Blame view
Documentation/accounting/getdelays.c
13.2 KB
a3baf649c
|
1 2 3 4 5 6 7 |
/* getdelays.c * * Utility to get per-pid and per-tgid delay accounting statistics * Also illustrates usage of the taskstats interface * * Copyright (C) Shailabh Nagar, IBM Corp. 2005 * Copyright (C) Balbir Singh, IBM Corp. 2006 |
9e06d3f9f
|
8 |
* Copyright (c) Jay Lan, SGI. 2006 |
a3baf649c
|
9 |
* |
d2f7bf134
|
10 11 |
* Compile with * gcc -I/usr/src/linux/include getdelays.c -o getdelays |
a3baf649c
|
12 13 14 15 16 17 18 19 20 21 22 23 |
*/ #include <stdio.h> #include <stdlib.h> #include <errno.h> #include <unistd.h> #include <poll.h> #include <string.h> #include <fcntl.h> #include <sys/types.h> #include <sys/stat.h> #include <sys/socket.h> |
db9e5679d
|
24 |
#include <sys/wait.h> |
a3baf649c
|
25 26 27 28 |
#include <signal.h> #include <linux/genetlink.h> #include <linux/taskstats.h> |
546040dc4
|
29 |
#include <linux/cgroupstats.h> |
a3baf649c
|
30 31 32 33 34 35 36 37 38 39 |
/* * Generic macros for dealing with netlink sockets. Might be duplicated * elsewhere. It is recommended that commercial grade applications use * libnl or libnetlink and use the interfaces provided by the library */ #define GENLMSG_DATA(glh) ((void *)(NLMSG_DATA(glh) + GENL_HDRLEN)) #define GENLMSG_PAYLOAD(glh) (NLMSG_PAYLOAD(glh, 0) - GENL_HDRLEN) #define NLA_DATA(na) ((void *)((char*)(na) + NLA_HDRLEN)) #define NLA_PAYLOAD(len) (len - NLA_HDRLEN) |
d2f7bf134
|
40 41 42 43 44 45 46 47 48 49 50 |
#define err(code, fmt, arg...) \ do { \ fprintf(stderr, fmt, ##arg); \ exit(code); \ } while (0) int done; int rcvbufsz; char name[100]; int dbg; int print_delays; |
cf709844d
|
51 |
int print_io_accounting; |
b663a79c1
|
52 |
int print_task_context_switch_counts; |
9e06d3f9f
|
53 |
__u64 stime, utime; |
d2f7bf134
|
54 |
|
9e06d3f9f
|
55 56 57 58 59 60 61 |
#define PRINTF(fmt, arg...) { \ if (dbg) { \ printf(fmt, ##arg); \ } \ } /* Maximum size of response requested or message sent */ |
880402306
|
62 |
#define MAX_MSG_SIZE 1024 |
9e06d3f9f
|
63 64 |
/* Maximum number of cpus expected to be specified in a cpumask */ #define MAX_CPUS 32 |
9e06d3f9f
|
65 66 67 68 69 70 71 72 |
struct msgtemplate { struct nlmsghdr n; struct genlmsghdr g; char buf[MAX_MSG_SIZE]; }; char cpumask[100+6*MAX_CPUS]; |
a3baf649c
|
73 |
|
f16825bbe
|
74 75 76 77 78 79 80 81 82 83 84 85 86 |
static void usage(void) { fprintf(stderr, "getdelays [-dilv] [-w logfile] [-r bufsize] " "[-m cpumask] [-t tgid] [-p pid] "); fprintf(stderr, " -d: print delayacct stats "); fprintf(stderr, " -i: print IO accounting (works only with -p) "); fprintf(stderr, " -l: listen forever "); fprintf(stderr, " -v: debug on "); |
546040dc4
|
87 88 |
fprintf(stderr, " -C: container path "); |
f16825bbe
|
89 |
} |
a3baf649c
|
90 91 92 |
