[linux-block.git] / tools / accounting / getdelays.c

// SPDX-License-Identifier: GPL-2.0
/* 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
 * Copyright (c) Jay Lan, SGI. 2006
 *
 * Compile with
 *	gcc -I/usr/src/linux/include getdelays.c -o getdelays
 */

#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>
#include <sys/wait.h>
#include <signal.h>

#include <linux/genetlink.h>
#include <linux/taskstats.h>
#include <linux/cgroupstats.h>

/*
 * 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)

#define err(code, fmt, arg...)			\
	do {					\
		fprintf(stderr, fmt, ##arg);	\
		exit(code);			\
	} while (0)

int rcvbufsz;
char name[100];
int dbg;
int print_delays;
int print_io_accounting;
int print_task_context_switch_counts;

#define PRINTF(fmt, arg...) {			\
	    if (dbg) {				\
		printf(fmt, ##arg);		\
	    }					\
	}

/* Maximum size of response requested or message sent */
#define MAX_MSG_SIZE	1024
/* Maximum number of cpus expected to be specified in a cpumask */
#define MAX_CPUS	32

struct msgtemplate {
	struct nlmsghdr n;
	struct genlmsghdr g;
	char buf[MAX_MSG_SIZE];
};

char cpumask[100+6*MAX_CPUS];

static void usage(void)
{
	fprintf(stderr, "getdelays [-dilv] [-w logfile] [-r bufsize] "
			"[-m cpumask] [-t tgid] [-p pid]\n");
	fprintf(stderr, "  -d: print delayacct stats\n");
	fprintf(stderr, "  -i: print IO accounting (works only with -p)\n");
	fprintf(stderr, "  -l: listen forever\n");
	fprintf(stderr, "  -v: debug on\n");
	fprintf(stderr, "  -C: container path\n");
}

/*
 * Create a raw netlink socket and bind
 */
static int create_nl_socket(int protocol)
{
	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) {
			fprintf(stderr, "Unable to set socket rcv buf size to %d\n",
				rcvbufsz);
			goto error;
		}

	memset(&local, 0, sizeof(local));
	local.nl_family = AF_NETLINK;

	if (bind(fd, (struct sockaddr *) &local, sizeof(local)) < 0)
		goto error;

	return fd;
error:
	close(fd);
	return -1;
}


static int send_cmd(int sd, __u16 nlmsg_type, __u32 nlmsg_pid,
	     __u8 genl_cmd, __u16 nla_type,
	     void *nla_data, int nla_len)
{
	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 + 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;
}


/*
 * Probe the controller in genetlink to find the family id
 * for the TASKSTATS family
 */
static int get_family_id(int sd)
{
	struct {
		struct nlmsghdr n;
		struct genlmsghdr g;
		char buf[256];
	} ans;

	int id = 0, rc;
	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);
	if (rc < 0)
		return 0;	/* sendto() failure? */

	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;

	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;
}

#define average_ms(t, c) (t / 1000000ULL / (c ? c : 1))

static void print_delayacct(struct taskstats *t)
{
	printf("\n\nCPU   %15s%15s%15s%15s%15s\n"
	       "      %15llu%15llu%15llu%15llu%15.3fms\n"
	       "IO    %15s%15s%15s\n"
	       "      %15llu%15llu%15.3fms\n"
	       "SWAP  %15s%15s%15s\n"
	       "      %15llu%15llu%15.3fms\n"
	       "RECLAIM  %12s%15s%15s\n"
	       "      %15llu%15llu%15.3fms\n"
	       "THRASHING%12s%15s%15s\n"
	       "      %15llu%15llu%15.3fms\n"
	       "COMPACT  %12s%15s%15s\n"
	       "      %15llu%15llu%15.3fms\n"
	       "WPCOPY   %12s%15s%15s\n"
	       "      %15llu%15llu%15.3fms\n"
	       "IRQ   %15s%15s%15s\n"
	       "      %15llu%15llu%15.3fms\n",
	       "count", "real total", "virtual total",
	       "delay total", "delay average",
	       (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,
	       average_ms((double)t->cpu_delay_total, t->cpu_count),
	       "count", "delay total", "delay average",
	       (unsigned long long)t->blkio_count,
	       (unsigned long long)t->blkio_delay_total,
	       average_ms((double)t->blkio_delay_total, t->blkio_count),
	       "count", "delay total", "delay average",
	       (unsigned long long)t->swapin_count,
	       (unsigned long long)t->swapin_delay_total,
	       average_ms((double)t->swapin_delay_total, t->swapin_count),
	       "count", "delay total", "delay average",
	       (unsigned long long)t->freepages_count,
	       (unsigned long long)t->freepages_delay_total,
	       average_ms((double)t->freepages_delay_total, t->freepages_count),
	       "count", "delay total", "delay average",
	       (unsigned long long)t->thrashing_count,
	       (unsigned long long)t->thrashing_delay_total,
	       average_ms((double)t->thrashing_delay_total, t->thrashing_count),
	       "count", "delay total", "delay average",
	       (unsigned long long)t->compact_count,
	       (unsigned long long)t->compact_delay_total,
	       average_ms((double)t->compact_delay_total, t->compact_count),
	       "count", "delay total", "delay average",
	       (unsigned long long)t->wpcopy_count,
	       (unsigned long long)t->wpcopy_delay_total,
	       average_ms((double)t->wpcopy_delay_total, t->wpcopy_count),
	       "count", "delay total", "delay average",
	       (unsigned long long)t->irq_count,
	       (unsigned long long)t->irq_delay_total,
	       average_ms((double)t->irq_delay_total, t->irq_count));
}

