stat/common/acquire.c

/*
 * CDDL HEADER START
 *
 * The contents of this file are subject to the terms of the
 * Common Development and Distribution License (the "License").
 * You may not use this file except in compliance with the License.
 *
 * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
 * or http://www.opensolaris.org/os/licensing.
 * See the License for the specific language governing permissions
 * and limitations under the License.
 *
 * When distributing Covered Code, include this CDDL HEADER in each
 * file and include the License file at usr/src/OPENSOLARIS.LICENSE.
 * If applicable, add the following below this CDDL HEADER, with the
 * fields enclosed by brackets "[]" replaced with your own identifying
 * information: Portions Copyright [yyyy] [name of copyright owner]
 *
 * CDDL HEADER END
 */
/*
 * Copyright 2007 Sun Microsystems, Inc.  All rights reserved.
 * Use is subject to license terms.
 */

#pragma ident	"%Z%%M%	%I%	%E% SMI"

#include "statcommon.h"
#include "dsr.h"

#include <stdlib.h>
#include <unistd.h>
#include <strings.h>
#include <errno.h>
#include <limits.h>
#include <poll.h>

#define	ARRAY_SIZE(a)	(sizeof (a) / sizeof (*a))

/*
 * The time we delay before retrying after an allocation
 * failure, in milliseconds
 */
#define	RETRY_DELAY 200

static char *cpu_states[] = {
	"cpu_ticks_idle",
	"cpu_ticks_user",
	"cpu_ticks_kernel",
	"cpu_ticks_wait"
};

static kstat_t *
kstat_lookup_read(kstat_ctl_t *kc, char *module,
		int instance, char *name)
{
	kstat_t *ksp = kstat_lookup(kc, module, instance, name);
	if (ksp == NULL)
		return (NULL);
	if (kstat_read(kc, ksp, NULL) == -1)
		return (NULL);
	return (ksp);
}

/*
 * Note: the following helpers do not clean up on the failure case,
 * because it is left to the free_snapshot() in the acquire_snapshot()
 * failure path.
 */

static int
acquire_cpus(struct snapshot *ss, kstat_ctl_t *kc)
{
	size_t i;

	ss->s_nr_cpus = sysconf(_SC_CPUID_MAX) + 1;
	ss->s_cpus = calloc(ss->s_nr_cpus, sizeof (struct cpu_snapshot));
	if (ss->s_cpus == NULL)
		goto out;

	for (i = 0; i < ss->s_nr_cpus; i++) {
		kstat_t *ksp;

		ss->s_cpus[i].cs_id = ID_NO_CPU;
		ss->s_cpus[i].cs_state = p_online(i, P_STATUS);
		/* If no valid CPU is present, move on to the next one */
		if (ss->s_cpus[i].cs_state == -1)
			continue;
		ss->s_cpus[i].cs_id = i;

		if ((ksp = kstat_lookup_read(kc, "cpu_info", i, NULL)) == NULL)
			goto out;

		(void) pset_assign(PS_QUERY, i, &ss->s_cpus[i].cs_pset_id);
		if (ss->s_cpus[i].cs_pset_id == PS_NONE)
			ss->s_cpus[i].cs_pset_id = ID_NO_PSET;

		if (!CPU_ACTIVE(&ss->s_cpus[i]))
			continue;

		if ((ksp = kstat_lookup_read(kc, "cpu", i, "vm")) == NULL)
			goto out;

		if (kstat_copy(ksp, &ss->s_cpus[i].cs_vm))
			goto out;

		if ((ksp = kstat_lookup_read(kc, "cpu", i, "sys")) == NULL)
			goto out;

		if (kstat_copy(ksp, &ss->s_cpus[i].cs_sys))
			goto out;
	}

	errno = 0;
out:
	return (errno);
}

static int
acquire_psets(struct snapshot *ss)
{
	psetid_t *pids = NULL;
	struct pset_snapshot *ps;
	size_t pids_nr;
	size_t i, j;