/* * Create a raw netlink socket and bind */ |
9e06d3f9f
|
93 |
static int create_nl_socket(int protocol) |
a3baf649c
|
94 |
{ |
9e06d3f9f
|
95 96 97 98 99 100 101 102 103 104 |
int fd; struct sockaddr_nl local; fd = socket(AF_NETLINK, SOCK_RAW, protocol); if (fd < 0) return -1; if (rcvbufsz) if (setsockopt(fd, SOL_SOCKET, SO_RCVBUF, &rcvbufsz, sizeof(rcvbufsz)) < 0) { |
d2f7bf134
|
105 106 107 108 |
fprintf(stderr, "Unable to set socket rcv buf size " "to %d ", rcvbufsz); |
9e06d3f9f
|
109 110 |
return -1; } |
a3baf649c
|
111 |
|
9e06d3f9f
|
112 113 |
memset(&local, 0, sizeof(local)); local.nl_family = AF_NETLINK; |
a3baf649c
|
114 |
|
9e06d3f9f
|
115 116 |
if (bind(fd, (struct sockaddr *) &local, sizeof(local)) < 0) goto error; |
a3baf649c
|
117 |
|
9e06d3f9f
|
118 119 120 121 |
return fd; error: close(fd); return -1; |
a3baf649c
|
122 |
} |
9e06d3f9f
|
123 |
|
b7ed698cc
|
124 |
static int send_cmd(int sd, __u16 nlmsg_type, __u32 nlmsg_pid, |
9e06d3f9f
|
125 126 |
__u8 genl_cmd, __u16 nla_type, void *nla_data, int nla_len) |
a3baf649c
|
127 |
{ |
9e06d3f9f
|
128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 |
struct nlattr *na; struct sockaddr_nl nladdr; int r, buflen; char *buf; struct msgtemplate msg; msg.n.nlmsg_len = NLMSG_LENGTH(GENL_HDRLEN); msg.n.nlmsg_type = nlmsg_type; msg.n.nlmsg_flags = NLM_F_REQUEST; msg.n.nlmsg_seq = 0; msg.n.nlmsg_pid = nlmsg_pid; msg.g.cmd = genl_cmd; msg.g.version = 0x1; na = (struct nlattr *) GENLMSG_DATA(&msg); na->nla_type = nla_type; na->nla_len = nla_len + 1 + NLA_HDRLEN; memcpy(NLA_DATA(na), nla_data, nla_len); msg.n.nlmsg_len += NLMSG_ALIGN(na->nla_len); buf = (char *) &msg; buflen = msg.n.nlmsg_len ; memset(&nladdr, 0, sizeof(nladdr)); nladdr.nl_family = AF_NETLINK; while ((r = sendto(sd, buf, buflen, 0, (struct sockaddr *) &nladdr, sizeof(nladdr))) < buflen) { if (r > 0) { buf += r; buflen -= r; } else if (errno != EAGAIN) return -1; } return 0; |
a3baf649c
|
161 |
} |
9e06d3f9f
|
162 |
|
a3baf649c
|
163 164 165 166 |
/* * Probe the controller in genetlink to find the family id * for the TASKSTATS family */ |
b7ed698cc
|
167 |
static int get_family_id(int sd) |
a3baf649c
|
168 |
{ |
9e06d3f9f
|
169 170 171 172 173 |
struct { struct nlmsghdr n; struct genlmsghdr g; char buf[256]; } ans; |
10e6f32bd
|
174 |
int id = 0, rc; |
9e06d3f9f
|
175 176 177 178 179 180 181 |
struct nlattr *na; int rep_len; strcpy(name, TASKSTATS_GENL_NAME); rc = send_cmd(sd, GENL_ID_CTRL, getpid(), CTRL_CMD_GETFAMILY, CTRL_ATTR_FAMILY_NAME, (void *)name, strlen(TASKSTATS_GENL_NAME)+1); |
4ed960b14
|
182 183 |
if (rc < 0) return 0; /* sendto() failure? */ |
9e06d3f9f
|
184 185 186 187 188 |
rep_len = recv(sd, &ans, sizeof(ans), 0); if (ans.n.nlmsg_type == NLMSG_ERROR || (rep_len < 0) || !NLMSG_OK((&ans.n), rep_len)) return 0; |
a3baf649c
|