static void task_context_switch_counts(struct taskstats *t)
{
	printf("\n\nTask   %15s%15s\n"
	       "       %15llu%15llu\n",
	       "voluntary", "nonvoluntary",
	       (unsigned long long)t->nvcsw, (unsigned long long)t->nivcsw);
}

static void print_cgroupstats(struct cgroupstats *c)
{
	printf("sleeping %llu, blocked %llu, running %llu, stopped %llu, "
		"uninterruptible %llu\n", (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);
}


static void print_ioacct(struct taskstats *t)
{
	printf("%s: read=%llu, write=%llu, cancelled_write=%llu\n",
		t->ac_comm,
		(unsigned long long)t->read_bytes,
		(unsigned long long)t->write_bytes,
		(unsigned long long)t->cancelled_write_bytes);
}

int main(int argc, char *argv[])
{
	int c, rc, rep_len, aggr_len, len2;
	int cmd_type = TASKSTATS_CMD_ATTR_UNSPEC;
	__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 write_file = 0;
	int maskset = 0;
	char *logfile = NULL;
	int loop = 0;
	int containerset = 0;
	char *containerpath = NULL;
	int cfd = 0;
	int forking = 0;
	sigset_t sigset;

	struct msgtemplate msg;

	while (!forking) {
		c = getopt(argc, argv, "qdiw:r:m:t:p:vlC:c:");
		if (c < 0)
			break;

		switch (c) {
		case 'd':
			printf("print delayacct stats ON\n");
			print_delays = 1;
			break;
		case 'i':
			printf("printing IO accounting\n");
			print_io_accounting = 1;
			break;
		case 'q':
			printf("printing task/process context switch rates\n");
			print_task_context_switch_counts = 1;
			break;
		case 'C':
			containerset = 1;
			containerpath = optarg;
			break;
		case 'w':
			logfile = strdup(optarg);
			printf("write to file %s\n", logfile);
			write_file = 1;
			break;
		case 'r':
			rcvbufsz = atoi(optarg);
			printf("receive buf size %d\n", rcvbufsz);
			if (rcvbufsz < 0)
				err(1, "Invalid rcv buf size\n");
			break;
		case 'm':
			strncpy(cpumask, optarg, sizeof(cpumask));
			cpumask[sizeof(cpumask) - 1] = '\0';
			maskset = 1;
			printf("cpumask %s maskset %d\n", cpumask, maskset);
			break;
		case 't':
			tid = atoi(optarg);
			if (!tid)
				err(1, "Invalid tgid\n");
			cmd_type = TASKSTATS_CMD_ATTR_TGID;
			break;
		case 'p':
			tid = atoi(optarg);
			if (!tid)
				err(1, "Invalid pid\n");
			cmd_type = TASKSTATS_CMD_ATTR_PID;
			break;
		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\n");
			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;
		case 'v':
			printf("debug on\n");
			dbg = 1;
			break;
		case 'l':
			printf("listen forever\n");
			loop = 1;
			break;
		default:
			usage();
			exit(-1);
		}
	}

	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\n");
			exit(1);
		}
	}

	nl_sd = create_nl_socket(NETLINK_GENERIC);
	if (nl_sd < 0)
		err(1, "error creating Netlink socket\n");


	mypid = getpid();
	id = get_family_id(nl_sd);
	if (!id) {
		fprintf(stderr, "Error getting family id, errno %d\n", errno);
		goto err;
	}
	PRINTF("family id %d\n", id);

	if (maskset) {
		rc = send_cmd(nl_sd, id, mypid, TASKSTATS_CMD_GET,
			      TASKSTATS_CMD_ATTR_REGISTER_CPUMASK,
			      &cpumask, strlen(cpumask) + 1);
		PRINTF("Sent register cpumask, retval %d\n", rc);
		if (rc < 0) {
			fprintf(stderr, "error sending register cpumask\n");
			goto err;
		}
	}

	if (tid && containerset) {
		fprintf(stderr, "Select either -t or -C, not both\n");
		goto err;
	}

	/*
	 * 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);
	}

	if (tid) {
		rc = send_cmd(nl_sd, id, mypid, TASKSTATS_CMD_GET,
			      cmd_type, &tid, sizeof(__u32));
		PRINTF("Sent pid/tgid, retval %d\n", rc);
		if (rc < 0) {
			fprintf(stderr, "error sending tid/tgid cmd\n");
			goto done;
		}
	}

