate the workload replayer/simulator. Example of a captured 'make -j' workload: $ perf sched run measurement overhead: 90 nsecs sleep measurement overhead: 2724743 nsecs the run test took 1000081 nsecs the sleep test took 2981111 nsecs version = 0.5 ... nr_run_events: 70 nr_sleep_events: 66 nr_wakeup_events: 9 target-less wakeups: 71 multi-target wakeups: 47 run events optimized: 139 task 0 ( perf: 6607), nr_events: 2 task 1 ( perf: 6608), nr_events: 6 task 2 ( : 0), nr_events: 1 task 3 ( make: 6609), nr_events: 5 task 4 ( sh: 6610), nr_events: 4 task 5 ( make: 6611), nr_events: 6 task 6 ( sh: 6612), nr_events: 4 task 7 ( make: 6613), nr_events: 5 task 8 ( migration/11: 25), nr_events: 1 task 9 ( migration/13: 29), nr_events: 1 task 10 ( migration/15: 33), nr_events: 1 task 11 ( migration/9: 21), nr_events: 1 task 12 ( sh: 6614), nr_events: 4 task 13 ( make: 6615), nr_events: 5 task 14 ( sh: 6616), nr_events: 4 task 15 ( make: 6617), nr_events: 7 task 16 ( migration/3: 9), nr_events: 1 task 17 ( migration/5: 13), nr_events: 1 task 18 ( migration/7: 17), nr_events: 1 task 19 ( migration/1: 5), nr_events: 1 task 20 ( sh: 6618), nr_events: 4 task 21 ( make: 6619), nr_events: 5 task 22 ( sh: 6620), nr_events: 4 task 23 ( make: 6621), nr_events: 10 task 24 ( sh: 6623), nr_events: 3 task 25 ( gcc: 6624), nr_events: 4 task 26 ( gcc: 6625), nr_events: 4 task 27 ( gcc: 6626), nr_events: 5 task 28 ( collect2: 6627), nr_events: 5 task 29 ( sh: 6622), nr_events: 1 task 30 ( make: 6628), nr_events: 7 task 31 ( sh: 6630), nr_events: 4 task 32 ( gcc: 6631), nr_events: 4 task 33 ( sh: 6629), nr_events: 1 task 34 ( gcc: 6632), nr_events: 4 task 35 ( gcc: 6633), nr_events: 4 task 36 ( collect2: 6634), nr_events: 4 task 37 ( make: 6635), nr_events: 8 task 38 ( sh: 6637), nr_events: 4 task 39 ( sh: 6636), nr_events: 1 task 40 ( gcc: 6638), nr_events: 4 task 41 ( gcc: 6639), nr_events: 4 task 42 ( gcc: 6640), nr_events: 4 task 43 ( collect2: 6641), nr_events: 4 task 44 ( make: 6642), nr_events: 6 task 45 ( sh: 6643), nr_events: 5 task 46 ( sh: 6644), nr_events: 3 task 47 ( sh: 6645), nr_events: 4 task 48 ( make: 6646), nr_events: 6 task 49 ( sh: 6647), nr_events: 3 task 50 ( make: 6648), nr_events: 5 task 51 ( sh: 6649), nr_events: 5 task 52 ( sh: 6650), nr_events: 6 task 53 ( make: 6651), nr_events: 4 task 54 ( make: 6652), nr_events: 5 task 55 ( make: 6653), nr_events: 4 task 56 ( make: 6654), nr_events: 4 task 57 ( make: 6655), nr_events: 5 task 58 ( sh: 6656), nr_events: 4 task 59 ( gcc: 6657), nr_events: 9 task 60 ( ksoftirqd/3: 10), nr_events: 1 task 61 ( gcc: 6658), nr_events: 4 task 62 ( make: 6659), nr_events: 5 task 63 ( sh: 6660), nr_events: 3 task 64 ( gcc: 6661), nr_events: 5 task 65 ( collect2: 6662), nr_events: 4 ------------------------------------------------------------ #1 : 256.745, ravg: 256.74, cpu: 0.00 / 0.00 #2 : 439.372, ravg: 275.01, cpu: 0.00 / 0.00 #3 : 411.971, ravg: 288.70, cpu: 0.00 / 0.00 #4 : 385.500, ravg: 298.38, cpu: 0.00 / 0.00 #5 : 366.526, ravg: 305.20, cpu: 0.00 / 0.00 #6 : 381.281, ravg: 312.81, cpu: 0.00 / 0.00 #7 : 410.756, ravg: 322.60, cpu: 0.00 / 0.00 #8 : 368.009, ravg: 327.14, cpu: 0.00 / 0.00 #9 : 408.098, ravg: 335.24, cpu: 0.00 / 0.00 #10 : 368.582, ravg: 338.57, cpu: 0.00 / 0.00 I.e. we successfully analyzed the trace, replayed it via real threads and measured the replayed workload's scheduling properties. This is how it looked like in 'top' output: PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND 7164 mingo 20 0 1434m 8080 888 R 57.0 0.1 0:02.04 :perf 7165 mingo 20 0 1434m 8080 888 R 41.8 0.1 0:01.52 :perf 7228 mingo 20 0 1434m 8080 888 R 39.8 0.1 0:01.44 :gcc 7225 mingo 20 0 1434m 8080 888 R 33.8 0.1 0:01.26 :gcc 7202 mingo 20 0 1434m 8080 888 R 31.2 0.1 0:01.16 :sh 7222 mingo 20 0 1434m 8080 888 R 25.2 0.1 0:00.96 :sh 7211 mingo 20 0 1434m 8080 888 R 21.9 0.1 0:00.82 :sh 7213 mingo 20 0 1434m 8080 888 D 19.2 0.1 0:00.74 :sh 7194 mingo 20 0 1434m 8080 888 D 18.6 0.1 0:00.72 :make There's still various kinks in it - more patches to come. Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Mike Galbraith <efault@gmx.de> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> LKML-Reference: <new-submission> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-09-11 06:12:54 -04:00
}
static void
replay_fork_event(struct trace_fork_event *fork_event,
struct event *event,
int cpu __used,
u64 timestamp __used,
struct thread *thread __used)
{
if (verbose) {
printf("sched_fork event %p\n", event);
printf("... parent: %s/%d\n", fork_event->parent_comm, fork_event->parent_pid);
printf("... child: %s/%d\n", fork_event->child_comm, fork_event->child_pid);
}
register_pid(fork_event->parent_pid, fork_event->parent_comm);
register_pid(fork_event->child_pid, fork_event->child_comm);
}
perf sched: Implement the scheduling workload replay engine Integrate the schedbench.c bits with the raw trace events that we get from the perf machinery, and activate the workload replayer/simulator. Example of a captured 'make -j' workload: $ perf sched run measurement overhead: 90 nsecs sleep measurement overhead: 2724743 nsecs the run test took 1000081 nsecs the sleep test took 2981111 nsecs version = 0.5 ... nr_run_events: 70 nr_sleep_events: 66 nr_wakeup_events: 9 target-less wakeups: 71 multi-target wakeups: 47 run events optimized: 139 task 0 ( perf: 6607), nr_events: 2 task 1 ( perf: 6608), nr_events: 6 task 2 ( : 0), nr_events: 1 task 3 ( make: 6609), nr_events: 5 task 4 ( sh: 6610), nr_events: 4 task 5 ( make: 6611), nr_events: 6 task 6 ( sh: 6612), nr_events: 4 task 7 ( make: 6613), nr_events: 5 task 8 ( migration/11: 25), nr_events: 1 task 9 ( migration/13: 29), nr_events: 1 task 10 ( migration/15: 33), nr_events: 1 task 11 ( migration/9: 21), nr_events: 1 task 12 ( sh: 6614), nr_events: 4 task 13 ( make: 6615), nr_events: 5 task 14 ( sh: 6616), nr_events: 4 task 15 ( make: 6617), nr_events: 7 task 16 ( migration/3: 9), nr_events: 1 task 17 ( migration/5: 13), nr_events: 1 task 18 ( migration/7: 17), nr_events: 1 task 19 ( migration/1: 5), nr_events: 1 task 20 ( sh: 6618), nr_events: 4 task 21 ( make: 6619), nr_events: 5 task 22 ( sh: 6620), nr_events: 4 task 23 ( make: 6621), nr_events: 10 task 24 ( sh: 6623), nr_events: 3 task 25 ( gcc: 6624), nr_events: 4 task 26 ( gcc: 6625), nr_events: 4 task 27 ( gcc: 6626), nr_events: 5 task 28 ( collect2: 6627), nr_events: 5 task 29 ( sh: 6622), nr_events: 1 task 30 ( make: 6628), nr_events: 7 task 31 ( sh: 6630), nr_events: 4 task 32 ( gcc: 6631), nr_events: 4 task 33 ( sh: 6629), nr_events: 1 task 34 ( gcc: 6632), nr_events: 4 task 35 ( gcc: 6633), nr_events: 4 task 36 ( collect2: 6634), nr_events: 4 task 37 ( make: 6635), nr_events: 8 task 38 ( sh: 6637), nr_events: 4 task 39 ( sh: 6636), nr_events: 1 task 40 ( gcc: 6638), nr_events: 4 task 41 ( gcc: 6639), nr_events: 4 task 42 ( gcc: 6640), nr_events: 4 task 43 ( collect2: 6641), nr_events: 4 task 44 ( make: 6642), nr_events: 6 task 45 ( sh: 6643), nr_events: 5 task 46 ( sh: 6644), nr_events: 3 task 47 ( sh: 6645), nr_events: 4 task 48 ( make: 6646), nr_events: 6 task 49 ( sh: 6647), nr_events: 3 task 50 ( make: 6648), nr_events: 5 task 51 ( sh: 6649), nr_events: 5 task 52 ( sh: 6650), nr_events: 6 task 53 ( make: 6651), nr_events: 4 task 54 ( make: 6652), nr_events: 5 task 55 ( make: 6653), nr_events: 4 task 56 ( make: 6654), nr_events: 4 task 57 ( make: 6655), nr_events: 5 task 58 ( sh: 6656), nr_events: 4 task 59 ( gcc: 6657), nr_events: 9 task 60 ( ksoftirqd/3: 10), nr_events: 1 task 61 ( gcc: 6658), nr_events: 4 task 62 ( make: 6659), nr_events: 5 task 63 ( sh: 6660), nr_events: 3 task 64 ( gcc: 6661), nr_events: 5 task 65 ( collect2: 6662), nr_events: 4 ------------------------------------------------------------ #1 : 256.745, ravg: 256.74, cpu: 0.00 / 0.00 #2 : 439.372, ravg: 275.01, cpu: 0.00 / 0.00 #3 : 411.971, ravg: 288.70, cpu: 0.00 / 0.00 #4 : 385.500, ravg: 298.38, cpu: 0.00 / 0.00 #5 : 366.526, ravg: 305.20, cpu: 0.00 / 0.00 #6 : 381.281, ravg: 312.81, cpu: 0.00 / 0.00 #7 : 410.756, ravg: 322.60, cpu: 0.00 / 0.00 #8 : 368.009, ravg: 327.14, cpu: 0.00 / 0.00 #9 : 408.098, ravg: 335.24, cpu: 0.00 / 0.00 #10 : 368.582, ravg: 338.57, cpu: 0.00 / 0.00 I.e. we successfully analyzed the trace, replayed it via real threads and measured the replayed workload's scheduling properties. This is how it looked like in 'top' output: PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND 7164 mingo 20 0 1434m 8080 888 R 57.0 0.1 0:02.04 :perf 7165 mingo 20 0 1434m 8080 888 R 41.8 0.1 0:01.52 :perf 7228 mingo 20 0 1434m 8080 888 R 39.8 0.1 0:01.44 :gcc 7225 mingo 20 0 1434m 8080 888 R 33.8 0.1 0:01.26 :gcc 7202 mingo 20 0 1434m 8080 888 R 31.2 0.1 0:01.16 :sh 7222 mingo 20 0 1434m 8080 888 R 25.2 0.1 0:00.96 :sh 7211 mingo 20 0 1434m 8080 888 R 21.9 0.1 0:00.82 :sh 7213 mingo 20 0 1434m 8080 888 D 19.2 0.1 0:00.74 :sh 7194 mingo 20 0 1434m 8080 888 D 18.6 0.1 0:00.72 :make There's still various kinks in it - more patches to come. Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Mike Galbraith <efault@gmx.de> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> LKML-Reference: <new-submission> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-09-11 06:12:54 -04:00
static struct trace_sched_handler replay_ops = {
.wakeup_event = replay_wakeup_event,
.switch_event = replay_switch_event,
.fork_event = replay_fork_event,