189 |
|
9e06d3f9f
|
190 191 192 193 194 195 |
na = (struct nlattr *) GENLMSG_DATA(&ans); na = (struct nlattr *) ((char *) na + NLA_ALIGN(na->nla_len)); if (na->nla_type == CTRL_ATTR_FAMILY_ID) { id = *(__u16 *) NLA_DATA(na); } return id; |
a3baf649c
|
196 |
} |
02d54f092
|
197 |
#define average_ms(t, c) (t / 1000000ULL / (c ? c : 1)) |
b7ed698cc
|
198 |
static void print_delayacct(struct taskstats *t) |
a3baf649c
|
199 |
{ |
02d54f092
|
200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 |
printf(" CPU %15s%15s%15s%15s%15s " " %15llu%15llu%15llu%15llu%15.3fms " "IO %15s%15s%15s " " %15llu%15llu%15llums " "SWAP %15s%15s%15s " " %15llu%15llu%15llums " "RECLAIM %12s%15s%15s " " %15llu%15llu%15llums ", "count", "real total", "virtual total", "delay total", "delay average", |
666593137
|
220 221 222 223 |
(unsigned long long)t->cpu_count, (unsigned long long)t->cpu_run_real_total, (unsigned long long)t->cpu_run_virtual_total, (unsigned long long)t->cpu_delay_total, |
02d54f092
|
224 225 |
average_ms((double)t->cpu_delay_total, t->cpu_count), "count", "delay total", "delay average", |
666593137
|
226 227 |
(unsigned long long)t->blkio_count, (unsigned long long)t->blkio_delay_total, |
02d54f092
|
228 229 |
average_ms(t->blkio_delay_total, t->blkio_count), "count", "delay total", "delay average", |
666593137
|
230 231 |
(unsigned long long)t->swapin_count, (unsigned long long)t->swapin_delay_total, |
02d54f092
|
232 233 |
average_ms(t->swapin_delay_total, t->swapin_count), "count", "delay total", "delay average", |
666593137
|
234 |
(unsigned long long)t->freepages_count, |
02d54f092
|
235 236 |
(unsigned long long)t->freepages_delay_total, average_ms(t->freepages_delay_total, t->freepages_count)); |
a3baf649c
|
237 |
} |
b7ed698cc
|
238 |
static void task_context_switch_counts(struct taskstats *t) |
b663a79c1
|
239 240 241 242 243 |
{ printf(" Task %15s%15s " |
10e6f32bd
|
244 245 |
" %15llu%15llu ", |
b663a79c1
|
246 |
"voluntary", "nonvoluntary", |
666593137
|
247 |
(unsigned long long)t->nvcsw, (unsigned long long)t->nivcsw); |
b663a79c1
|
248 |
} |
b7ed698cc
|
249 |
static void print_cgroupstats(struct cgroupstats *c) |
546040dc4
|
250 251 |
{ printf("sleeping %llu, blocked %llu, running %llu, stopped %llu, " |
666593137
|
252 253 254 255 256 257 |
"uninterruptible %llu ", (unsigned long long)c->nr_sleeping, (unsigned long long)c->nr_io_wait, (unsigned long long)c->nr_running, (unsigned long long)c->nr_stopped, (unsigned long long)c->nr_uninterruptible); |
546040dc4
|
258 |
} |
b7ed698cc
|
259 |
static void print_ioacct(struct taskstats *t) |
cf709844d
|
260 261 262 263 264 265 266 267 |
{ printf("%s: read=%llu, write=%llu, cancelled_write=%llu ", t->ac_comm, (unsigned long long)t->read_bytes, (unsigned long long)t->write_bytes, (unsigned long long)t->cancelled_write_bytes); } |