	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;
		}
	}
	if (!maskset && !tid && !containerset) {
		usage();
		goto err;
	}

	do {
		rep_len = recv(nl_sd, &msg, sizeof(msg), 0);
		PRINTF("received %d bytes\n", rep_len);

		if (rep_len < 0) {
			fprintf(stderr, "nonfatal reply error: errno %d\n",
				errno);
			continue;
		}
		if (msg.n.nlmsg_type == NLMSG_ERROR ||
		    !NLMSG_OK((&msg.n), rep_len)) {
			struct nlmsgerr *err = NLMSG_DATA(&msg);
			fprintf(stderr, "fatal reply error,  errno %d\n",
				err->error);
			goto done;
		}

		PRINTF("nlmsghdr size=%zu, nlmsg_len=%d, rep_len=%d\n",
		       sizeof(struct nlmsghdr), msg.n.nlmsg_len, rep_len);


		rep_len = GENLMSG_PAYLOAD(&msg.n);

		na = (struct nlattr *) GENLMSG_DATA(&msg);
		len = 0;
		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);
				while (len2 < aggr_len) {
					switch (na->nla_type) {
					case TASKSTATS_TYPE_PID:
						rtid = *(int *) NLA_DATA(na);
						if (print_delays)
							printf("PID\t%d\n", rtid);
						break;
					case TASKSTATS_TYPE_TGID:
						rtid = *(int *) NLA_DATA(na);
						if (print_delays)
							printf("TGID\t%d\n", rtid);
						break;
					case TASKSTATS_TYPE_STATS:
						if (print_delays)
							print_delayacct((struct taskstats *) NLA_DATA(na));
						if (print_io_accounting)
							print_ioacct((struct taskstats *) NLA_DATA(na));
						if (print_task_context_switch_counts)
							task_context_switch_counts((struct taskstats *) NLA_DATA(na));
						if (fd) {
							if (write(fd, NLA_DATA(na), na->nla_len) < 0) {
								err(1,"write error\n");
							}
						}
						if (!loop)
							goto done;
						break;
					case TASKSTATS_TYPE_NULL:
						break;
					default:
						fprintf(stderr, "Unknown nested"
							" nla_type %d\n",
							na->nla_type);
						break;
					}
					len2 += NLA_ALIGN(na->nla_len);
					na = (struct nlattr *)((char *)na +
							       NLA_ALIGN(na->nla_len));
				}
				break;