perf sched: Implement the scheduling workload replay engine Integrate the schedbench.c bits with the raw trace events that we get from the perf machinery, and activate the workload replayer/simulator. Example of a captured 'make -j' workload: $ perf sched run measurement overhead: 90 nsecs sleep measurement overhead: 2724743 nsecs the run test took 1000081 nsecs the sleep test took 2981111 nsecs version = 0.5 ... nr_run_events: 70 nr_sleep_events: 66 nr_wakeup_events: 9 target-less wakeups: 71 multi-target wakeups: 47 run events optimized: 139 task 0 ( perf: 6607), nr_events: 2 task 1 ( perf: 6608), nr_events: 6 task 2 ( : 0), nr_events: 1 task 3 ( make: 6609), nr_events: 5 task 4 ( sh: 6610), nr_events: 4 task 5 ( make: 6611), nr_events: 6 task 6 ( sh: 6612), nr_events: 4 task 7 ( make: 6613), nr_events: 5 task 8 ( migration/11: 25), nr_events: 1 task 9 ( migration/13: 29), nr_events: 1 task 10 ( migration/15: 33), nr_events: 1 task 11 ( migration/9: 21), nr_events: 1 task 12 ( sh: 6614), nr_events: 4 task 13 ( make: 6615), nr_events: 5 task 14 ( sh: 6616), nr_events: 4 task 15 ( make: 6617), nr_events: 7 task 16 ( migration/3: 9), nr_events: 1 task 17 ( migration/5: 13), nr_events: 1 task 18 ( migration/7: 17), nr_events: 1 task 19 ( migration/1: 5), nr_events: 1 task 20 ( sh: 6618), nr_events: 4 task 21 ( make: 6619), nr_events: 5 task 22 ( sh: 6620), nr_events: 4 task 23 ( make: 6621), nr_events: 10 task 24 ( sh: 6623), nr_events: 3 task 25 ( gcc: 6624), nr_events: 4 task 26 ( gcc: 6625), nr_events: 4 task 27 ( gcc: 6626), nr_events: 5 task 28 ( collect2: 6627), nr_events: 5 task 29 ( sh: 6622), nr_events: 1 task 30 ( make: 6628), nr_events: 7 task 31 ( sh: 6630), nr_events: 4 task 32 ( gcc: 6631), nr_events: 4 task 33 ( sh: 6629), nr_events: 1 task 34 ( gcc: 6632), nr_events: 4 task 35 ( gcc: 6633), nr_events: 4 task 36 ( collect2: 6634), nr_events: 4 task 37 ( make: 6635), nr_events: 8 task 38 ( sh: 6637), nr_events: 4 task 39 ( sh: 6636), nr_events: 1 task 40 ( gcc: 6638), nr_events: 4 task 41 ( gcc: 6639), nr_events: 4 task 42 ( gcc: 6640), nr_events: 4 task 43 ( collect2: 6641), nr_events: 4 task 44 ( make: 6642), nr_events: 6 task 45 ( sh: 6643), nr_events: 5 task 46 ( sh: 6644), nr_events: 3 task 47 ( sh: 6645), nr_events: 4 task 48 ( make: 6646), nr_events: 6 task 49 ( sh: 6647), nr_events: 3 task 50 ( make: 6648), nr_events: 5 task 51 ( sh: 6649), nr_events: 5 task 52 ( sh: 6650), nr_events: 6 task 53 ( make: 6651), nr_events: 4 task 54 ( make: 6652), nr_events: 5 task 55 ( make: 6653), nr_events: 4 task 56 ( make: 6654), nr_events: 4 task 57 ( make: 6655), nr_events: 5 task 58 ( sh: 6656), nr_events: 4 task 59 ( gcc: 6657), nr_events: 9 task 60 ( ksoftirqd/3: 10), nr_events: 1 task 61 ( gcc: 6658), nr_events: 4 task 62 ( make: 6659), nr_events: 5 task 63 ( sh: 6660), nr_events: 3 task 64 ( gcc: 6661), nr_events: 5 task 65 ( collect2: 6662), nr_events: 4 ------------------------------------------------------------ #1 : 256.745, ravg: 256.74, cpu: 0.00 / 0.00 #2 : 439.372, ravg: 275.01, cpu: 0.00 / 0.00 #3 : 411.971, ravg: 288.70, cpu: 0.00 / 0.00 #4 : 385.500, ravg: 298.38, cpu: 0.00 / 0.00 #5 : 366.526, ravg: 305.20, cpu: 0.00 / 0.00 #6 : 381.281, ravg: 312.81, cpu: 0.00 / 0.00 #7 : 410.756, ravg: 322.60, cpu: 0.00 / 0.00 #8 : 368.009, ravg: 327.14, cpu: 0.00 / 0.00 #9 : 408.098, ravg: 335.24, cpu: 0.00 / 0.00 #10 : 368.582, ravg: 338.57, cpu: 0.00 / 0.00 I.e. we successfully analyzed the trace, replayed it via real threads and measured the replayed workload's scheduling properties. This is how it looked like in 'top' output: PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND 7164 mingo 20 0 1434m 8080 888 R 57.0 0.1 0:02.04 :perf 7165 mingo 20 0 1434m 8080 888 R 41.8 0.1 0:01.52 :perf 7228 mingo 20 0 1434m 8080 888 R 39.8 0.1 0:01.44 :gcc 7225 mingo 20 0 1434m 8080 888 R 33.8 0.1 0:01.26 :gcc 7202 mingo 20 0 1434m 8080 888 R 31.2 0.1 0:01.16 :sh 7222 mingo 20 0 1434m 8080 888 R 25.2 0.1 0:00.96 :sh 7211 mingo 20 0 1434m 8080 888 R 21.9 0.1 0:00.82 :sh 7213 mingo 20 0 1434m 8080 888 D 19.2 0.1 0:00.74 :sh 7194 mingo 20 0 1434m 8080 888 D 18.6 0.1 0:00.72 :make There's still various kinks in it - more patches to come. Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Mike Galbraith <efault@gmx.de> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> LKML-Reference: <new-submission> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-09-11 06:12:54 -04:00
};
#define TASK_STATE_TO_CHAR_STR "RSDTtZX"
enum thread_state {
THREAD_SLEEPING,
THREAD_WAKED_UP,
THREAD_SCHED_IN,
THREAD_IGNORE
};
struct lat_snapshot {
struct list_head list;
enum thread_state state;
u64 wake_up_time;
u64 sched_in_time;
u64 runtime;
};
struct thread_latency {
struct list_head snapshot_list;
struct thread *thread;
struct rb_node node;
};
static struct rb_root lat_snapshot_root;
static struct thread_latency *
thread_latency_search(struct rb_root *root, struct thread *thread)
{
struct rb_node *node = root->rb_node;
while (node) {
struct thread_latency *lat;
lat = container_of(node, struct thread_latency, node);
if (thread->pid < lat->thread->pid)
node = node->rb_left;
else if (thread->pid > lat->thread->pid)
node = node->rb_right;
else {
return lat;
}
}
return NULL;
}
static void
__thread_latency_insert(struct rb_root *root, struct thread_latency *data)
{
struct rb_node **new = &(root->rb_node), *parent = NULL;
while (*new) {
struct thread_latency *this;
this = container_of(*new, struct thread_latency, node);
parent = *new;
if (data->thread->pid < this->thread->pid)
new = &((*new)->rb_left);
else if (data->thread->pid > this->thread->pid)
new = &((*new)->rb_right);
else
die("Double thread insertion\n");
}
rb_link_node(&data->node, parent, new);
rb_insert_color(&data->node, root);
}
static void thread_latency_insert(struct thread *thread)
{
struct thread_latency *lat;
lat = calloc(sizeof(*lat), 1);
if (!lat)
die("No memory");
lat->thread = thread;
INIT_LIST_HEAD(&lat->snapshot_list);
__thread_latency_insert(&lat_snapshot_root, lat);
}
static void
latency_fork_event(struct trace_fork_event *fork_event __used,
struct event *event __used,
int cpu __used,
u64 timestamp __used,
struct thread *thread __used)
{
/* should insert the newcomer */
}
__used
static char sched_out_state(struct trace_switch_event *switch_event)
{
const char *str = TASK_STATE_TO_CHAR_STR;
return str[switch_event->prev_state];
}
static void
lat_sched_out(struct thread_latency *lat,
struct trace_switch_event *switch_event __used, u64 delta)
{
struct lat_snapshot *snapshot;
snapshot = calloc(sizeof(*snapshot), 1);
if (!snapshot)
die("Non memory");
snapshot->runtime = delta;
list_add_tail(&snapshot->list, &lat->snapshot_list);
}
static void
lat_sched_in(struct thread_latency *lat, u64 timestamp)
{
struct lat_snapshot *snapshot;
if (list_empty(&lat->snapshot_list))
return;
snapshot = list_entry(lat->snapshot_list.prev, struct lat_snapshot,
list);
if (snapshot->state != THREAD_WAKED_UP)
return;
if (timestamp < snapshot->wake_up_time) {
snapshot->state = THREAD_IGNORE;
return;
}
snapshot->state = THREAD_SCHED_IN;
snapshot->sched_in_time = timestamp;
}
static void
latency_switch_event(struct trace_switch_event *switch_event,
struct event *event __used,
int cpu,
u64 timestamp,
struct thread *thread __used)
{
struct thread_latency *out_lat, *in_lat;
struct thread *sched_out, *sched_in;
u64 timestamp0;
s64 delta;
if (cpu >= MAX_CPUS || cpu < 0)
return;
timestamp0 = cpu_last_switched[cpu];
cpu_last_switched[cpu] = timestamp;
if (timestamp0)
delta = timestamp - timestamp0;
else
delta = 0;
if (delta < 0)
die("hm, delta: %Ld < 0 ?\n", delta);
sched_out = threads__findnew(switch_event->prev_pid, &threads, &last_match);
sched_in = threads__findnew(switch_event->next_pid, &threads, &last_match);
in_lat = thread_latency_search(&lat_snapshot_root, sched_in);
if (!in_lat) {
thread_latency_insert(sched_in);
in_lat = thread_latency_search(&lat_snapshot_root, sched_in);
if (!in_lat)
die("Internal latency tree error");
}
out_lat = thread_latency_search(&lat_snapshot_root, sched_out);
if (!out_lat) {
thread_latency_insert(sched_out);
out_lat = thread_latency_search(&lat_snapshot_root, sched_out);
if (!out_lat)
die("Internal latency tree error");
}
lat_sched_in(in_lat, timestamp);
lat_sched_out(out_lat, switch_event, delta);
}
static void
latency_wakeup_event(struct trace_wakeup_event *wakeup_event,
struct event *event __used,
int cpu __used,
u64 timestamp,
struct thread *thread __used)
{
struct thread_latency *lat;
struct lat_snapshot *snapshot;
struct thread *wakee;
/* Note for later, it may be interesting to observe the failing cases */
if (!wakeup_event->success)
return;
wakee = threads__findnew(wakeup_event->pid, &threads, &last_match);
lat = thread_latency_search(&lat_snapshot_root, wakee);
if (!lat) {
thread_latency_insert(wakee);
return;
}
if (list_empty(&lat->snapshot_list))
return;
snapshot = list_entry(lat->snapshot_list.prev, struct lat_snapshot,
list);
if (snapshot->state != THREAD_SLEEPING)
return;
snapshot->state = THREAD_WAKED_UP;
snapshot->wake_up_time = timestamp;
}
static struct trace_sched_handler lat_ops = {
.wakeup_event = latency_wakeup_event,
.switch_event = latency_switch_event,
.fork_event = latency_fork_event,
};
static u64 all_runtime;
static u64 all_count;
static void output_lat_thread(struct thread_latency *lat)
{
struct lat_snapshot *shot;
int count = 0;
int i;
int ret;
u64 max = 0, avg;
u64 total = 0, delta;
u64 total_runtime = 0;
list_for_each_entry(shot, &lat->snapshot_list, list) {
total_runtime += shot->runtime;
if (shot->state != THREAD_SCHED_IN)
continue;
count++;
delta = shot->sched_in_time - shot->wake_up_time;
if (delta > max)
max = delta;
total += delta;
}
all_runtime += total_runtime;
all_count += count;
if (!count)
return;