a3baf649c
|
268 269 |
int main(int argc, char *argv[]) { |
b8d9a8659
|
270 271 |
int c, rc, rep_len, aggr_len, len2; int cmd_type = TASKSTATS_CMD_ATTR_UNSPEC; |
9e06d3f9f
|
272 273 274 275 276 277 278 279 280 281 282 283 284 |
__u16 id; __u32 mypid; struct nlattr *na; int nl_sd = -1; int len = 0; pid_t tid = 0; pid_t rtid = 0; int fd = 0; int count = 0; int write_file = 0; int maskset = 0; |
7f76c4037
|
285 |
char *logfile = NULL; |
9e06d3f9f
|
286 |
int loop = 0; |
546040dc4
|
287 288 289 |
int containerset = 0; char containerpath[1024]; int cfd = 0; |
db9e5679d
|
290 291 |
int forking = 0; sigset_t sigset; |
9e06d3f9f
|
292 293 |
struct msgtemplate msg; |
db9e5679d
|
294 295 |
while (!forking) { c = getopt(argc, argv, "qdiw:r:m:t:p:vlC:c:"); |
9e06d3f9f
|
296 297 |
if (c < 0) break; |
a3baf649c
|
298 |
|
9e06d3f9f
|
299 300 301 302 303 304 |
switch (c) { case 'd': printf("print delayacct stats ON "); print_delays = 1; break; |
cf709844d
|
305 306 307 308 309 |
case 'i': printf("printing IO accounting "); print_io_accounting = 1; break; |
b663a79c1
|
310 311 312 313 314 |
case 'q': printf("printing task/process context switch rates "); print_task_context_switch_counts = 1; break; |
546040dc4
|
315 316 317 318 |
case 'C': containerset = 1; strncpy(containerpath, optarg, strlen(optarg) + 1); break; |
9e06d3f9f
|
319 |
case 'w': |
7f76c4037
|
320 |
logfile = strdup(optarg); |
9e06d3f9f
|
321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 |
printf("write to file %s ", logfile); write_file = 1; break; case 'r': rcvbufsz = atoi(optarg); printf("receive buf size %d ", rcvbufsz); if (rcvbufsz < 0) err(1, "Invalid rcv buf size "); break; case 'm': strncpy(cpumask, optarg, sizeof(cpumask)); maskset = 1; printf("cpumask %s maskset %d ", cpumask, maskset); break; case 't': tid = atoi(optarg); if (!tid) err(1, "Invalid tgid "); cmd_type = TASKSTATS_CMD_ATTR_TGID; |
9e06d3f9f
|
345 346 347 348 349 350 351 |
break; case 'p': tid = atoi(optarg); if (!tid) err(1, "Invalid pid "); cmd_type = TASKSTATS_CMD_ATTR_PID; |
9e06d3f9f
|
352 |
break; |
db9e5679d
|
353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 |
case 'c': /* Block SIGCHLD for sigwait() later */ if (sigemptyset(&sigset) == -1) err(1, "Failed to empty sigset"); if (sigaddset(&sigset, SIGCHLD)) err(1, "Failed to set sigchld in sigset"); sigprocmask(SIG_BLOCK, &sigset, NULL); /* fork/exec a child */ tid = fork(); if (tid < 0) err(1, "Fork failed "); if (tid == 0) if (execvp(argv[optind - 1], &argv[optind - 1]) < 0) exit(-1); /* Set the command type and avoid further processing */ cmd_type = TASKSTATS_CMD_ATTR_PID; forking = 1; break; |
9e06d3f9f
|
376 377 378 379 380 381 382 383 384 385 386 |
case 'v': printf("debug on "); dbg = 1; break; case 'l': printf("listen forever "); loop = 1; break; default: |