			case CGROUPSTATS_TYPE_CGROUP_STATS:
				print_cgroupstats(NLA_DATA(na));
				break;
			default:
				fprintf(stderr, "Unknown nla_type %d\n",
					na->nla_type);
			case TASKSTATS_TYPE_NULL:
				break;
			}
			na = (struct nlattr *) (GENLMSG_DATA(&msg) + len);
		}
	} while (loop);
done:
	if (maskset) {
		rc = send_cmd(nl_sd, id, mypid, TASKSTATS_CMD_GET,
			      TASKSTATS_CMD_ATTR_DEREGISTER_CPUMASK,
			      &cpumask, strlen(cpumask) + 1);
		printf("Sent deregister mask, retval %d\n", rc);
		if (rc < 0)
			err(rc, "error sending deregister cpumask\n");
	}
err:
	close(nl_sd);
	if (fd)
		close(fd);
	if (cfd)
		close(cfd);
	return 0;
}
Commit	Line	Data
b2441318	1	// SPDX-License-Identifier: GPL-2.0
a3baf649 SN	2	/* getdelays.c
	3	*
	4	* Utility to get per-pid and per-tgid delay accounting statistics
	5	* Also illustrates usage of the taskstats interface
	6	*
	7	* Copyright (C) Shailabh Nagar, IBM Corp. 2005
	8	* Copyright (C) Balbir Singh, IBM Corp. 2006
9e06d3f9	9	* Copyright (c) Jay Lan, SGI. 2006
a3baf649	10	*
d2f7bf13 AM	11	* Compile with
d2f7bf13 AM	12	* gcc -I/usr/src/linux/include getdelays.c -o getdelays
a3baf649 SN	13	*/
	14
	15	#include <stdio.h>
	16	#include <stdlib.h>
	17	#include <errno.h>
	18	#include <unistd.h>
	19	#include <poll.h>
	20	#include <string.h>
	21	#include <fcntl.h>
	22	#include <sys/types.h>
	23	#include <sys/stat.h>
	24	#include <sys/socket.h>
db9e5679	25	#include <sys/wait.h>
a3baf649 SN	26	#include <signal.h>
	27
	28	#include <linux/genetlink.h>
	29	#include <linux/taskstats.h>
546040dc	30	#include <linux/cgroupstats.h>
a3baf649 SN	31
	32	/*
	33	* Generic macros for dealing with netlink sockets. Might be duplicated
	34	* elsewhere. It is recommended that commercial grade applications use
	35	* libnl or libnetlink and use the interfaces provided by the library
	36	*/
	37	#define GENLMSG_DATA(glh) ((void *)(NLMSG_DATA(glh) + GENL_HDRLEN))
	38	#define GENLMSG_PAYLOAD(glh) (NLMSG_PAYLOAD(glh, 0) - GENL_HDRLEN)
	39	#define NLA_DATA(na) ((void )((char)(na) + NLA_HDRLEN))
	40	#define NLA_PAYLOAD(len) (len - NLA_HDRLEN)
	41
d2f7bf13 AM	42	#define err(code, fmt, arg...) \
	43	do { \
	44	fprintf(stderr, fmt, ##arg); \
	45	exit(code); \
	46	} while (0)
	47
d2f7bf13 AM	48	int rcvbufsz;
	49	char name[100];
	50	int dbg;
	51	int print_delays;
cf709844	52	int print_io_accounting;
b663a79c	53	int print_task_context_switch_counts;
d2f7bf13	54
9e06d3f9 SN	55	#define PRINTF(fmt, arg...) { \
	56	if (dbg) { \
	57	printf(fmt, ##arg); \
	58	} \
	59	}
	60
	61	/* Maximum size of response requested or message sent */
88040230	62	#define MAX_MSG_SIZE 1024
9e06d3f9 SN	63	/* Maximum number of cpus expected to be specified in a cpumask */
9e06d3f9 SN	64	#define MAX_CPUS 32
9e06d3f9 SN	65
	66	struct msgtemplate {
	67	struct nlmsghdr n;
	68	struct genlmsghdr g;
	69	char buf[MAX_MSG_SIZE];
	70	};
	71
	72	char cpumask[100+6*MAX_CPUS];
a3baf649	73
f16825bb RD	74	static void usage(void)
	75	{
	76	fprintf(stderr, "getdelays [-dilv] [-w logfile] [-r bufsize] "
	77	"[-m cpumask] [-t tgid] [-p pid]\n");
	78	fprintf(stderr, " -d: print delayacct stats\n");
	79	fprintf(stderr, " -i: print IO accounting (works only with -p)\n");
	80	fprintf(stderr, " -l: listen forever\n");
	81	fprintf(stderr, " -v: debug on\n");
546040dc	82	fprintf(stderr, " -C: container path\n");
f16825bb RD	83	}
f16825bb RD	84
a3baf649 SN	85	/*
	86	* Create a raw netlink socket and bind
	87	*/
9e06d3f9	88	static int create_nl_socket(int protocol)
a3baf649	89	{
9e06d3f9 SN	90	int fd;
	91	struct sockaddr_nl local;
	92
	93	fd = socket(AF_NETLINK, SOCK_RAW, protocol);
	94	if (fd < 0)
	95	return -1;
	96
	97	if (rcvbufsz)
	98	if (setsockopt(fd, SOL_SOCKET, SO_RCVBUF,
	99	&rcvbufsz, sizeof(rcvbufsz)) < 0) {
0de4b954	100	fprintf(stderr, "Unable to set socket rcv buf size to %d\n",