perf sched: Display time in milliseconds, reorganize output After: ----------------------------------------------------------------------------------- Task | runtime ms | switches | average delay ms | maximum delay ms | ----------------------------------------------------------------------------------- migration/0 | 0.000 ms | 1 | avg: 0.047 ms | max: 0.047 ms | ksoftirqd/0 | 0.000 ms | 1 | avg: 0.039 ms | max: 0.039 ms | migration/1 | 0.000 ms | 3 | avg: 0.013 ms | max: 0.016 ms | migration/3 | 0.000 ms | 2 | avg: 0.003 ms | max: 0.004 ms | migration/4 | 0.000 ms | 1 | avg: 0.022 ms | max: 0.022 ms | distccd | 0.000 ms | 1 | avg: 0.004 ms | max: 0.004 ms | distccd | 0.000 ms | 1 | avg: 0.014 ms | max: 0.014 ms | distccd | 0.000 ms | 2 | avg: 0.000 ms | max: 0.000 ms | distccd | 0.000 ms | 2 | avg: 0.012 ms | max: 0.019 ms | distccd | 0.000 ms | 1 | avg: 0.002 ms | max: 0.002 ms | as | 0.000 ms | 2 | avg: 0.019 ms | max: 0.019 ms | as | 0.000 ms | 3 | avg: 0.015 ms | max: 0.017 ms | as | 0.000 ms | 1 | avg: 0.009 ms | max: 0.009 ms | perf | 0.000 ms | 1 | avg: 0.001 ms | max: 0.001 ms | gcc | 0.000 ms | 1 | avg: 0.021 ms | max: 0.021 ms | run-mozilla.sh | 0.000 ms | 2 | avg: 0.010 ms | max: 0.017 ms | mozilla-plugin- | 0.000 ms | 1 | avg: 0.006 ms | max: 0.006 ms | gcc | 0.000 ms | 2 | avg: 0.013 ms | max: 0.013 ms | ----------------------------------------------------------------------------------- (The runtime ms column is not filled in yet.) Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Mike Galbraith <efault@gmx.de> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> LKML-Reference: <new-submission> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-09-12 04:08:34 -04:00
ret = printf(" %s ", lat->thread->comm);
perf sched: Display time in milliseconds, reorganize output After: ----------------------------------------------------------------------------------- Task | runtime ms | switches | average delay ms | maximum delay ms | ----------------------------------------------------------------------------------- migration/0 | 0.000 ms | 1 | avg: 0.047 ms | max: 0.047 ms | ksoftirqd/0 | 0.000 ms | 1 | avg: 0.039 ms | max: 0.039 ms | migration/1 | 0.000 ms | 3 | avg: 0.013 ms | max: 0.016 ms | migration/3 | 0.000 ms | 2 | avg: 0.003 ms | max: 0.004 ms | migration/4 | 0.000 ms | 1 | avg: 0.022 ms | max: 0.022 ms | distccd | 0.000 ms | 1 | avg: 0.004 ms | max: 0.004 ms | distccd | 0.000 ms | 1 | avg: 0.014 ms | max: 0.014 ms | distccd | 0.000 ms | 2 | avg: 0.000 ms | max: 0.000 ms | distccd | 0.000 ms | 2 | avg: 0.012 ms | max: 0.019 ms | distccd | 0.000 ms | 1 | avg: 0.002 ms | max: 0.002 ms | as | 0.000 ms | 2 | avg: 0.019 ms | max: 0.019 ms | as | 0.000 ms | 3 | avg: 0.015 ms | max: 0.017 ms | as | 0.000 ms | 1 | avg: 0.009 ms | max: 0.009 ms | perf | 0.000 ms | 1 | avg: 0.001 ms | max: 0.001 ms | gcc | 0.000 ms | 1 | avg: 0.021 ms | max: 0.021 ms | run-mozilla.sh | 0.000 ms | 2 | avg: 0.010 ms | max: 0.017 ms | mozilla-plugin- | 0.000 ms | 1 | avg: 0.006 ms | max: 0.006 ms | gcc | 0.000 ms | 2 | avg: 0.013 ms | max: 0.013 ms | ----------------------------------------------------------------------------------- (The runtime ms column is not filled in yet.) Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Mike Galbraith <efault@gmx.de> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> LKML-Reference: <new-submission> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-09-12 04:08:34 -04:00
for (i = 0; i < 19 - ret; i++)
printf(" ");
avg = total / count;
perf sched: Display time in milliseconds, reorganize output After: ----------------------------------------------------------------------------------- Task | runtime ms | switches | average delay ms | maximum delay ms | ----------------------------------------------------------------------------------- migration/0 | 0.000 ms | 1 | avg: 0.047 ms | max: 0.047 ms | ksoftirqd/0 | 0.000 ms | 1 | avg: 0.039 ms | max: 0.039 ms | migration/1 | 0.000 ms | 3 | avg: 0.013 ms | max: 0.016 ms | migration/3 | 0.000 ms | 2 | avg: 0.003 ms | max: 0.004 ms | migration/4 | 0.000 ms | 1 | avg: 0.022 ms | max: 0.022 ms | distccd | 0.000 ms | 1 | avg: 0.004 ms | max: 0.004 ms | distccd | 0.000 ms | 1 | avg: 0.014 ms | max: 0.014 ms | distccd | 0.000 ms | 2 | avg: 0.000 ms | max: 0.000 ms | distccd | 0.000 ms | 2 | avg: 0.012 ms | max: 0.019 ms | distccd | 0.000 ms | 1 | avg: 0.002 ms | max: 0.002 ms | as | 0.000 ms | 2 | avg: 0.019 ms | max: 0.019 ms | as | 0.000 ms | 3 | avg: 0.015 ms | max: 0.017 ms | as | 0.000 ms | 1 | avg: 0.009 ms | max: 0.009 ms | perf | 0.000 ms | 1 | avg: 0.001 ms | max: 0.001 ms | gcc | 0.000 ms | 1 | avg: 0.021 ms | max: 0.021 ms | run-mozilla.sh | 0.000 ms | 2 | avg: 0.010 ms | max: 0.017 ms | mozilla-plugin- | 0.000 ms | 1 | avg: 0.006 ms | max: 0.006 ms | gcc | 0.000 ms | 2 | avg: 0.013 ms | max: 0.013 ms | ----------------------------------------------------------------------------------- (The runtime ms column is not filled in yet.) Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Mike Galbraith <efault@gmx.de> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> LKML-Reference: <new-submission> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-09-12 04:08:34 -04:00
printf("|%9.3f ms |%9d | avg:%9.3f ms | max:%9.3f ms |\n",
(double)total_runtime/1e9, count, (double)avg/1e9, (double)max/1e9);
}
static void __cmd_lat(void)
{
struct rb_node *next;
setup_pager();
read_events();
perf sched: Display time in milliseconds, reorganize output After: ----------------------------------------------------------------------------------- Task | runtime ms | switches | average delay ms | maximum delay ms | ----------------------------------------------------------------------------------- migration/0 | 0.000 ms | 1 | avg: 0.047 ms | max: 0.047 ms | ksoftirqd/0 | 0.000 ms | 1 | avg: 0.039 ms | max: 0.039 ms | migration/1 | 0.000 ms | 3 | avg: 0.013 ms | max: 0.016 ms | migration/3 | 0.000 ms | 2 | avg: 0.003 ms | max: 0.004 ms | migration/4 | 0.000 ms | 1 | avg: 0.022 ms | max: 0.022 ms | distccd | 0.000 ms | 1 | avg: 0.004 ms | max: 0.004 ms | distccd | 0.000 ms | 1 | avg: 0.014 ms | max: 0.014 ms | distccd | 0.000 ms | 2 | avg: 0.000 ms | max: 0.000 ms | distccd | 0.000 ms | 2 | avg: 0.012 ms | max: 0.019 ms | distccd | 0.000 ms | 1 | avg: 0.002 ms | max: 0.002 ms | as | 0.000 ms | 2 | avg: 0.019 ms | max: 0.019 ms | as | 0.000 ms | 3 | avg: 0.015 ms | max: 0.017 ms | as | 0.000 ms | 1 | avg: 0.009 ms | max: 0.009 ms | perf | 0.000 ms | 1 | avg: 0.001 ms | max: 0.001 ms | gcc | 0.000 ms | 1 | avg: 0.021 ms | max: 0.021 ms | run-mozilla.sh | 0.000 ms | 2 | avg: 0.010 ms | max: 0.017 ms | mozilla-plugin- | 0.000 ms | 1 | avg: 0.006 ms | max: 0.006 ms | gcc | 0.000 ms | 2 | avg: 0.013 ms | max: 0.013 ms | ----------------------------------------------------------------------------------- (The runtime ms column is not filled in yet.) Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Mike Galbraith <efault@gmx.de> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> LKML-Reference: <new-submission> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-09-12 04:08:34 -04:00
printf("-----------------------------------------------------------------------------------\n");
printf(" Task | Runtime ms | Switches | Average delay ms | Maximum delay ms |\n");
perf sched: Display time in milliseconds, reorganize output After: ----------------------------------------------------------------------------------- Task | runtime ms | switches | average delay ms | maximum delay ms | ----------------------------------------------------------------------------------- migration/0 | 0.000 ms | 1 | avg: 0.047 ms | max: 0.047 ms | ksoftirqd/0 | 0.000 ms | 1 | avg: 0.039 ms | max: 0.039 ms | migration/1 | 0.000 ms | 3 | avg: 0.013 ms | max: 0.016 ms | migration/3 | 0.000 ms | 2 | avg: 0.003 ms | max: 0.004 ms | migration/4 | 0.000 ms | 1 | avg: 0.022 ms | max: 0.022 ms | distccd | 0.000 ms | 1 | avg: 0.004 ms | max: 0.004 ms | distccd | 0.000 ms | 1 | avg: 0.014 ms | max: 0.014 ms | distccd | 0.000 ms | 2 | avg: 0.000 ms | max: 0.000 ms | distccd | 0.000 ms | 2 | avg: 0.012 ms | max: 0.019 ms | distccd | 0.000 ms | 1 | avg: 0.002 ms | max: 0.002 ms | as | 0.000 ms | 2 | avg: 0.019 ms | max: 0.019 ms | as | 0.000 ms | 3 | avg: 0.015 ms | max: 0.017 ms | as | 0.000 ms | 1 | avg: 0.009 ms | max: 0.009 ms | perf | 0.000 ms | 1 | avg: 0.001 ms | max: 0.001 ms | gcc | 0.000 ms | 1 | avg: 0.021 ms | max: 0.021 ms | run-mozilla.sh | 0.000 ms | 2 | avg: 0.010 ms | max: 0.017 ms | mozilla-plugin- | 0.000 ms | 1 | avg: 0.006 ms | max: 0.006 ms | gcc | 0.000 ms | 2 | avg: 0.013 ms | max: 0.013 ms | ----------------------------------------------------------------------------------- (The runtime ms column is not filled in yet.) Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Mike Galbraith <efault@gmx.de> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> LKML-Reference: <new-submission> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-09-12 04:08:34 -04:00
printf("-----------------------------------------------------------------------------------\n");
next = rb_first(&lat_snapshot_root);
while (next) {
struct thread_latency *lat;
lat = rb_entry(next, struct thread_latency, node);
output_lat_thread(lat);
next = rb_next(next);
}
perf sched: Display time in milliseconds, reorganize output After: ----------------------------------------------------------------------------------- Task | runtime ms | switches | average delay ms | maximum delay ms | ----------------------------------------------------------------------------------- migration/0 | 0.000 ms | 1 | avg: 0.047 ms | max: 0.047 ms | ksoftirqd/0 | 0.000 ms | 1 | avg: 0.039 ms | max: 0.039 ms | migration/1 | 0.000 ms | 3 | avg: 0.013 ms | max: 0.016 ms | migration/3 | 0.000 ms | 2 | avg: 0.003 ms | max: 0.004 ms | migration/4 | 0.000 ms | 1 | avg: 0.022 ms | max: 0.022 ms | distccd | 0.000 ms | 1 | avg: 0.004 ms | max: 0.004 ms | distccd | 0.000 ms | 1 | avg: 0.014 ms | max: 0.014 ms | distccd | 0.000 ms | 2 | avg: 0.000 ms | max: 0.000 ms | distccd | 0.000 ms | 2 | avg: 0.012 ms | max: 0.019 ms | distccd | 0.000 ms | 1 | avg: 0.002 ms | max: 0.002 ms | as | 0.000 ms | 2 | avg: 0.019 ms | max: 0.019 ms | as | 0.000 ms | 3 | avg: 0.015 ms | max: 0.017 ms | as | 0.000 ms | 1 | avg: 0.009 ms | max: 0.009 ms | perf | 0.000 ms | 1 | avg: 0.001 ms | max: 0.001 ms | gcc | 0.000 ms | 1 | avg: 0.021 ms | max: 0.021 ms | run-mozilla.sh | 0.000 ms | 2 | avg: 0.010 ms | max: 0.017 ms | mozilla-plugin- | 0.000 ms | 1 | avg: 0.006 ms | max: 0.006 ms | gcc | 0.000 ms | 2 | avg: 0.013 ms | max: 0.013 ms | ----------------------------------------------------------------------------------- (The runtime ms column is not filled in yet.) Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Mike Galbraith <efault@gmx.de> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> LKML-Reference: <new-submission> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-09-12 04:08:34 -04:00