	/*
	 * Careful in this code. We have to use pset_list
	 * twice, but inbetween pids_nr can change at will.
	 * We delay the setting of s_nr_psets until we have
	 * the "final" value of pids_nr.
	 */

	if (pset_list(NULL, &pids_nr) < 0)
		return (errno);

	if ((pids = calloc(pids_nr, sizeof (psetid_t))) == NULL)
		goto out;

	if (pset_list(pids, &pids_nr) < 0)
		goto out;

	ss->s_psets = calloc(pids_nr + 1, sizeof (struct pset_snapshot));
	if (ss->s_psets == NULL)
		goto out;
	ss->s_nr_psets = pids_nr + 1;

	/* CPUs not in any actual pset */
	ps = &ss->s_psets[0];
	ps->ps_id = 0;
	ps->ps_cpus = calloc(ss->s_nr_cpus, sizeof (struct cpu_snapshot *));
	if (ps->ps_cpus == NULL)
		goto out;

	/* CPUs in a a pset */
	for (i = 1; i < ss->s_nr_psets; i++) {
		ps = &ss->s_psets[i];

		ps->ps_id = pids[i - 1];
		ps->ps_cpus =
			calloc(ss->s_nr_cpus, sizeof (struct cpu_snapshot *));
		if (ps->ps_cpus == NULL)
			goto out;
	}

	for (i = 0; i < ss->s_nr_psets; i++) {
		ps = &ss->s_psets[i];

		for (j = 0; j < ss->s_nr_cpus; j++) {
			if (!CPU_ACTIVE(&ss->s_cpus[j]))
				continue;
			if (ss->s_cpus[j].cs_pset_id != ps->ps_id)
				continue;

			ps->ps_cpus[ps->ps_nr_cpus++] = &ss->s_cpus[j];
		}
	}

	errno = 0;
out:
	free(pids);
	return (errno);
}

static int
acquire_intrs(struct snapshot *ss, kstat_ctl_t *kc)
{
	kstat_t *ksp;
	size_t i = 0;
	kstat_t *sys_misc;
	kstat_named_t *clock;

	/* clock interrupt */
	ss->s_nr_intrs = 1;

	for (ksp = kc->kc_chain; ksp; ksp = ksp->ks_next) {
		if (ksp->ks_type == KSTAT_TYPE_INTR)
			ss->s_nr_intrs++;
	}

	ss->s_intrs = calloc(ss->s_nr_intrs, sizeof (struct intr_snapshot));
	if (ss->s_intrs == NULL)
		return (errno);

	sys_misc = kstat_lookup_read(kc, "unix", 0, "system_misc");
	if (sys_misc == NULL)
		goto out;

	clock = (kstat_named_t *)kstat_data_lookup(sys_misc, "clk_intr");
	if (clock == NULL)
		goto out;

	(void) strlcpy(ss->s_intrs[0].is_name, "clock", KSTAT_STRLEN);
	ss->s_intrs[0].is_total = clock->value.ui32;

	i = 1;

	for (ksp = kc->kc_chain; ksp; ksp = ksp->ks_next) {
		kstat_intr_t *ki;
		int j;

		if (ksp->ks_type != KSTAT_TYPE_INTR)
			continue;
		if (kstat_read(kc, ksp, NULL) == -1)
			goto out;

		ki = KSTAT_INTR_PTR(ksp);

		(void) strlcpy(ss->s_intrs[i].is_name, ksp->ks_name,
			KSTAT_STRLEN);
		ss->s_intrs[i].is_total = 0;

		for (j = 0; j < KSTAT_NUM_INTRS; j++)
			ss->s_intrs[i].is_total += ki->intrs[j];

		i++;
	}

	errno = 0;
out:
	return (errno);
}

int
acquire_sys(struct snapshot *ss, kstat_ctl_t *kc)
{
	size_t i;
	kstat_named_t *knp;
	kstat_t *ksp;

	if ((ksp = kstat_lookup(kc, "unix", 0, "sysinfo")) == NULL)
		return (errno);

	if (kstat_read(kc, ksp, &ss->s_sys.ss_sysinfo) == -1)
		return (errno);

	if ((ksp = kstat_lookup(kc, "unix", 0, "vminfo")) == NULL)
		return (errno);

	if (kstat_read(kc, ksp, &ss->s_sys.ss_vminfo) == -1)
		return (errno);

	if ((ksp = kstat_lookup(kc, "unix", 0, "dnlcstats")) == NULL)
		return (errno);