f16825bbe
|
387 |
usage(); |
9e06d3f9f
|
388 |
exit(-1); |
a3baf649c
|
389 |
} |
a3baf649c
|
390 |
} |
a3baf649c
|
391 |
|
9e06d3f9f
|
392 393 394 395 396 397 398 399 400 |
if (write_file) { fd = open(logfile, O_WRONLY | O_CREAT | O_TRUNC, S_IRUSR | S_IWUSR | S_IRGRP | S_IROTH); if (fd == -1) { perror("Cannot open output file "); exit(1); } } |
a3baf649c
|
401 |
|
9e06d3f9f
|
402 403 404 |
if ((nl_sd = create_nl_socket(NETLINK_GENERIC)) < 0) err(1, "error creating Netlink socket "); |
a3baf649c
|
405 |
|
a3baf649c
|
406 |
|
9e06d3f9f
|
407 408 409 |
mypid = getpid(); id = get_family_id(nl_sd); if (!id) { |
d2f7bf134
|
410 411 |
fprintf(stderr, "Error getting family id, errno %d ", errno); |
9e06d3f9f
|
412 |
goto err; |
a3baf649c
|
413 |
} |
9e06d3f9f
|
414 415 416 417 418 419 |
PRINTF("family id %d ", id); if (maskset) { rc = send_cmd(nl_sd, id, mypid, TASKSTATS_CMD_GET, TASKSTATS_CMD_ATTR_REGISTER_CPUMASK, |
7d1bdca9b
|
420 |
&cpumask, strlen(cpumask) + 1); |
9e06d3f9f
|
421 422 423 |
PRINTF("Sent register cpumask, retval %d ", rc); if (rc < 0) { |
d2f7bf134
|
424 425 |
fprintf(stderr, "error sending register cpumask "); |
9e06d3f9f
|
426 427 |
goto err; } |
a3baf649c
|
428 |
} |
546040dc4
|
429 430 431 432 433 |
if (tid && containerset) { fprintf(stderr, "Select either -t or -C, not both "); goto err; } |
db9e5679d
|
434 435 436 437 438 439 440 441 |
/* * If we forked a child, wait for it to exit. Cannot use waitpid() * as all the delicious data would be reaped as part of the wait */ if (tid && forking) { int sig_received; sigwait(&sigset, &sig_received); } |
9e06d3f9f
|
442 443 444 445 446 447 |
if (tid) { rc = send_cmd(nl_sd, id, mypid, TASKSTATS_CMD_GET, cmd_type, &tid, sizeof(__u32)); PRINTF("Sent pid/tgid, retval %d ", rc); if (rc < 0) { |
d2f7bf134
|
448 449 |
fprintf(stderr, "error sending tid/tgid cmd "); |
9e06d3f9f
|
450 451 |
goto done; } |
a3baf649c
|
452 |
} |
546040dc4
|
453 454 455 456 457 458 459 460 461 462 463 464 465 |
if (containerset) { cfd = open(containerpath, O_RDONLY); if (cfd < 0) { perror("error opening container file"); goto err; } rc = send_cmd(nl_sd, id, mypid, CGROUPSTATS_CMD_GET, CGROUPSTATS_CMD_ATTR_FD, &cfd, sizeof(__u32)); if (rc < 0) { perror("error sending cgroupstats command"); goto err; } } |
65a67bd26
|
466 467 468 469 |
if (!maskset && !tid && !containerset) { usage(); goto err; } |
546040dc4
|
470 |
|
9e06d3f9f
|
471 |
do { |
9e06d3f9f
|
472 473 474 |
rep_len = recv(nl_sd, &msg, sizeof(msg), 0); PRINTF("received %d bytes ", rep_len); |
a3baf649c
|
475 |
|
9e06d3f9f
|
476 |
if (rep_len < 0) { |
d2f7bf134
|
477 478 479 |
fprintf(stderr, "nonfatal reply error: errno %d ", errno); |
9e06d3f9f
|
480 481 482 483 |
continue; } if (msg.n.nlmsg_type == NLMSG_ERROR || !NLMSG_OK((&msg.n), rep_len)) { |