printf("-----------------------------------------------------------------------------------\n");
printf(" TOTAL: |%9.3f ms |%9Ld |\n",
(double)all_runtime/1e9, all_count);
printf("---------------------------------------------\n");
}
static struct trace_sched_handler *trace_handler;
perf sched: Implement the scheduling workload replay engine Integrate the schedbench.c bits with the raw trace events that we get from the perf machinery, and activate the workload replayer/simulator. Example of a captured 'make -j' workload: $ perf sched run measurement overhead: 90 nsecs sleep measurement overhead: 2724743 nsecs the run test took 1000081 nsecs the sleep test took 2981111 nsecs version = 0.5 ... nr_run_events: 70 nr_sleep_events: 66 nr_wakeup_events: 9 target-less wakeups: 71 multi-target wakeups: 47 run events optimized: 139 task 0 ( perf: 6607), nr_events: 2 task 1 ( perf: 6608), nr_events: 6 task 2 ( : 0), nr_events: 1 task 3 ( make: 6609), nr_events: 5 task 4 ( sh: 6610), nr_events: 4 task 5 ( make: 6611), nr_events: 6 task 6 ( sh: 6612), nr_events: 4 task 7 ( make: 6613), nr_events: 5 task 8 ( migration/11: 25), nr_events: 1 task 9 ( migration/13: 29), nr_events: 1 task 10 ( migration/15: 33), nr_events: 1 task 11 ( migration/9: 21), nr_events: 1 task 12 ( sh: 6614), nr_events: 4 task 13 ( make: 6615), nr_events: 5 task 14 ( sh: 6616), nr_events: 4 task 15 ( make: 6617), nr_events: 7 task 16 ( migration/3: 9), nr_events: 1 task 17 ( migration/5: 13), nr_events: 1 task 18 ( migration/7: 17), nr_events: 1 task 19 ( migration/1: 5), nr_events: 1 task 20 ( sh: 6618), nr_events: 4 task 21 ( make: 6619), nr_events: 5 task 22 ( sh: 6620), nr_events: 4 task 23 ( make: 6621), nr_events: 10 task 24 ( sh: 6623), nr_events: 3 task 25 ( gcc: 6624), nr_events: 4 task 26 ( gcc: 6625), nr_events: 4 task 27 ( gcc: 6626), nr_events: 5 task 28 ( collect2: 6627), nr_events: 5 task 29 ( sh: 6622), nr_events: 1 task 30 ( make: 6628), nr_events: 7 task 31 ( sh: 6630), nr_events: 4 task 32 ( gcc: 6631), nr_events: 4 task 33 ( sh: 6629), nr_events: 1 task 34 ( gcc: 6632), nr_events: 4 task 35 ( gcc: 6633), nr_events: 4 task 36 ( collect2: 6634), nr_events: 4 task 37 ( make: 6635), nr_events: 8 task 38 ( sh: 6637), nr_events: 4 task 39 ( sh: 6636), nr_events: 1 task 40 ( gcc: 6638), nr_events: 4 task 41 ( gcc: 6639), nr_events: 4 task 42 ( gcc: 6640), nr_events: 4 task 43 ( collect2: 6641), nr_events: 4 task 44 ( make: 6642), nr_events: 6 task 45 ( sh: 6643), nr_events: 5 task 46 ( sh: 6644), nr_events: 3 task 47 ( sh: 6645), nr_events: 4 task 48 ( make: 6646), nr_events: 6 task 49 ( sh: 6647), nr_events: 3 task 50 ( make: 6648), nr_events: 5 task 51 ( sh: 6649), nr_events: 5 task 52 ( sh: 6650), nr_events: 6 task 53 ( make: 6651), nr_events: 4 task 54 ( make: 6652), nr_events: 5 task 55 ( make: 6653), nr_events: 4 task 56 ( make: 6654), nr_events: 4 task 57 ( make: 6655), nr_events: 5 task 58 ( sh: 6656), nr_events: 4 task 59 ( gcc: 6657), nr_events: 9 task 60 ( ksoftirqd/3: 10), nr_events: 1 task 61 ( gcc: 6658), nr_events: 4 task 62 ( make: 6659), nr_events: 5 task 63 ( sh: 6660), nr_events: 3 task 64 ( gcc: 6661), nr_events: 5 task 65 ( collect2: 6662), nr_events: 4 ------------------------------------------------------------ #1 : 256.745, ravg: 256.74, cpu: 0.00 / 0.00 #2 : 439.372, ravg: 275.01, cpu: 0.00 / 0.00 #3 : 411.971, ravg: 288.70, cpu: 0.00 / 0.00 #4 : 385.500, ravg: 298.38, cpu: 0.00 / 0.00 #5 : 366.526, ravg: 305.20, cpu: 0.00 / 0.00 #6 : 381.281, ravg: 312.81, cpu: 0.00 / 0.00 #7 : 410.756, ravg: 322.60, cpu: 0.00 / 0.00 #8 : 368.009, ravg: 327.14, cpu: 0.00 / 0.00 #9 : 408.098, ravg: 335.24, cpu: 0.00 / 0.00 #10 : 368.582, ravg: 338.57, cpu: 0.00 / 0.00 I.e. we successfully analyzed the trace, replayed it via real threads and measured the replayed workload's scheduling properties. This is how it looked like in 'top' output: PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND 7164 mingo 20 0 1434m 8080 888 R 57.0 0.1 0:02.04 :perf 7165 mingo 20 0 1434m 8080 888 R 41.8 0.1 0:01.52 :perf 7228 mingo 20 0 1434m 8080 888 R 39.8 0.1 0:01.44 :gcc 7225 mingo 20 0 1434m 8080 888 R 33.8 0.1 0:01.26 :gcc 7202 mingo 20 0 1434m 8080 888 R 31.2 0.1 0:01.16 :sh 7222 mingo 20 0 1434m 8080 888 R 25.2 0.1 0:00.96 :sh 7211 mingo 20 0 1434m 8080 888 R 21.9 0.1 0:00.82 :sh 7213 mingo 20 0 1434m 8080 888 D 19.2 0.1 0:00.74 :sh 7194 mingo 20 0 1434m 8080 888 D 18.6 0.1 0:00.72 :make There's still various kinks in it - more patches to come. Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Mike Galbraith <efault@gmx.de> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> LKML-Reference: <new-submission> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-09-11 06:12:54 -04:00
static void
process_sched_wakeup_event(struct raw_event_sample *raw,
struct event *event,
int cpu __used,
u64 timestamp __used,
struct thread *thread __used)
{
struct trace_wakeup_event wakeup_event;
FILL_COMMON_FIELDS(wakeup_event, event, raw->data);
FILL_ARRAY(wakeup_event, comm, event, raw->data);
FILL_FIELD(wakeup_event, pid, event, raw->data);
FILL_FIELD(wakeup_event, prio, event, raw->data);
FILL_FIELD(wakeup_event, success, event, raw->data);
FILL_FIELD(wakeup_event, cpu, event, raw->data);
trace_handler->wakeup_event(&wakeup_event, event, cpu, timestamp, thread);
}
static void
process_sched_switch_event(struct raw_event_sample *raw,
struct event *event,
int cpu __used,
u64 timestamp __used,
struct thread *thread __used)
{
struct trace_switch_event switch_event;
FILL_COMMON_FIELDS(switch_event, event, raw->data);
FILL_ARRAY(switch_event, prev_comm, event, raw->data);
FILL_FIELD(switch_event, prev_pid, event, raw->data);
FILL_FIELD(switch_event, prev_prio, event, raw->data);
FILL_FIELD(switch_event, prev_state, event, raw->data);
FILL_ARRAY(switch_event, next_comm, event, raw->data);
FILL_FIELD(switch_event, next_pid, event, raw->data);
FILL_FIELD(switch_event, next_prio, event, raw->data);
trace_handler->switch_event(&switch_event, event, cpu, timestamp, thread);
}
static void
process_sched_fork_event(struct raw_event_sample *raw,
struct event *event,
int cpu __used,
u64 timestamp __used,
struct thread *thread __used)
perf sched: Implement the scheduling workload replay engine Integrate the schedbench.c bits with the raw trace events that we get from the perf machinery, and activate the workload replayer/simulator. Example of a captured 'make -j' workload: $ perf sched run measurement overhead: 90 nsecs sleep measurement overhead: 2724743 nsecs the run test took 1000081 nsecs the sleep test took 2981111 nsecs version = 0.5 ... nr_run_events: 70 nr_sleep_events: 66 nr_wakeup_events: 9 target-less wakeups: 71 multi-target wakeups: 47 run events optimized: 139 task 0 ( perf: 6607), nr_events: 2 task 1 ( perf: 6608), nr_events: 6 task 2 ( : 0), nr_events: 1 task 3 ( make: 6609), nr_events: 5 task 4 ( sh: 6610), nr_events: 4 task 5 ( make: 6611), nr_events: 6 task 6 ( sh: 6612), nr_events: 4 task 7 ( make: 6613), nr_events: 5 task 8 ( migration/11: 25), nr_events: 1 task 9 ( migration/13: 29), nr_events: 1 task 10 ( migration/15: 33), nr_events: 1 task 11 ( migration/9: 21), nr_events: 1 task 12 ( sh: 6614), nr_events: 4 task 13 ( make: 6615), nr_events: 5 task 14 ( sh: 6616), nr_events: 4 task 15 ( make: 6617), nr_events: 7 task 16 ( migration/3: 9), nr_events: 1 task 17 ( migration/5: 13), nr_events: 1 task 18 ( migration/7: 17), nr_events: 1 task 19 ( migration/1: 5), nr_events: 1 task 20 ( sh: 6618), nr_events: 4 task 21 ( make: 6619), nr_events: 5 task 22 ( sh: 6620), nr_events: 4 task 23 ( make: 6621), nr_events: 10 task 24 ( sh: 6623), nr_events: 3 task 25 ( gcc: 6624), nr_events: 4 task 26 ( gcc: 6625), nr_events: 4 task 27 ( gcc: 6626), nr_events: 5 task 28 ( collect2: 6627), nr_events: 5 task 29 ( sh: 6622), nr_events: 1 task 30 ( make: 6628), nr_events: 7 task 31 ( sh: 6630), nr_events: 4 task 32 ( gcc: 6631), nr_events: 4 task 33 ( sh: 6629), nr_events: 1 task 34 ( gcc: 6632), nr_events: 4 task 35 ( gcc: 6633), nr_events: 4 task 36 ( collect2: 6634), nr_events: 4 task 37 ( make: 6635), nr_events: 8 task 38 ( sh: 6637), nr_events: 4 task 39 ( sh: 6636), nr_events: 1 task 40 ( gcc: 6638), nr_events: 4 task 41 ( gcc: 6639), nr_events: 4 task 42 ( gcc: 6640), nr_events: 4 task 43 ( collect2: 6641), nr_events: 4 task 44 ( make: 6642), nr_events: 6 task 45 ( sh: 6643), nr_events: 5 task 46 ( sh: 6644), nr_events: 3 task 47 ( sh: 6645), nr_events: 4 task 48 ( make: 6646), nr_events: 6 task 49 ( sh: 6647), nr_events: 3 task 50 ( make: 6648), nr_events: 5 task 51 ( sh: 6649), nr_events: 5 task 52 ( sh: 6650), nr_events: 6 task 53 ( make: 6651), nr_events: 4 task 54 ( make: 6652), nr_events: 5 task 55 ( make: 6653), nr_events: 4 task 56 ( make: 6654), nr_events: 4 task 57 ( make: 6655), nr_events: 5 task 58 ( sh: 6656), nr_events: 4 task 59 ( gcc: 6657), nr_events: 9 task 60 ( ksoftirqd/3: 10), nr_events: 1 task 61 ( gcc: 6658), nr_events: 4 task 62 ( make: 6659), nr_events: 5 task 63 ( sh: 6660), nr_events: 3 task 64 ( gcc: 6661), nr_events: 5 task 65 ( collect2: 6662), nr_events: 4 ------------------------------------------------------------ #1 : 256.745, ravg: 256.74, cpu: 0.00 / 0.00 #2 : 439.372, ravg: 275.01, cpu: 0.00 / 0.00 #3 : 411.971, ravg: 288.70, cpu: 0.00 / 0.00 #4 : 385.500, ravg: 298.38, cpu: 0.00 / 0.00 #5 : 366.526, ravg: 305.20, cpu: 0.00 / 0.00 #6 : 381.281, ravg: 312.81, cpu: 0.00 / 0.00 #7 : 410.756, ravg: 322.60, cpu: 0.00 / 0.00 #8 : 368.009, ravg: 327.14, cpu: 0.00 / 0.00 #9 : 408.098, ravg: 335.24, cpu: 0.00 / 0.00 #10 : 368.582, ravg: 338.57, cpu: 0.00 / 0.00 I.e. we successfully analyzed the trace, replayed it via real threads and measured the replayed workload's scheduling properties. This is how it looked like in 'top' output: PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND 7164 mingo 20 0 1434m 8080 888 R 57.0 0.1 0:02.04 :perf 7165 mingo 20 0 1434m 8080 888 R 41.8 0.1 0:01.52 :perf 7228 mingo 20 0 1434m 8080 888 R 39.8 0.1 0:01.44 :gcc 7225 mingo 20 0 1434m 8080 888 R 33.8 0.1 0:01.26 :gcc 7202 mingo 20 0 1434m 8080 888 R 31.2 0.1 0:01.16 :sh 7222 mingo 20 0 1434m 8080 888 R 25.2 0.1 0:00.96 :sh 7211 mingo 20 0 1434m 8080 888 R 21.9 0.1 0:00.82 :sh 7213 mingo 20 0 1434m 8080 888 D 19.2 0.1 0:00.74 :sh 7194 mingo 20 0 1434m 8080 888 D 18.6 0.1 0:00.72 :make There's still various kinks in it - more patches to come. Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Mike Galbraith <efault@gmx.de> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> LKML-Reference: <new-submission> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-09-11 06:12:54 -04:00