	if (kstat_read(kc, ksp, &ss->s_sys.ss_nc) == -1)
		return (errno);

	if ((ksp = kstat_lookup(kc, "unix", 0, "system_misc")) == NULL)
		return (errno);

	if (kstat_read(kc, ksp, NULL) == -1)
		return (errno);

	knp = (kstat_named_t *)kstat_data_lookup(ksp, "clk_intr");
	if (knp == NULL)
		return (errno);

	ss->s_sys.ss_ticks = knp->value.l;

	knp = (kstat_named_t *)kstat_data_lookup(ksp, "deficit");
	if (knp == NULL)
		return (errno);

	ss->s_sys.ss_deficit = knp->value.l;

	for (i = 0; i < ss->s_nr_cpus; i++) {
		if (!CPU_ACTIVE(&ss->s_cpus[i]))
			continue;

		if (kstat_add(&ss->s_cpus[i].cs_sys, &ss->s_sys.ss_agg_sys))
			return (errno);
		if (kstat_add(&ss->s_cpus[i].cs_vm, &ss->s_sys.ss_agg_vm))
			return (errno);
	}

	return (0);
}

struct snapshot *
acquire_snapshot(kstat_ctl_t *kc, int types, struct iodev_filter *iodev_filter)
{
	struct snapshot *ss = NULL;
	int err;

retry:
	err = 0;
	/* ensure any partial resources are freed on a retry */
	free_snapshot(ss);

	ss = safe_alloc(sizeof (struct snapshot));

	(void) memset(ss, 0, sizeof (struct snapshot));

	ss->s_types = types;

	/* wait for a possibly up-to-date chain */
	while (kstat_chain_update(kc) == -1) {
		if (errno == EAGAIN)
			(void) poll(NULL, 0, RETRY_DELAY);
		else
			fail(1, "kstat_chain_update failed");
	}

	if (types & SNAP_FLUSHES) {
		kstat_t *ksp;
		ksp = kstat_lookup(kc, "unix", 0, "flushmeter");
		if (ksp == NULL) {
			fail(0, "This machine does not have "
				"a virtual address cache");
		}
		if (kstat_read(kc, ksp, &ss->s_flushes) == -1)
			err = errno;
	}

	if (!err && (types & SNAP_INTERRUPTS))
		err = acquire_intrs(ss, kc);

	if (!err && (types & (SNAP_CPUS | SNAP_SYSTEM | SNAP_PSETS)))
		err = acquire_cpus(ss, kc);

	if (!err && (types & SNAP_PSETS))
		err = acquire_psets(ss);

	if (!err && (types & (SNAP_IODEVS | SNAP_CONTROLLERS |
	    SNAP_IOPATHS_LI | SNAP_IOPATHS_LTI)))
		err = acquire_iodevs(ss, kc, iodev_filter);

	if (!err && (types & SNAP_SYSTEM))
		err = acquire_sys(ss, kc);

	switch (err) {
		case 0:
			break;
		case EAGAIN:
			(void) poll(NULL, 0, RETRY_DELAY);
		/* a kstat disappeared from under us */
		/*FALLTHRU*/
		case ENXIO:
		case ENOENT:
			goto retry;
		default:
			fail(1, "acquiring snapshot failed");
	}

	return (ss);
}

void
free_snapshot(struct snapshot *ss)
{
	size_t i;

	if (ss == NULL)
		return;

	while (ss->s_iodevs) {
		struct iodev_snapshot *tmp = ss->s_iodevs;
		ss->s_iodevs = ss->s_iodevs->is_next;
		free_iodev(tmp);
	}

	if (ss->s_cpus) {
		for (i = 0; i < ss->s_nr_cpus; i++) {
			free(ss->s_cpus[i].cs_vm.ks_data);
			free(ss->s_cpus[i].cs_sys.ks_data);
		}
		free(ss->s_cpus);
	}

	if (ss->s_psets) {
		for (i = 0; i < ss->s_nr_psets; i++)
			free(ss->s_psets[i].ps_cpus);
		free(ss->s_psets);
	}

	free(ss->s_sys.ss_agg_sys.ks_data);
	free(ss->s_sys.ss_agg_vm.ks_data);
	free(ss);
}