d2f7bf13	101	rcvbufsz);
0de4b954	102	goto error;
9e06d3f9	103	}
a3baf649	104
9e06d3f9 SN	105	memset(&local, 0, sizeof(local));
9e06d3f9 SN	106	local.nl_family = AF_NETLINK;
a3baf649	107
9e06d3f9 SN	108	if (bind(fd, (struct sockaddr *) &local, sizeof(local)) < 0)
9e06d3f9 SN	109	goto error;
a3baf649	110
9e06d3f9 SN	111	return fd;
	112	error:
	113	close(fd);
	114	return -1;
a3baf649 SN	115	}
a3baf649 SN	116
9e06d3f9	117
b7ed698c	118	static int send_cmd(int sd, __u16 nlmsg_type, __u32 nlmsg_pid,
9e06d3f9 SN	119	__u8 genl_cmd, __u16 nla_type,
9e06d3f9 SN	120	void *nla_data, int nla_len)
a3baf649	121	{
9e06d3f9 SN	122	struct nlattr *na;
	123	struct sockaddr_nl nladdr;
	124	int r, buflen;
	125	char *buf;
	126
	127	struct msgtemplate msg;
	128
	129	msg.n.nlmsg_len = NLMSG_LENGTH(GENL_HDRLEN);
	130	msg.n.nlmsg_type = nlmsg_type;
	131	msg.n.nlmsg_flags = NLM_F_REQUEST;
	132	msg.n.nlmsg_seq = 0;
	133	msg.n.nlmsg_pid = nlmsg_pid;
	134	msg.g.cmd = genl_cmd;
	135	msg.g.version = 0x1;
	136	na = (struct nlattr *) GENLMSG_DATA(&msg);
	137	na->nla_type = nla_type;
4054ab64	138	na->nla_len = nla_len + NLA_HDRLEN;
9e06d3f9 SN	139	memcpy(NLA_DATA(na), nla_data, nla_len);
	140	msg.n.nlmsg_len += NLMSG_ALIGN(na->nla_len);
	141
	142	buf = (char *) &msg;
	143	buflen = msg.n.nlmsg_len ;
	144	memset(&nladdr, 0, sizeof(nladdr));
	145	nladdr.nl_family = AF_NETLINK;
	146	while ((r = sendto(sd, buf, buflen, 0, (struct sockaddr *) &nladdr,
	147	sizeof(nladdr))) < buflen) {
	148	if (r > 0) {
	149	buf += r;
	150	buflen -= r;
	151	} else if (errno != EAGAIN)
	152	return -1;
	153	}
	154	return 0;
a3baf649 SN	155	}
a3baf649 SN	156
9e06d3f9	157
a3baf649 SN	158	/*
	159	* Probe the controller in genetlink to find the family id
	160	* for the TASKSTATS family
	161	*/
b7ed698c	162	static int get_family_id(int sd)
a3baf649	163	{
9e06d3f9 SN	164	struct {
	165	struct nlmsghdr n;
	166	struct genlmsghdr g;
	167	char buf[256];
	168	} ans;
	169
10e6f32b	170	int id = 0, rc;
9e06d3f9 SN	171	struct nlattr *na;
	172	int rep_len;
	173
	174	strcpy(name, TASKSTATS_GENL_NAME);
	175	rc = send_cmd(sd, GENL_ID_CTRL, getpid(), CTRL_CMD_GETFAMILY,
	176	CTRL_ATTR_FAMILY_NAME, (void *)name,
	177	strlen(TASKSTATS_GENL_NAME)+1);
4ed960b1 AM	178	if (rc < 0)
4ed960b1 AM	179	return 0; /* sendto() failure? */
9e06d3f9 SN	180
	181	rep_len = recv(sd, &ans, sizeof(ans), 0);
	182	if (ans.n.nlmsg_type == NLMSG_ERROR \|\|
	183	(rep_len < 0) \|\| !NLMSG_OK((&ans.n), rep_len))
	184	return 0;
a3baf649	185
9e06d3f9 SN	186	na = (struct nlattr *) GENLMSG_DATA(&ans);
	187	na = (struct nlattr ) ((char ) na + NLA_ALIGN(na->nla_len));
	188	if (na->nla_type == CTRL_ATTR_FAMILY_ID) {
	189	id = (__u16 ) NLA_DATA(na);
	190	}
	191	return id;
a3baf649 SN	192	}
a3baf649 SN	193
02d54f09 WF	194	#define average_ms(t, c) (t / 1000000ULL / (c ? c : 1))
02d54f09 WF	195
b7ed698c	196	static void print_delayacct(struct taskstats *t)
a3baf649	197	{
02d54f09 WF	198	printf("\n\nCPU %15s%15s%15s%15s%15s\n"
	199	" %15llu%15llu%15llu%15llu%15.3fms\n"
	200	"IO %15s%15s%15s\n"
a3b2aeac	201	" %15llu%15llu%15.3fms\n"
02d54f09	202	"SWAP %15s%15s%15s\n"
a3b2aeac	203	" %15llu%15llu%15.3fms\n"
02d54f09	204	"RECLAIM %12s%15s%15s\n"
a3b2aeac	205	" %15llu%15llu%15.3fms\n"
b1d29ba8	206	"THRASHING%12s%15s%15s\n"
a3b2aeac	207	" %15llu%15llu%15.3fms\n"
5bf18281	208	"COMPACT %12s%15s%15s\n"
a3b2aeac	209	" %15llu%15llu%15.3fms\n"
662ce1dc	210	"WPCOPY %12s%15s%15s\n"
a3b2aeac YY	211	" %15llu%15llu%15.3fms\n"
	212	"IRQ %15s%15s%15s\n"
	213	" %15llu%15llu%15.3fms\n",
02d54f09 WF	214	"count", "real total", "virtual total",
02d54f09 WF	215	"delay total", "delay average",
66659313 RD	216	(unsigned long long)t->cpu_count,
	217	(unsigned long long)t->cpu_run_real_total,
	218	(unsigned long long)t->cpu_run_virtual_total,
	219	(unsigned long long)t->cpu_delay_total,