{
perf sched: Fix bad event alignment perf sched raises the following error when it meets a sched switch event: perf: builtin-sched.c:286: register_pid: Assertion `!(pid >= 65536)' failed. Abandon Currently in x86-64, the sched switch events have a hole in the middle of the structure: u16 common_type; u8 common_flags; u8 common_preempt_count; u32 common_pid; u32 common_tgid; char prev_comm[16]; u32 prev_pid; u32 prev_prio; <--- there u64 prev_state; char next_comm[16]; u32 next_pid; u32 next_prio; Gcc inserts a 4 bytes hole there for prev_state to be u64 aligned. And the events are exported to userspace with this hole. But in userspace, from perf sched, we fetch it using a structure that has a new field in the beginning: u32 size. This is because our trace is exported with its size as a field. But now that we have this new field, the hole in the middle disappears because it makes prev_state becoming well aligned. And since we are using a pointer to the raw trace using this struct, instead of reading prev_state, we are reading the hole. We could fix it by keeping the size seperate from the struct but actually there a lot of other potential problems: some fields may be saved as long in a 64 bits system and later read as long in a 32 bits system. Also this direct cast doesn't care about the endianness differences between the host traced machine and the machine in which we do the post processing. So instead of using such dangerous direct casts, fetch the values using the trace parsing API that already takes care of all these problems. Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Mike Galbraith <efault@gmx.de> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-09-11 20:43:45 -04:00
struct trace_fork_event fork_event;
FILL_COMMON_FIELDS(fork_event, event, raw->data);
FILL_ARRAY(fork_event, parent_comm, event, raw->data);
FILL_FIELD(fork_event, parent_pid, event, raw->data);
FILL_ARRAY(fork_event, child_comm, event, raw->data);
FILL_FIELD(fork_event, child_pid, event, raw->data);
trace_handler->fork_event(&fork_event, event, cpu, timestamp, thread);
perf sched: Implement the scheduling workload replay engine Integrate the schedbench.c bits with the raw trace events that we get from the perf machinery, and activate the workload replayer/simulator. Example of a captured 'make -j' workload: $ perf sched run measurement overhead: 90 nsecs sleep measurement overhead: 2724743 nsecs the run test took 1000081 nsecs the sleep test took 2981111 nsecs version = 0.5 ... nr_run_events: 70 nr_sleep_events: 66 nr_wakeup_events: 9 target-less wakeups: 71 multi-target wakeups: 47 run events optimized: 139 task 0 ( perf: 6607), nr_events: 2 task 1 ( perf: 6608), nr_events: 6 task 2 ( : 0), nr_events: 1 task 3 ( make: 6609), nr_events: 5 task 4 ( sh: 6610), nr_events: 4 task 5 ( make: 6611), nr_events: 6 task 6 ( sh: 6612), nr_events: 4 task 7 ( make: 6613), nr_events: 5 task 8 ( migration/11: 25), nr_events: 1 task 9 ( migration/13: 29), nr_events: 1 task 10 ( migration/15: 33), nr_events: 1 task 11 ( migration/9: 21), nr_events: 1 task 12 ( sh: 6614), nr_events: 4 task 13 ( make: 6615), nr_events: 5 task 14 ( sh: 6616), nr_events: 4 task 15 ( make: 6617), nr_events: 7 task 16 ( migration/3: 9), nr_events: 1 task 17 ( migration/5: 13), nr_events: 1 task 18 ( migration/7: 17), nr_events: 1 task 19 ( migration/1: 5), nr_events: 1 task 20 ( sh: 6618), nr_events: 4 task 21 ( make: 6619), nr_events: 5 task 22 ( sh: 6620), nr_events: 4 task 23 ( make: 6621), nr_events: 10 task 24 ( sh: 6623), nr_events: 3 task 25 ( gcc: 6624), nr_events: 4 task 26 ( gcc: 6625), nr_events: 4 task 27 ( gcc: 6626), nr_events: 5 task 28 ( collect2: 6627), nr_events: 5 task 29 ( sh: 6622), nr_events: 1 task 30 ( make: 6628), nr_events: 7 task 31 ( sh: 6630), nr_events: 4 task 32 ( gcc: 6631), nr_events: 4 task 33 ( sh: 6629), nr_events: 1 task 34 ( gcc: 6632), nr_events: 4 task 35 ( gcc: 6633), nr_events: 4 task 36 ( collect2: 6634), nr_events: 4 task 37 ( make: 6635), nr_events: 8 task 38 ( sh: 6637), nr_events: 4 task 39 ( sh: 6636), nr_events: 1 task 40 ( gcc: 6638), nr_events: 4 task 41 ( gcc: 6639), nr_events: 4 task 42 ( gcc: 6640), nr_events: 4 task 43 ( collect2: 6641), nr_events: 4 task 44 ( make: 6642), nr_events: 6 task 45 ( sh: 6643), nr_events: 5 task 46 ( sh: 6644), nr_events: 3 task 47 ( sh: 6645), nr_events: 4 task 48 ( make: 6646), nr_events: 6 task 49 ( sh: 6647), nr_events: 3 task 50 ( make: 6648), nr_events: 5 task 51 ( sh: 6649), nr_events: 5 task 52 ( sh: 6650), nr_events: 6 task 53 ( make: 6651), nr_events: 4 task 54 ( make: 6652), nr_events: 5 task 55 ( make: 6653), nr_events: 4 task 56 ( make: 6654), nr_events: 4 task 57 ( make: 6655), nr_events: 5 task 58 ( sh: 6656), nr_events: 4 task 59 ( gcc: 6657), nr_events: 9 task 60 ( ksoftirqd/3: 10), nr_events: 1 task 61 ( gcc: 6658), nr_events: 4 task 62 ( make: 6659), nr_events: 5 task 63 ( sh: 6660), nr_events: 3 task 64 ( gcc: 6661), nr_events: 5 task 65 ( collect2: 6662), nr_events: 4 ------------------------------------------------------------ #1 : 256.745, ravg: 256.74, cpu: 0.00 / 0.00 #2 : 439.372, ravg: 275.01, cpu: 0.00 / 0.00 #3 : 411.971, ravg: 288.70, cpu: 0.00 / 0.00 #4 : 385.500, ravg: 298.38, cpu: 0.00 / 0.00 #5 : 366.526, ravg: 305.20, cpu: 0.00 / 0.00 #6 : 381.281, ravg: 312.81, cpu: 0.00 / 0.00 #7 : 410.756, ravg: 322.60, cpu: 0.00 / 0.00 #8 : 368.009, ravg: 327.14, cpu: 0.00 / 0.00 #9 : 408.098, ravg: 335.24, cpu: 0.00 / 0.00 #10 : 368.582, ravg: 338.57, cpu: 0.00 / 0.00 I.e. we successfully analyzed the trace, replayed it via real threads and measured the replayed workload's scheduling properties. This is how it looked like in 'top' output: PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND 7164 mingo 20 0 1434m 8080 888 R 57.0 0.1 0:02.04 :perf 7165 mingo 20 0 1434m 8080 888 R 41.8 0.1 0:01.52 :perf 7228 mingo 20 0 1434m 8080 888 R 39.8 0.1 0:01.44 :gcc 7225 mingo 20 0 1434m 8080 888 R 33.8 0.1 0:01.26 :gcc 7202 mingo 20 0 1434m 8080 888 R 31.2 0.1 0:01.16 :sh 7222 mingo 20 0 1434m 8080 888 R 25.2 0.1 0:00.96 :sh 7211 mingo 20 0 1434m 8080 888 R 21.9 0.1 0:00.82 :sh 7213 mingo 20 0 1434m 8080 888 D 19.2 0.1 0:00.74 :sh 7194 mingo 20 0 1434m 8080 888 D 18.6 0.1 0:00.72 :make There's still various kinks in it - more patches to come. Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Mike Galbraith <efault@gmx.de> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> LKML-Reference: <new-submission> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-09-11 06:12:54 -04:00
}
static void
process_sched_exit_event(struct event *event,
int cpu __used,
u64 timestamp __used,
struct thread *thread __used)
perf sched: Implement the scheduling workload replay engine Integrate the schedbench.c bits with the raw trace events that we get from the perf machinery, and activate the workload replayer/simulator. Example of a captured 'make -j' workload: $ perf sched run measurement overhead: 90 nsecs sleep measurement overhead: 2724743 nsecs the run test took 1000081 nsecs the sleep test took 2981111 nsecs version = 0.5 ... nr_run_events: 70 nr_sleep_events: 66 nr_wakeup_events: 9 target-less wakeups: 71 multi-target wakeups: 47 run events optimized: 139 task 0 ( perf: 6607), nr_events: 2 task 1 ( perf: 6608), nr_events: 6 task 2 ( : 0), nr_events: 1 task 3 ( make: 6609), nr_events: 5 task 4 ( sh: 6610), nr_events: 4 task 5 ( make: 6611), nr_events: 6 task 6 ( sh: 6612), nr_events: 4 task 7 ( make: 6613), nr_events: 5 task 8 ( migration/11: 25), nr_events: 1 task 9 ( migration/13: 29), nr_events: 1 task 10 ( migration/15: 33), nr_events: 1 task 11 ( migration/9: 21), nr_events: 1 task 12 ( sh: 6614), nr_events: 4 task 13 ( make: 6615), nr_events: 5 task 14 ( sh: 6616), nr_events: 4 task 15 ( make: 6617), nr_events: 7 task 16 ( migration/3: 9), nr_events: 1 task 17 ( migration/5: 13), nr_events: 1 task 18 ( migration/7: 17), nr_events: 1 task 19 ( migration/1: 5), nr_events: 1 task 20 ( sh: 6618), nr_events: 4 task 21 ( make: 6619), nr_events: 5 task 22 ( sh: 6620), nr_events: 4 task 23 ( make: 6621), nr_events: 10 task 24 ( sh: 6623), nr_events: 3 task 25 ( gcc: 6624), nr_events: 4 task 26 ( gcc: 6625), nr_events: 4 task 27 ( gcc: 6626), nr_events: 5 task 28 ( collect2: 6627), nr_events: 5 task 29 ( sh: 6622), nr_events: 1 task 30 ( make: 6628), nr_events: 7 task 31 ( sh: 6630), nr_events: 4 task 32 ( gcc: 6631), nr_events: 4 task 33 ( sh: 6629), nr_events: 1 task 34 ( gcc: 6632), nr_events: 4 task 35 ( gcc: 6633), nr_events: 4 task 36 ( collect2: 6634), nr_events: 4 task 37 ( make: 6635), nr_events: 8 task 38 ( sh: 6637), nr_events: 4 task 39 ( sh: 6636), nr_events: 1 task 40 ( gcc: 6638), nr_events: 4 task 41 ( gcc: 6639), nr_events: 4 task 42 ( gcc: 6640), nr_events: 4 task 43 ( collect2: 6641), nr_events: 4 