kstat_ctl_t *
open_kstat(void)
{
	kstat_ctl_t *kc;

	while ((kc = kstat_open()) == NULL) {
		if (errno == EAGAIN)
			(void) poll(NULL, 0, RETRY_DELAY);
		else
			fail(1, "kstat_open failed");
	}

	return (kc);
}

void *
safe_alloc(size_t size)
{
	void *ptr;

	while ((ptr = malloc(size)) == NULL) {
		if (errno == EAGAIN)
			(void) poll(NULL, 0, RETRY_DELAY);
		else
			fail(1, "malloc failed");
	}
	return (ptr);
}

char *
safe_strdup(char *str)
{
	char *ret;

	if (str == NULL)
		return (NULL);

	while ((ret = strdup(str)) == NULL) {
		if (errno == EAGAIN)
			(void) poll(NULL, 0, RETRY_DELAY);
		else
			fail(1, "malloc failed");
	}
	return (ret);
}

uint64_t
kstat_delta(kstat_t *old, kstat_t *new, char *name)
{
	kstat_named_t *knew = kstat_data_lookup(new, name);
	if (old && old->ks_data) {
		kstat_named_t *kold = kstat_data_lookup(old, name);
		return (knew->value.ui64 - kold->value.ui64);
	}
	return (knew->value.ui64);
}

int
kstat_copy(const kstat_t *src, kstat_t *dst)
{
	*dst = *src;

	if (src->ks_data != NULL) {
		if ((dst->ks_data = malloc(src->ks_data_size)) == NULL)
			return (-1);
		bcopy(src->ks_data, dst->ks_data, src->ks_data_size);
	} else {
		dst->ks_data = NULL;
		dst->ks_data_size = 0;
	}
	return (0);
}

int
kstat_add(const kstat_t *src, kstat_t *dst)
{
	size_t i;
	kstat_named_t *from;
	kstat_named_t *to;

	if (dst->ks_data == NULL)
		return (kstat_copy(src, dst));

	from = src->ks_data;
	to = dst->ks_data;

	for (i = 0; i < src->ks_ndata; i++) {
		/* "addition" makes little sense for strings */
		if (from->data_type != KSTAT_DATA_CHAR &&
			from->data_type != KSTAT_DATA_STRING)
			(to)->value.ui64 += (from)->value.ui64;
		from++;
		to++;
	}

	return (0);
}

uint64_t
cpu_ticks_delta(kstat_t *old, kstat_t *new)
{
	uint64_t ticks = 0;
	size_t i;
	for (i = 0; i < ARRAY_SIZE(cpu_states); i++)
		ticks += kstat_delta(old, new, cpu_states[i]);
	return (ticks);
}

int
nr_active_cpus(struct snapshot *ss)
{
	size_t i;
	int count = 0;
	for (i = 0; i < ss->s_nr_cpus; i++) {
		if (CPU_ACTIVE(&ss->s_cpus[i]))
			count++;
	}

	return (count);
}

/*
 * Return the number of ticks delta between two hrtime_t
 * values. Attempt to cater for various kinds of overflow
 * in hrtime_t - no matter how improbable.
 */
uint64_t
hrtime_delta(hrtime_t old, hrtime_t new)
{
	uint64_t del;

	if ((new >= old) && (old >= 0L))
		return (new - old);
	else {
		/*
		 * We've overflowed the positive portion of an
		 * hrtime_t.
		 */
		if (new < 0L) {
			/*
			 * The new value is negative. Handle the
			 * case where the old value is positive or
			 * negative.
			 */
			uint64_t n1;
			uint64_t o1;

			n1 = -new;
			if (old > 0L)
				return (n1 - old);
			else {
				o1 = -old;
				del = n1 - o1;
				return (del);
			}
		} else {
			/*
			 * Either we've just gone from being negative
			 * to positive *or* the last entry was positive
			 * and the new entry is also positive but *less*
			 * than the old entry. This implies we waited
			 * quite a few days on a very fast system between
			 * iostat displays.
			 */
			if (old < 0L) {
				uint64_t o2;

				o2 = -old;
				del = UINT64_MAX - o2;
			} else {
				del = UINT64_MAX - old;
			}
			del += new;
			return (del);
		}
	}
}