02d54f09 WF	220	average_ms((double)t->cpu_delay_total, t->cpu_count),
02d54f09 WF	221	"count", "delay total", "delay average",
66659313 RD	222	(unsigned long long)t->blkio_count,
66659313 RD	223	(unsigned long long)t->blkio_delay_total,
a3b2aeac	224	average_ms((double)t->blkio_delay_total, t->blkio_count),
02d54f09	225	"count", "delay total", "delay average",
66659313 RD	226	(unsigned long long)t->swapin_count,
66659313 RD	227	(unsigned long long)t->swapin_delay_total,
a3b2aeac	228	average_ms((double)t->swapin_delay_total, t->swapin_count),
02d54f09	229	"count", "delay total", "delay average",
66659313	230	(unsigned long long)t->freepages_count,
02d54f09	231	(unsigned long long)t->freepages_delay_total,
a3b2aeac	232	average_ms((double)t->freepages_delay_total, t->freepages_count),
b1d29ba8 JW	233	"count", "delay total", "delay average",
	234	(unsigned long long)t->thrashing_count,
	235	(unsigned long long)t->thrashing_delay_total,
a3b2aeac	236	average_ms((double)t->thrashing_delay_total, t->thrashing_count),
5bf18281	237	"count", "delay total", "delay average",
	238	(unsigned long long)t->compact_count,
	239	(unsigned long long)t->compact_delay_total,
a3b2aeac	240	average_ms((double)t->compact_delay_total, t->compact_count),
662ce1dc YY	241	"count", "delay total", "delay average",
	242	(unsigned long long)t->wpcopy_count,
	243	(unsigned long long)t->wpcopy_delay_total,
a3b2aeac YY	244	average_ms((double)t->wpcopy_delay_total, t->wpcopy_count),
	245	"count", "delay total", "delay average",
	246	(unsigned long long)t->irq_count,
	247	(unsigned long long)t->irq_delay_total,
	248	average_ms((double)t->irq_delay_total, t->irq_count));
a3baf649 SN	249	}
a3baf649 SN	250
b7ed698c	251	static void task_context_switch_counts(struct taskstats *t)
b663a79c MU	252	{
b663a79c MU	253	printf("\n\nTask %15s%15s\n"
10e6f32b	254	" %15llu%15llu\n",
b663a79c	255	"voluntary", "nonvoluntary",
66659313	256	(unsigned long long)t->nvcsw, (unsigned long long)t->nivcsw);
b663a79c MU	257	}
b663a79c MU	258
b7ed698c	259	static void print_cgroupstats(struct cgroupstats *c)
546040dc BS	260	{
546040dc BS	261	printf("sleeping %llu, blocked %llu, running %llu, stopped %llu, "
66659313 RD	262	"uninterruptible %llu\n", (unsigned long long)c->nr_sleeping,
	263	(unsigned long long)c->nr_io_wait,
	264	(unsigned long long)c->nr_running,
	265	(unsigned long long)c->nr_stopped,
	266	(unsigned long long)c->nr_uninterruptible);
546040dc BS	267	}
	268
	269
b7ed698c	270	static void print_ioacct(struct taskstats *t)
cf709844 AM	271	{
	272	printf("%s: read=%llu, write=%llu, cancelled_write=%llu\n",
	273	t->ac_comm,
	274	(unsigned long long)t->read_bytes,
	275	(unsigned long long)t->write_bytes,
	276	(unsigned long long)t->cancelled_write_bytes);
	277	}
	278
a3baf649 SN	279	int main(int argc, char *argv[])
a3baf649 SN	280	{
b8d9a865 JSR	281	int c, rc, rep_len, aggr_len, len2;
b8d9a865 JSR	282	int cmd_type = TASKSTATS_CMD_ATTR_UNSPEC;
9e06d3f9 SN	283	__u16 id;
	284	__u32 mypid;
	285
	286	struct nlattr *na;
	287	int nl_sd = -1;
	288	int len = 0;
	289	pid_t tid = 0;
	290	pid_t rtid = 0;
	291
	292	int fd = 0;
9e06d3f9 SN	293	int write_file = 0;
9e06d3f9 SN	294	int maskset = 0;
7f76c403	295	char *logfile = NULL;
9e06d3f9	296	int loop = 0;
546040dc	297	int containerset = 0;
8da01af4	298	char *containerpath = NULL;
546040dc	299	int cfd = 0;
db9e5679 MG	300	int forking = 0;
db9e5679 MG	301	sigset_t sigset;
9e06d3f9 SN	302
	303	struct msgtemplate msg;
	304
db9e5679 MG	305	while (!forking) {
db9e5679 MG	306	c = getopt(argc, argv, "qdiw:r:m:t:p:vlC:c:");
9e06d3f9 SN	307	if (c < 0)
9e06d3f9 SN	308	break;
a3baf649	309
9e06d3f9 SN	310	switch (c) {
	311	case 'd':
	312	printf("print delayacct stats ON\n");
	313	print_delays = 1;
	314	break;
cf709844 AM	315	case 'i':