task 44 ( make: 6642), nr_events: 6 task 45 ( sh: 6643), nr_events: 5 task 46 ( sh: 6644), nr_events: 3 task 47 ( sh: 6645), nr_events: 4 task 48 ( make: 6646), nr_events: 6 task 49 ( sh: 6647), nr_events: 3 task 50 ( make: 6648), nr_events: 5 task 51 ( sh: 6649), nr_events: 5 task 52 ( sh: 6650), nr_events: 6 task 53 ( make: 6651), nr_events: 4 task 54 ( make: 6652), nr_events: 5 task 55 ( make: 6653), nr_events: 4 task 56 ( make: 6654), nr_events: 4 task 57 ( make: 6655), nr_events: 5 task 58 ( sh: 6656), nr_events: 4 task 59 ( gcc: 6657), nr_events: 9 task 60 ( ksoftirqd/3: 10), nr_events: 1 task 61 ( gcc: 6658), nr_events: 4 task 62 ( make: 6659), nr_events: 5 task 63 ( sh: 6660), nr_events: 3 task 64 ( gcc: 6661), nr_events: 5 task 65 ( collect2: 6662), nr_events: 4 ------------------------------------------------------------ #1 : 256.745, ravg: 256.74, cpu: 0.00 / 0.00 #2 : 439.372, ravg: 275.01, cpu: 0.00 / 0.00 #3 : 411.971, ravg: 288.70, cpu: 0.00 / 0.00 #4 : 385.500, ravg: 298.38, cpu: 0.00 / 0.00 #5 : 366.526, ravg: 305.20, cpu: 0.00 / 0.00 #6 : 381.281, ravg: 312.81, cpu: 0.00 / 0.00 #7 : 410.756, ravg: 322.60, cpu: 0.00 / 0.00 #8 : 368.009, ravg: 327.14, cpu: 0.00 / 0.00 #9 : 408.098, ravg: 335.24, cpu: 0.00 / 0.00 #10 : 368.582, ravg: 338.57, cpu: 0.00 / 0.00 I.e. we successfully analyzed the trace, replayed it via real threads and measured the replayed workload's scheduling properties. This is how it looked like in 'top' output: PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND 7164 mingo 20 0 1434m 8080 888 R 57.0 0.1 0:02.04 :perf 7165 mingo 20 0 1434m 8080 888 R 41.8 0.1 0:01.52 :perf 7228 mingo 20 0 1434m 8080 888 R 39.8 0.1 0:01.44 :gcc 7225 mingo 20 0 1434m 8080 888 R 33.8 0.1 0:01.26 :gcc 7202 mingo 20 0 1434m 8080 888 R 31.2 0.1 0:01.16 :sh 7222 mingo 20 0 1434m 8080 888 R 25.2 0.1 0:00.96 :sh 7211 mingo 20 0 1434m 8080 888 R 21.9 0.1 0:00.82 :sh 7213 mingo 20 0 1434m 8080 888 D 19.2 0.1 0:00.74 :sh 7194 mingo 20 0 1434m 8080 888 D 18.6 0.1 0:00.72 :make There's still various kinks in it - more patches to come. Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Mike Galbraith <efault@gmx.de> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> LKML-Reference: <new-submission> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-09-11 06:12:54 -04:00
{
if (verbose)
printf("sched_exit event %p\n", event);
}
static void
process_raw_event(event_t *raw_event __used, void *more_data,
int cpu, u64 timestamp, struct thread *thread)
{
perf sched: Fix bad event alignment perf sched raises the following error when it meets a sched switch event: perf: builtin-sched.c:286: register_pid: Assertion `!(pid >= 65536)' failed. Abandon Currently in x86-64, the sched switch events have a hole in the middle of the structure: u16 common_type; u8 common_flags; u8 common_preempt_count; u32 common_pid; u32 common_tgid; char prev_comm[16]; u32 prev_pid; u32 prev_prio; <--- there u64 prev_state; char next_comm[16]; u32 next_pid; u32 next_prio; Gcc inserts a 4 bytes hole there for prev_state to be u64 aligned. And the events are exported to userspace with this hole. But in userspace, from perf sched, we fetch it using a structure that has a new field in the beginning: u32 size. This is because our trace is exported with its size as a field. But now that we have this new field, the hole in the middle disappears because it makes prev_state becoming well aligned. And since we are using a pointer to the raw trace using this struct, instead of reading prev_state, we are reading the hole. We could fix it by keeping the size seperate from the struct but actually there a lot of other potential problems: some fields may be saved as long in a 64 bits system and later read as long in a 32 bits system. Also this direct cast doesn't care about the endianness differences between the host traced machine and the machine in which we do the post processing. So instead of using such dangerous direct casts, fetch the values using the trace parsing API that already takes care of all these problems. Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Mike Galbraith <efault@gmx.de> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-09-11 20:43:45 -04:00
struct raw_event_sample *raw = more_data;
struct event *event;
int type;
type = trace_parse_common_type(raw->data);
event = trace_find_event(type);
if (!strcmp(event->name, "sched_switch"))
perf sched: Fix bad event alignment perf sched raises the following error when it meets a sched switch event: perf: builtin-sched.c:286: register_pid: Assertion `!(pid >= 65536)' failed. Abandon Currently in x86-64, the sched switch events have a hole in the middle of the structure: u16 common_type; u8 common_flags; u8 common_preempt_count; u32 common_pid; u32 common_tgid; char prev_comm[16]; u32 prev_pid; u32 prev_prio; <--- there u64 prev_state; char next_comm[16]; u32 next_pid; u32 next_prio; Gcc inserts a 4 bytes hole there for prev_state to be u64 aligned. And the events are exported to userspace with this hole. But in userspace, from perf sched, we fetch it using a structure that has a new field in the beginning: u32 size. This is because our trace is exported with its size as a field. But now that we have this new field, the hole in the middle disappears because it makes prev_state becoming well aligned. And since we are using a pointer to the raw trace using this struct, instead of reading prev_state, we are reading the hole. We could fix it by keeping the size seperate from the struct but actually there a lot of other potential problems: some fields may be saved as long in a 64 bits system and later read as long in a 32 bits system. Also this direct cast doesn't care about the endianness differences between the host traced machine and the machine in which we do the post processing. So instead of using such dangerous direct casts, fetch the values using the trace parsing API that already takes care of all these problems. Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Mike Galbraith <efault@gmx.de> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-09-11 20:43:45 -04:00
process_sched_switch_event(raw, event, cpu, timestamp, thread);
if (!strcmp(event->name, "sched_wakeup"))
perf sched: Fix bad event alignment perf sched raises the following error when it meets a sched switch event: perf: builtin-sched.c:286: register_pid: Assertion `!(pid >= 65536)' failed. Abandon Currently in x86-64, the sched switch events have a hole in the middle of the structure: u16 common_type; u8 common_flags; u8 common_preempt_count; u32 common_pid; u32 common_tgid; char prev_comm[16]; u32 prev_pid; u32 prev_prio; <--- there u64 prev_state; char next_comm[16]; u32 next_pid; u32 next_prio; Gcc inserts a 4 bytes hole there for prev_state to be u64 aligned. And the events are exported to userspace with this hole. But in userspace, from perf sched, we fetch it using a structure that has a new field in the beginning: u32 size. This is because our trace is exported with its size as a field. But now that we have this new field, the hole in the middle disappears because it makes prev_state becoming well aligned. And since we are using a pointer to the raw trace using this struct, instead of reading prev_state, we are reading the hole. We could fix it by keeping the size seperate from the struct but actually there a lot of other potential problems: some fields may be saved as long in a 64 bits system and later read as long in a 32 bits system. Also this direct cast doesn't care about the endianness differences between the host traced machine and the machine in which we do the post processing. So instead of using such dangerous direct casts, fetch the values using the trace parsing API that already takes care of all these problems. Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Mike Galbraith <efault@gmx.de> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-09-11 20:43:45 -04:00
process_sched_wakeup_event(raw, event, cpu, timestamp, thread);
perf sched: Implement the scheduling workload replay engine Integrate the schedbench.c bits with the raw trace events that we get from the perf machinery, and activate the workload replayer/simulator. Example of a captured 'make -j' workload: $ perf sched run measurement overhead: 90 nsecs sleep measurement overhead: 2724743 nsecs the run test took 1000081 nsecs the sleep test took 2981111 nsecs version = 0.5 ... nr_run_events: 70 nr_sleep_events: 66 nr_wakeup_events: 9 target-less wakeups: 71 multi-target wakeups: 47 run events optimized: 139 task 0 ( perf: 6607), nr_events: 2 task 1 ( perf: 6608), nr_events: 6 task 2 ( : 0), nr_events: 1 task 3 ( make: 6609), nr_events: 5 task 4 ( sh: 6610), nr_events: 4 task 5 ( make: 6611), nr_events: 6 task 6 ( sh: 6612), nr_events: 4 task 7 ( make: 6613), nr_events: 5 task 8 ( migration/11: 25), nr_events: 1 task 9 ( migration/13: 29), nr_events: 1 task 10 ( migration/15: 33), nr_events: 1 task 11 ( migration/9: 21), nr_events: 1 task 12 ( sh: 6614), nr_events: 4 task 13 ( make: 6615), nr_events: 5 task 14 ( sh: 6616), nr_events: 4 task 15 ( make: 6617), nr_events: 7 task 16 ( migration/3: 9), nr_events: 1 task 17 ( migration/5: 13), nr_events: 1 task 18 ( migration/7: 17), nr_events: 1 task 19 ( migration/1: 5), nr_events: 1 task 20 ( sh: 6618), nr_events: 4 task 21 ( make: 6619), nr_events: 5 task 22 ( sh: 6620), nr_events: 4 task 23 ( make: 6621), nr_events: 10 task 24 ( sh: 6623), nr_events: 3 task 25 ( gcc: 6624), nr_events: 4 task 26 ( gcc: 6625), nr_events: 4 task 27 ( gcc: 6626), nr_events: 5 task 28 ( collect2: 6627), nr_events: 5 task 29 ( sh: 6622), nr_events: 1 task 30 ( make: 6628), nr_events: 7 task 31 ( sh: 6630), nr_events: 4 task 32 ( gcc: 6631), nr_events: 4 task 33 ( sh: 6629), nr_events: 1 task 34 ( gcc: 6632), nr_events: 4 task 35 ( gcc: 6633), nr_events: 4 task 36 ( collect2: 6634), nr_events: 4 task 37 ( make: 6635), nr_events: 8 task 38 ( sh: 