7d1bdca9b
|
484 |
struct nlmsgerr *err = NLMSG_DATA(&msg); |
d2f7bf134
|
485 486 487 |
fprintf(stderr, "fatal reply error, errno %d ", err->error); |
9e06d3f9f
|
488 489 |
goto done; } |
10e6f32bd
|
490 491 |
PRINTF("nlmsghdr size=%zu, nlmsg_len=%d, rep_len=%d ", |
9e06d3f9f
|
492 493 494 495 496 497 498 |
sizeof(struct nlmsghdr), msg.n.nlmsg_len, rep_len); rep_len = GENLMSG_PAYLOAD(&msg.n); na = (struct nlattr *) GENLMSG_DATA(&msg); len = 0; |
9e06d3f9f
|
499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 |
while (len < rep_len) { len += NLA_ALIGN(na->nla_len); switch (na->nla_type) { case TASKSTATS_TYPE_AGGR_TGID: /* Fall through */ case TASKSTATS_TYPE_AGGR_PID: aggr_len = NLA_PAYLOAD(na->nla_len); len2 = 0; /* For nested attributes, na follows */ na = (struct nlattr *) NLA_DATA(na); done = 0; while (len2 < aggr_len) { switch (na->nla_type) { case TASKSTATS_TYPE_PID: rtid = *(int *) NLA_DATA(na); if (print_delays) printf("PID\t%d ", rtid); break; case TASKSTATS_TYPE_TGID: rtid = *(int *) NLA_DATA(na); if (print_delays) printf("TGID\t%d ", rtid); break; case TASKSTATS_TYPE_STATS: count++; if (print_delays) print_delayacct((struct taskstats *) NLA_DATA(na)); |
cf709844d
|
528 529 |
if (print_io_accounting) print_ioacct((struct taskstats *) NLA_DATA(na)); |
b663a79c1
|
530 531 |
if (print_task_context_switch_counts) task_context_switch_counts((struct taskstats *) NLA_DATA(na)); |
9e06d3f9f
|
532 533 534 535 536 537 538 539 540 541 |
if (fd) { if (write(fd, NLA_DATA(na), na->nla_len) < 0) { err(1,"write error "); } } if (!loop) goto done; break; default: |
d2f7bf134
|
542 543 544 545 |
fprintf(stderr, "Unknown nested" " nla_type %d ", na->nla_type); |
9e06d3f9f
|
546 547 548 549 550 551 |
break; } len2 += NLA_ALIGN(na->nla_len); na = (struct nlattr *) ((char *) na + len2); } break; |
546040dc4
|
552 553 554 |
case CGROUPSTATS_TYPE_CGROUP_STATS: print_cgroupstats(NLA_DATA(na)); break; |
9e06d3f9f
|
555 |
default: |
d2f7bf134
|
556 557 558 |
fprintf(stderr, "Unknown nla_type %d ", na->nla_type); |
4be2c95d1
|
559 |
case TASKSTATS_TYPE_NULL: |
9e06d3f9f
|
560 |
break; |
a3baf649c
|
561 |
} |
9e06d3f9f
|
562 |
na = (struct nlattr *) (GENLMSG_DATA(&msg) + len); |
a3baf649c
|
563 |
} |
9e06d3f9f
|
564 565 566 567 568 |
} while (loop); done: if (maskset) { rc = send_cmd(nl_sd, id, mypid, TASKSTATS_CMD_GET, TASKSTATS_CMD_ATTR_DEREGISTER_CPUMASK, |
7d1bdca9b
|
569 |
&cpumask, strlen(cpumask) + 1); |
9e06d3f9f
|
570 571 572 573 574 |
printf("Sent deregister mask, retval %d ", rc); if (rc < 0) err(rc, "error sending deregister cpumask "); |
a3baf649c
|
575 |
} |
9e06d3f9f
|
576 577 578 579 |
err: close(nl_sd); if (fd) close(fd); |
546040dc4
|
580 581 |
if (cfd) close(cfd); |
9e06d3f9f
|
582 |
return 0; |
a3baf649c
|
583 |
} |