	316	printf("printing IO accounting\n");
	317	print_io_accounting = 1;
	318	break;
b663a79c MU	319	case 'q':
	320	printf("printing task/process context switch rates\n");
	321	print_task_context_switch_counts = 1;
	322	break;
546040dc BS	323	case 'C':
546040dc BS	324	containerset = 1;
8da01af4	325	containerpath = optarg;
546040dc	326	break;
9e06d3f9	327	case 'w':
7f76c403	328	logfile = strdup(optarg);
9e06d3f9 SN	329	printf("write to file %s\n", logfile);
	330	write_file = 1;
	331	break;
	332	case 'r':
	333	rcvbufsz = atoi(optarg);
	334	printf("receive buf size %d\n", rcvbufsz);
	335	if (rcvbufsz < 0)
	336	err(1, "Invalid rcv buf size\n");
	337	break;
	338	case 'm':
	339	strncpy(cpumask, optarg, sizeof(cpumask));
88e15ce4	340	cpumask[sizeof(cpumask) - 1] = '\0';
9e06d3f9 SN	341	maskset = 1;
	342	printf("cpumask %s maskset %d\n", cpumask, maskset);
	343	break;
	344	case 't':
	345	tid = atoi(optarg);
	346	if (!tid)
	347	err(1, "Invalid tgid\n");
	348	cmd_type = TASKSTATS_CMD_ATTR_TGID;
9e06d3f9 SN	349	break;
	350	case 'p':
	351	tid = atoi(optarg);
	352	if (!tid)
	353	err(1, "Invalid pid\n");
	354	cmd_type = TASKSTATS_CMD_ATTR_PID;
9e06d3f9	355	break;
db9e5679 MG	356	case 'c':
	357
	358	/* Block SIGCHLD for sigwait() later */
	359	if (sigemptyset(&sigset) == -1)
	360	err(1, "Failed to empty sigset");
	361	if (sigaddset(&sigset, SIGCHLD))
	362	err(1, "Failed to set sigchld in sigset");
	363	sigprocmask(SIG_BLOCK, &sigset, NULL);
	364
	365	/* fork/exec a child */
	366	tid = fork();
	367	if (tid < 0)
	368	err(1, "Fork failed\n");
	369	if (tid == 0)
	370	if (execvp(argv[optind - 1],
	371	&argv[optind - 1]) < 0)
	372	exit(-1);
	373
	374	/* Set the command type and avoid further processing */
	375	cmd_type = TASKSTATS_CMD_ATTR_PID;
	376	forking = 1;
	377	break;
9e06d3f9 SN	378	case 'v':
	379	printf("debug on\n");
	380	dbg = 1;
	381	break;
	382	case 'l':
	383	printf("listen forever\n");
	384	loop = 1;
	385	break;
	386	default:
f16825bb	387	usage();
9e06d3f9	388	exit(-1);
a3baf649	389	}
a3baf649	390	}
a3baf649	391
9e06d3f9 SN	392	if (write_file) {
	393	fd = open(logfile, O_WRONLY \| O_CREAT \| O_TRUNC,
	394	S_IRUSR \| S_IWUSR \| S_IRGRP \| S_IROTH);
	395	if (fd == -1) {
	396	perror("Cannot open output file\n");
	397	exit(1);
	398	}
	399	}
a3baf649	400
563c17cb ME	401	nl_sd = create_nl_socket(NETLINK_GENERIC);
563c17cb ME	402	if (nl_sd < 0)
9e06d3f9	403	err(1, "error creating Netlink socket\n");
a3baf649	404
a3baf649	405
9e06d3f9 SN	406	mypid = getpid();
	407	id = get_family_id(nl_sd);
	408	if (!id) {
d2f7bf13	409	fprintf(stderr, "Error getting family id, errno %d\n", errno);
9e06d3f9	410	goto err;
a3baf649	411	}
9e06d3f9 SN	412	PRINTF("family id %d\n", id);
	413
	414	if (maskset) {
	415	rc = send_cmd(nl_sd, id, mypid, TASKSTATS_CMD_GET,
	416	TASKSTATS_CMD_ATTR_REGISTER_CPUMASK,
7d1bdca9	417	&cpumask, strlen(cpumask) + 1);
9e06d3f9 SN	418	PRINTF("Sent register cpumask, retval %d\n", rc);
9e06d3f9 SN	419	if (rc < 0) {
d2f7bf13	420	fprintf(stderr, "error sending register cpumask\n");
9e06d3f9 SN	421	goto err;
9e06d3f9 SN	422	}
a3baf649 SN	423	}
a3baf649 SN	424
546040dc BS	425	if (tid && containerset) {
	426	fprintf(stderr, "Select either -t or -C, not both\n");
	427	goto err;
	428	}
	429
db9e5679 MG	430	/*
	431	* If we forked a child, wait for it to exit. Cannot use waitpid()
	432	* as all the delicious data would be reaped as part of the wait
	433	*/
	434	if (tid && forking) {
	435	int sig_received;
	436	sigwait(&sigset, &sig_received);
	437	}
	438
9e06d3f9 SN	439	if (tid) {
	440	rc = send_cmd(nl_sd, id, mypid, TASKSTATS_CMD_GET,
	441	cmd_type, &tid, sizeof(__u32));
	442	PRINTF("Sent pid/tgid, retval %d\n", rc);
	443	if (rc < 0) {
d2f7bf13	444	fprintf(stderr, "error sending tid/tgid cmd\n");
9e06d3f9 SN	445	goto done;
9e06d3f9 SN	446	}
a3baf649 SN	447	}