6637), nr_events: 4 task 39 ( sh: 6636), nr_events: 1 task 40 ( gcc: 6638), nr_events: 4 task 41 ( gcc: 6639), nr_events: 4 task 42 ( gcc: 6640), nr_events: 4 task 43 ( collect2: 6641), nr_events: 4 task 44 ( make: 6642), nr_events: 6 task 45 ( sh: 6643), nr_events: 5 task 46 ( sh: 6644), nr_events: 3 task 47 ( sh: 6645), nr_events: 4 task 48 ( make: 6646), nr_events: 6 task 49 ( sh: 6647), nr_events: 3 task 50 ( make: 6648), nr_events: 5 task 51 ( sh: 6649), nr_events: 5 task 52 ( sh: 6650), nr_events: 6 task 53 ( make: 6651), nr_events: 4 task 54 ( make: 6652), nr_events: 5 task 55 ( make: 6653), nr_events: 4 task 56 ( make: 6654), nr_events: 4 task 57 ( make: 6655), nr_events: 5 task 58 ( sh: 6656), nr_events: 4 task 59 ( gcc: 6657), nr_events: 9 task 60 ( ksoftirqd/3: 10), nr_events: 1 task 61 ( gcc: 6658), nr_events: 4 task 62 ( make: 6659), nr_events: 5 task 63 ( sh: 6660), nr_events: 3 task 64 ( gcc: 6661), nr_events: 5 task 65 ( collect2: 6662), nr_events: 4 ------------------------------------------------------------ #1 : 256.745, ravg: 256.74, cpu: 0.00 / 0.00 #2 : 439.372, ravg: 275.01, cpu: 0.00 / 0.00 #3 : 411.971, ravg: 288.70, cpu: 0.00 / 0.00 #4 : 385.500, ravg: 298.38, cpu: 0.00 / 0.00 #5 : 366.526, ravg: 305.20, cpu: 0.00 / 0.00 #6 : 381.281, ravg: 312.81, cpu: 0.00 / 0.00 #7 : 410.756, ravg: 322.60, cpu: 0.00 / 0.00 #8 : 368.009, ravg: 327.14, cpu: 0.00 / 0.00 #9 : 408.098, ravg: 335.24, cpu: 0.00 / 0.00 #10 : 368.582, ravg: 338.57, cpu: 0.00 / 0.00 I.e. we successfully analyzed the trace, replayed it via real threads and measured the replayed workload's scheduling properties. This is how it looked like in 'top' output: PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND 7164 mingo 20 0 1434m 8080 888 R 57.0 0.1 0:02.04 :perf 7165 mingo 20 0 1434m 8080 888 R 41.8 0.1 0:01.52 :perf 7228 mingo 20 0 1434m 8080 888 R 39.8 0.1 0:01.44 :gcc 7225 mingo 20 0 1434m 8080 888 R 33.8 0.1 0:01.26 :gcc 7202 mingo 20 0 1434m 8080 888 R 31.2 0.1 0:01.16 :sh 7222 mingo 20 0 1434m 8080 888 R 25.2 0.1 0:00.96 :sh 7211 mingo 20 0 1434m 8080 888 R 21.9 0.1 0:00.82 :sh 7213 mingo 20 0 1434m 8080 888 D 19.2 0.1 0:00.74 :sh 7194 mingo 20 0 1434m 8080 888 D 18.6 0.1 0:00.72 :make There's still various kinks in it - more patches to come. Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Mike Galbraith <efault@gmx.de> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> LKML-Reference: <new-submission> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-09-11 06:12:54 -04:00
if (!strcmp(event->name, "sched_wakeup_new"))
perf sched: Fix bad event alignment perf sched raises the following error when it meets a sched switch event: perf: builtin-sched.c:286: register_pid: Assertion `!(pid >= 65536)' failed. Abandon Currently in x86-64, the sched switch events have a hole in the middle of the structure: u16 common_type; u8 common_flags; u8 common_preempt_count; u32 common_pid; u32 common_tgid; char prev_comm[16]; u32 prev_pid; u32 prev_prio; <--- there u64 prev_state; char next_comm[16]; u32 next_pid; u32 next_prio; Gcc inserts a 4 bytes hole there for prev_state to be u64 aligned. And the events are exported to userspace with this hole. But in userspace, from perf sched, we fetch it using a structure that has a new field in the beginning: u32 size. This is because our trace is exported with its size as a field. But now that we have this new field, the hole in the middle disappears because it makes prev_state becoming well aligned. And since we are using a pointer to the raw trace using this struct, instead of reading prev_state, we are reading the hole. We could fix it by keeping the size seperate from the struct but actually there a lot of other potential problems: some fields may be saved as long in a 64 bits system and later read as long in a 32 bits system. Also this direct cast doesn't care about the endianness differences between the host traced machine and the machine in which we do the post processing. So instead of using such dangerous direct casts, fetch the values using the trace parsing API that already takes care of all these problems. Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Mike Galbraith <efault@gmx.de> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-09-11 20:43:45 -04:00
process_sched_wakeup_event(raw, event, cpu, timestamp, thread);
perf sched: Implement the scheduling workload replay engine Integrate the schedbench.c bits with the raw trace events that we get from the perf machinery, and activate the workload replayer/simulator. Example of a captured 'make -j' workload: $ perf sched run measurement overhead: 90 nsecs sleep measurement overhead: 2724743 nsecs the run test took 1000081 nsecs the sleep test took 2981111 nsecs version = 0.5 ... nr_run_events: 70 nr_sleep_events: 66 nr_wakeup_events: 9 target-less wakeups: 71 multi-target wakeups: 47 run events optimized: 139 task 0 ( perf: 6607), nr_events: 2 task 1 ( perf: 6608), nr_events: 6 task 2 ( : 0), nr_events: 1 task 3 ( make: 6609), nr_events: 5 task 4 ( sh: 6610), nr_events: 4 task 5 ( make: 6611), nr_events: 6 task 6 ( sh: 6612), nr_events: 4 task 7 ( make: 6613), nr_events: 5 task 8 ( migration/11: 25), nr_events: 1 task 9 ( migration/13: 29), nr_events: 1 task 10 ( migration/15: 33), nr_events: 1 task 11 ( migration/9: 21), nr_events: 1 task 12 ( sh: 6614), nr_events: 4 task 13 ( make: 6615), nr_events: 5 task 14 ( sh: 6616), nr_events: 4 task 15 ( make: 6617), nr_events: 7 task 16 ( migration/3: 9), nr_events: 1 task 17 ( migration/5: 13), nr_events: 1 task 18 ( migration/7: 17), nr_events: 1 task 19 ( migration/1: 5), nr_events: 1 task 20 ( sh: 6618), nr_events: 4 task 21 ( make: 6619), nr_events: 5 task 22 ( sh: 6620), nr_events: 4 task 23 ( make: 6621), nr_events: 10 task 24 ( sh: 6623), nr_events: 3 task 25 ( gcc: 6624), nr_events: 4 task 26 ( gcc: 6625), nr_events: 4 task 27 ( gcc: 6626), nr_events: 5 task 28 ( collect2: 6627), nr_events: 5 task 29 ( sh: 6622), nr_events: 1 task 30 ( make: 6628), nr_events: 7 task 31 ( sh: 6630), nr_events: 4 task 32 ( gcc: 6631), nr_events: 4 task 33 ( sh: 6629), nr_events: 1 task 34 ( gcc: 6632), nr_events: 4 task 35 ( gcc: 6633), nr_events: 4 task 36 ( collect2: 6634), nr_events: 4 task 37 ( make: 6635), nr_events: 8 task 38 ( sh: 6637), nr_events: 4 task 39 ( sh: 6636), nr_events: 1 task 40 ( gcc: 6638), nr_events: 4 task 41 ( gcc: 6639), nr_events: 4 task 42 ( gcc: 6640), nr_events: 4 task 43 ( collect2: 6641), nr_events: 4 task 44 ( make: 6642), nr_events: 6 task 45 ( sh: 6643), nr_events: 5 task 46 ( sh: 6644), nr_events: 3 task 47 ( sh: 6645), nr_events: 4 task 48 ( make: 6646), nr_events: 6 task 49 ( sh: 6647), nr_events: 3 task 50 ( make: 6648), nr_events: 5 task 51 ( sh: 6649), nr_events: 5 task 52 ( sh: 6650), nr_events: 6 task 53 ( make: 6651), nr_events: 4 task 54 ( make: 6652), nr_events: 5 task 55 ( make: 6653), nr_events: 4 task 56 ( make: 6654), nr_events: 4 task 57 ( make: 6655), nr_events: 5 task 58 ( sh: 6656), nr_events: 4 task 59 ( gcc: 6657), nr_events: 9 task 60 ( ksoftirqd/3: 10), nr_events: 1 task 61 ( gcc: 6658), nr_events: 4 task 62 ( make: 6659), nr_events: 5 task 63 ( sh: 6660), nr_events: 3 task 64 ( gcc: 6661), nr_events: 5 task 65 ( collect2: 6662), nr_events: 4 ------------------------------------------------------------ #1 : 256.745, ravg: 256.74, cpu: 0.00 / 0.00 #2 : 439.372, ravg: 275.01, cpu: 0.00 / 0.00 #3 : 411.971, ravg: 288.70, cpu: 0.00 / 0.00 #4 : 385.500, ravg: 298.38, cpu: 0.00 / 0.00 #5 : 366.526, ravg: 305.20, cpu: 0.00 / 0.00 #6 : 381.281, ravg: 312.81, cpu: 0.00 / 0.00 #7 : 410.756, ravg: 322.60, cpu: 0.00 / 0.00 #8 : 368.009, ravg: 327.14, cpu: 0.00 / 0.00 #9 : 408.098, ravg: 335.24, cpu: 0.00 / 0.00 #10 : 368.582, ravg: 338.57, cpu: 0.00 / 0.00 I.e. we successfully analyzed the trace, replayed it via real threads and measured the replayed workload's scheduling properties. This is how it looked like in 'top' output: PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND 7164 mingo 20 0 1434m 8080 888 R 57.0 0.1 0:02.04 :perf 7165 mingo 20 0 1434m 8080 888 R 41.8 0.1 0:01.52 :perf 7228 mingo 20 0 1434m 8080 888 R 39.8 0.1 0:01.44 :gcc 7225 mingo 20 0 1434m 8080 888 R 33.8 0.1 0:01.26 :gcc 7202 mingo 20 0 1434m 8080 888 R 31.2 0.1 0:01.16 :sh 7222 mingo 20 0 1434m 8080 888 R 25.2 0.1 0:00.96 :sh 7211 mingo 20 0 1434m 8080 888 R 21.9 0.1 0:00.82 :sh 7213 mingo 20 0 1434m 8080 888 D 19.2 0.1 0:00.74 :sh 7194 mingo 20 0 1434m 8080 888 D 18.6 0.1 0:00.72 :make There's still various kinks in it - more patches to come. Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Mike Galbraith <efault@gmx.de> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> LKML-Reference: <new-submission> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-09-11 06:12:54 -04:00
if (!strcmp(event->name, "sched_process_fork"))
perf sched: Fix bad event alignment perf sched raises the following error when it meets a sched switch event: perf: builtin-sched.c:286: register_pid: Assertion `!(pid >= 65536)' failed. Abandon Currently in x86-64, the sched switch events have a hole in the middle of the structure: u16 common_type; u8 common_flags; u8 common_preempt_count; u32 common_pid; u32 common_tgid; char prev_comm[16]; u32 prev_pid; u32 prev_prio; <--- there u64 prev_state; char next_comm[16]; u32 next_pid; u32 next_prio; Gcc inserts a 4 bytes hole there for prev_state to be u64 aligned. And the events are exported to userspace with this hole. But in userspace, from perf sched, we fetch it using a structure that has a new field in the beginning: u32 size. This is because our trace is exported with its size as a field. But now that we have this new field, the hole in the middle disappears because it makes prev_state becoming well aligned. And since we are using a pointer to the raw trace using this struct, instead of reading prev_state, we are reading the hole. We could fix it by keeping the size seperate from the struct but actually there a lot of other potential problems: some fields may be saved as long in a 64 bits system and later read as long in a 32 bits system. Also this direct cast doesn't care about the endianness differences between the host traced machine and the machine in which we do the post processing. So instead of using such dangerous direct casts, fetch the values using the trace parsing API that already takes care of all these problems. Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Mike Galbraith <efault@gmx.de> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-09-11 20:43:45 -04:00
process_sched_fork_event(raw, event, cpu, timestamp, thread);
perf sched: Implement the scheduling workload replay engine Integrate the schedbench.c bits with the raw trace events that we get from the perf machinery, and activate the workload replayer/simulator. Example of a captured 'make -j' workload: $ perf sched run measurement overhead: 90 nsecs sleep measurement overhead: 2724743 nsecs the run test took 1000081 nsecs the sleep test took 2981111 nsecs version = 0.5 ... nr_run_events: 70 nr_sleep_events: 66 nr_wakeup_events: 9 target-less wakeups: 71 multi-target wakeups: 47 run events optimized: 139 task 0 ( perf: 6607), nr_events: 2 task 1 ( perf: 6608), nr_events: 6 task 2 ( : 0), nr_events: 1 task 3 ( make: 6609), nr_events: 5 task 4 ( sh: 6610), nr_events: 4 task 5 ( make: 6611), nr_events: 6 task 6 ( sh: 6612), nr_events: 4 task 7 ( make: 6613), nr_events: 5 task 8 ( migration/11: 25), nr_events: 1 task 9 ( migration/13: 29), nr_events: 1 task 10 ( migration/15: 33), nr_events: 1 task 11 ( migration/9: 21), nr_events: 1 task 12 ( sh: 6614), nr_events: 4 task 13 ( make: 6615), nr_events: 5 task 14 ( sh: 6616), nr_events: 4 task 15 ( make: 6617), nr_events: 7 task 16 ( migration/3: 9), nr_events: 1 task 17 ( migration/5: 13), nr_events: 1 task 18 ( migration/7: 17), nr_events: 1 task 19 ( migration/1: 5), nr_events: 1 task 20 ( sh: 6618), nr_events: 4 task 21 ( make: 6619), nr_events: 5 task 22 ( sh: 6620), nr_events: 4 task 23 ( make: 6621), nr_events: 10 task 24 ( sh: 6623), nr_events: 3 task 25 ( gcc: 6624), nr_events: 4 task 26 ( gcc: 6625), nr_events: 4 task 27 ( gcc: 6626), nr_events: 5 task 28 ( collect2: 6627), nr_events: 5 task 29 ( sh: 6622), nr_events: 1 task 30 ( make: 6628), nr_events: 7 task 31 ( sh: 6630), nr_events: 4 task 32 ( gcc: 6631), nr_events: 4 task 33 ( sh: 6629), nr_events: 1 task 34 ( gcc: 6632), nr_events: 4 task 35 ( gcc: 6633), nr_events: 4 task 36 ( collect2: 6634), nr_events: 4 task 37 ( make: 6635), nr_events: 8 task 38 ( sh: 6637), nr_events: 4 task 39 ( sh: 6636), nr_events: 1 task 40 ( gcc: 6638), nr_events: 4 task 41 ( gcc: 6639), nr_events: 4 task 42 ( gcc: 6640), nr_events: 4 task 43 ( collect2: 6641), nr_events: 4 task 44 ( make: 6642), nr_events: 6 task 45 ( sh: 6643), nr_events: 5 task 46 ( sh: 6644), nr_events: 3 task 47 ( sh: 6645), nr_events: 4 task 48 ( make: 6646), nr_events: 6 task 49 ( sh: 6647), nr_events: 3 task 50 ( make: 6648), nr_events: 5 task 51 ( sh: 6649), nr_events: 5 task 52 ( sh: 6650), nr_events: 6 task 53 ( make: 6651), nr_events: 4 task 54 ( make: 6652), nr_events: 5 task 55 ( make: 6653), nr_events: 4 task 56 ( make: 6654), nr_events: 4 task 57 ( make: 6655), nr_events: 5 task 58 ( sh: 6656), nr_events: 4 task 59 ( gcc: 6657), nr_events: 9 task 60 ( ksoftirqd/3: 10), nr_events: 1 task 61 ( gcc: 6658), nr_events: 4 task 62 ( make: 6659), nr_events: 5 task 63 ( sh: 6660), nr_events: 3 task 64 ( gcc: 6661), nr_events: 5 task 65 ( collect2: 6662), nr_events: 4 ------------------------------------------------------------ #1 : 256.745, ravg: 256.74, cpu: 0.00 / 0.00 #2 : 439.372, ravg: 275.01, cpu: 0.00 / 0.00 #3 : 411.971, ravg: 288.70, cpu: 0.00 / 0.00 #4 : 385.500, ravg: 298.38, cpu: 0.00 / 0.00 #5 : 366.526, ravg: 305.20, cpu: 0.00 / 0.00 #6 : 381.281, ravg: 312.81, cpu: 0.00 / 0.00 #7 : 410.756, ravg: 322.60, cpu: 0.00 / 0.00 #8 : 368.009, ravg: 327.14, cpu: 0.00 / 0.00 #9 : 408.098, ravg: 335.24, cpu: 0.00 / 0.00 #10 : 368.582, ravg: 338.57, cpu: 0.00 / 0.00 I.e. we successfully analyzed the trace, replayed it via real threads and measured the replayed workload's scheduling properties. This is how it looked like in 'top' output: PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND 7164 mingo 20 0 1434m 8080 888 R 57.0 0.1 0:02.04 :perf 7165 mingo 20 0 1434m 8080 888 R 41.8 0.1 0:01.52 :perf 7228 mingo 20 0 1434m 8080 888 R 39.8 0.1 0:01.44 :gcc 7225 mingo 20 0 1434m 8080 888 R 33.8 0.1 0:01.26 :gcc 7202 mingo 20 0 1434m 8080 888 R 31.2 0.1 0:01.16 :sh 7222 mingo 20 0 1434m 8080 888 R 25.2 0.1 0:00.96 :sh 7211 mingo 20 0 1434m 8080 888 R 21.9 0.1 0:00.82 :sh 7213 mingo 20 0 1434m 8080 888 D 19.2 0.1 0:00.74 :sh 7194 mingo 20 0 1434m 8080 888 D 18.6 0.1 0:00.72 :make There's still various kinks in it - more patches to come. Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Mike Galbraith <efault@gmx.de> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> LKML-Reference: <new-submission> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-09-11 06:12:54 -04:00
if (!strcmp(event->name, "sched_process_exit"))
process_sched_exit_event(event, cpu, timestamp, thread);
}
static int
process_sample_event(event_t *event, unsigned long offset, unsigned long head)
{
char level;
int show = 0;
struct dso *dso = NULL;
struct thread *thread;
u64 ip = event->ip.ip;
u64 timestamp = -1;
u32 cpu = -1;
u64 period = 1;
void *more_data = event->ip.__more_data;
int cpumode;
thread = threads__findnew(event->ip.pid, &threads, &last_match);
if (sample_type & PERF_SAMPLE_TIME) {
timestamp = *(u64 *)more_data;
more_data += sizeof(u64);
}
if (sample_type & PERF_SAMPLE_CPU) {
cpu = *(u32 *)more_data;
more_data += sizeof(u32);
more_data += sizeof(u32); /* reserved */
}
if (sample_type & PERF_SAMPLE_PERIOD) {
period = *(u64 *)more_data;
more_data += sizeof(u64);
}
dump_printf("%p [%p]: PERF_EVENT_SAMPLE (IP, %d): %d/%d: %p period: %Ld\n",
(void *)(offset + head),
(void *)(long)(event->header.size),
event->header.misc,
event->ip.pid, event->ip.tid,
(void *)(long)ip,
(long long)period);
dump_printf(" ... thread: %s:%d\n", thread->comm, thread->pid);
if (thread == NULL) {
eprintf("problem processing %d event, skipping it.\n",
event->header.type);
return -1;
}
cpumode = event->header.misc & PERF_EVENT_MISC_CPUMODE_MASK;
if (cpumode == PERF_EVENT_MISC_KERNEL) {
show = SHOW_KERNEL;
level = 'k';
dso = kernel_dso;
dump_printf(" ...... dso: %s\n", dso->name);
} else if (cpumode == PERF_EVENT_MISC_USER) {
show = SHOW_USER;
level = '.';
} else {
show = SHOW_HV;
level = 'H';
dso = hypervisor_dso;
dump_printf(" ...... dso: [hypervisor]\n");
}
if (sample_type & PERF_SAMPLE_RAW)
process_raw_event(event, more_data, cpu, timestamp, thread);
return 0;
}
static int
process_event(event_t *event, unsigned long offset, unsigned long head)
{
trace_event(event);
switch (event->header.type) {
case PERF_EVENT_MMAP ... PERF_EVENT_LOST:
return 0;
case PERF_EVENT_COMM:
return process_comm_event(event, offset, head);
case PERF_EVENT_EXIT ... PERF_EVENT_READ:
return 0;
case PERF_EVENT_SAMPLE:
return process_sample_event(event, offset, head);
case PERF_EVENT_MAX:
default:
return -1;
}
return 0;
}
static int read_events(void)
{
int ret, rc = EXIT_FAILURE;
unsigned long offset = 0;
unsigned long head = 0;
struct stat perf_stat;
event_t *event;
uint32_t size;
char *buf;
trace_report();
register_idle_thread(&threads, &last_match);
input = open(input_name, O_RDONLY);
if (input < 0) {
perror("failed to open file");
exit(-1);
}
ret = fstat(input, &perf_stat);
if (ret < 0) {
perror("failed to stat file");
exit(-1);
}
if (!perf_stat.st_size) {
fprintf(stderr, "zero-sized file, nothing to do!\n");
exit(0);
}
header = perf_header__read(input);
head = header->data_offset;
sample_type = perf_header__sample_type(header);
if (!(sample_type & PERF_SAMPLE_RAW))
die("No trace sample to read. Did you call perf record "
"without -R?");
if (load_kernel() < 0) {
perror("failed to load kernel symbols");
return EXIT_FAILURE;
}
remap:
buf = (char *)mmap(NULL, page_size * mmap_window, PROT_READ,
MAP_SHARED, input, offset);
if (buf == MAP_FAILED) {
perror("failed to mmap file");
exit(-1);
}
more:
event = (event_t *)(buf + head);
size = event->header.size;
if (!size)
size = 8;
if (head + event->header.size >= page_size * mmap_window) {
unsigned long shift = page_size * (head / page_size);
int res;
res = munmap(buf, page_size * mmap_window);
assert(res == 0);
offset += shift;
head -= shift;
goto remap;
}
size = event->header.size;
if (!size || process_event(event, offset, head) < 0) {
/*
* assume we lost track of the stream, check alignment, and
* increment a single u64 in the hope to catch on again 'soon'.
*/
if (unlikely(head & 7))
head &= ~7ULL;
size = 8;
}
head += size;
if (offset + head < (unsigned long)perf_stat.st_size)
goto more;
rc = EXIT_SUCCESS;
close(input);
return rc;
}
static const char * const sched_usage[] = {
"perf sched [<options>] <command>",
NULL
};
static const struct option options[] = {
OPT_BOOLEAN('D', "dump-raw-trace", &dump_trace,
"dump raw trace in ASCII"),
OPT_BOOLEAN('r', "replay", &replay_mode,
"replay sched behaviour from traces"),
OPT_BOOLEAN('l', "latency", &lat_mode,
"measure various latencies"),
OPT_BOOLEAN('v', "verbose", &verbose,
"be more verbose (show symbol address, etc)"),
OPT_END()
};
int cmd_sched(int argc, const char **argv, const char *prefix __used)
{
symbol__init();
page_size = getpagesize();
argc = parse_options(argc, argv, options, sched_usage, 0);
if (argc) {
/*
* Special case: if there's an argument left then assume tha
* it's a symbol filter:
*/
if (argc > 1)
usage_with_options(sched_usage, options);
}
if (replay_mode)
trace_handler = &replay_ops;
else if (lat_mode)
trace_handler = &lat_ops;
else
usage_with_options(sched_usage, options);
if (replay_mode)
__cmd_replay();
else if (lat_mode)
__cmd_lat();
return 0;
}