a3baf649 SN	448
546040dc BS	449	if (containerset) {
	450	cfd = open(containerpath, O_RDONLY);
	451	if (cfd < 0) {
	452	perror("error opening container file");
	453	goto err;
	454	}
	455	rc = send_cmd(nl_sd, id, mypid, CGROUPSTATS_CMD_GET,
	456	CGROUPSTATS_CMD_ATTR_FD, &cfd, sizeof(__u32));
	457	if (rc < 0) {
	458	perror("error sending cgroupstats command");
	459	goto err;
	460	}
	461	}
65a67bd2 MM	462	if (!maskset && !tid && !containerset) {
	463	usage();
	464	goto err;
	465	}
546040dc	466
9e06d3f9	467	do {
9e06d3f9 SN	468	rep_len = recv(nl_sd, &msg, sizeof(msg), 0);
9e06d3f9 SN	469	PRINTF("received %d bytes\n", rep_len);
a3baf649	470
9e06d3f9	471	if (rep_len < 0) {
d2f7bf13 AM	472	fprintf(stderr, "nonfatal reply error: errno %d\n",
d2f7bf13 AM	473	errno);
9e06d3f9 SN	474	continue;
	475	}
	476	if (msg.n.nlmsg_type == NLMSG_ERROR \|\|
	477	!NLMSG_OK((&msg.n), rep_len)) {
7d1bdca9	478	struct nlmsgerr *err = NLMSG_DATA(&msg);
d2f7bf13 AM	479	fprintf(stderr, "fatal reply error, errno %d\n",
d2f7bf13 AM	480	err->error);
9e06d3f9 SN	481	goto done;
	482	}
	483
10e6f32b	484	PRINTF("nlmsghdr size=%zu, nlmsg_len=%d, rep_len=%d\n",
9e06d3f9 SN	485	sizeof(struct nlmsghdr), msg.n.nlmsg_len, rep_len);
	486
	487
	488	rep_len = GENLMSG_PAYLOAD(&msg.n);
	489
	490	na = (struct nlattr *) GENLMSG_DATA(&msg);
	491	len = 0;
9e06d3f9 SN	492	while (len < rep_len) {
	493	len += NLA_ALIGN(na->nla_len);
	494	switch (na->nla_type) {
	495	case TASKSTATS_TYPE_AGGR_TGID:
	496	/* Fall through */
	497	case TASKSTATS_TYPE_AGGR_PID:
	498	aggr_len = NLA_PAYLOAD(na->nla_len);
	499	len2 = 0;
	500	/* For nested attributes, na follows */
	501	na = (struct nlattr *) NLA_DATA(na);
9e06d3f9 SN	502	while (len2 < aggr_len) {
	503	switch (na->nla_type) {
	504	case TASKSTATS_TYPE_PID:
	505	rtid = (int ) NLA_DATA(na);
	506	if (print_delays)
	507	printf("PID\t%d\n", rtid);
	508	break;
	509	case TASKSTATS_TYPE_TGID:
	510	rtid = (int ) NLA_DATA(na);
	511	if (print_delays)
	512	printf("TGID\t%d\n", rtid);
	513	break;
	514	case TASKSTATS_TYPE_STATS:
9e06d3f9 SN	515	if (print_delays)
9e06d3f9 SN	516	print_delayacct((struct taskstats *) NLA_DATA(na));
cf709844 AM	517	if (print_io_accounting)
cf709844 AM	518	print_ioacct((struct taskstats *) NLA_DATA(na));
b663a79c MU	519	if (print_task_context_switch_counts)
b663a79c MU	520	task_context_switch_counts((struct taskstats *) NLA_DATA(na));
9e06d3f9 SN	521	if (fd) {
	522	if (write(fd, NLA_DATA(na), na->nla_len) < 0) {
	523	err(1,"write error\n");
	524	}
	525	}
	526	if (!loop)
	527	goto done;
	528	break;
570d8e93 ND	529	case TASKSTATS_TYPE_NULL:
570d8e93 ND	530	break;
9e06d3f9	531	default:
d2f7bf13 AM	532	fprintf(stderr, "Unknown nested"
	533	" nla_type %d\n",
	534	na->nla_type);
9e06d3f9 SN	535	break;
	536	}
	537	len2 += NLA_ALIGN(na->nla_len);
570d8e93 ND	538	na = (struct nlattr )((char )na +
570d8e93 ND	539	NLA_ALIGN(na->nla_len));
9e06d3f9 SN	540	}
	541	break;
	542
546040dc BS	543	case CGROUPSTATS_TYPE_CGROUP_STATS:
	544	print_cgroupstats(NLA_DATA(na));
	545	break;
9e06d3f9	546	default:
d2f7bf13 AM	547	fprintf(stderr, "Unknown nla_type %d\n",
d2f7bf13 AM	548	na->nla_type);
4be2c95d	549	case TASKSTATS_TYPE_NULL:
9e06d3f9	550	break;
a3baf649	551	}
9e06d3f9	552	na = (struct nlattr *) (GENLMSG_DATA(&msg) + len);
a3baf649	553	}
9e06d3f9 SN	554	} while (loop);
	555	done:
	556	if (maskset) {
	557	rc = send_cmd(nl_sd, id, mypid, TASKSTATS_CMD_GET,
	558	TASKSTATS_CMD_ATTR_DEREGISTER_CPUMASK,
7d1bdca9	559	&cpumask, strlen(cpumask) + 1);
9e06d3f9 SN	560	printf("Sent deregister mask, retval %d\n", rc);
	561	if (rc < 0)
	562	err(rc, "error sending deregister cpumask\n");
a3baf649	563	}
9e06d3f9 SN	564	err:
	565	close(nl_sd);
	566	if (fd)
	567	close(fd);
546040dc BS	568	if (cfd)
546040dc BS	569	close(cfd);
9e06d3f9	570	return 0;
a3baf649	571	}