dev/sysmon/sysmon_envsys_events.c

/* $NetBSD: sysmon_envsys_events.c,v 1.20 2007/08/30 18:01:26 xtraeme Exp $ */

/*-
 * Copyright (c) 2007 The NetBSD Foundation, Inc.
 * All rights reserved.
 *
 * This code is derived from software contributed to The NetBSD Foundation
 * by Juan Romero Pardines.
 *
 * Redistribution and use in source and binary forms, with or without
 * modification, are permitted provided that the following conditions
 * are met:
 * 1. Redistributions of source code must retain the above copyright
 *    notice, this list of conditions and the following disclaimer.
 * 2. Redistributions in binary form must reproduce the above copyright
 *    notice, this list of conditions and the following disclaimer in the
 *    documentation and/or other materials provided with the distribution.
 * 3. All advertising materials mentioning features or use of this software
 *    must display the following acknowledgement:
 *      This product includes software developed by Juan Romero Pardines
 *      for the NetBSD Foundation, Inc. and its contributors.
 * 4. Neither the name of The NetBSD Foundation nor the names of its
 *    contributors may be used to endorse or promote products derived
 *    from this software without specific prior written permission.
 *
 * THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS
 * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED
 * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
 * PURPOSE ARE DISCLAIMED.  IN NO EVENT SHALL THE FOUNDATION OR CONTRIBUTORS
 * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
 * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
 * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
 * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
 * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
 * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
 * POSSIBILITY OF SUCH DAMAGE.
 */

/*
 * sysmon_envsys(9) events framework.
 */

#include <sys/cdefs.h>
__KERNEL_RCSID(0, "$NetBSD: sysmon_envsys_events.c,v 1.20 2007/08/30 18:01:26 xtraeme Exp $");

#include <sys/param.h>
#include <sys/types.h>
#include <sys/conf.h>
#include <sys/errno.h>
#include <sys/kernel.h>
#include <sys/sysctl.h>
#include <sys/systm.h>
#include <sys/proc.h>
#include <sys/mutex.h>
#include <sys/kmem.h>
#include <sys/callout.h>

#include <dev/sysmon/sysmonvar.h>
#include <dev/sysmon/sysmon_envsysvar.h>

struct sme_sensor_state {
	int		type;
	const char 	*desc;
};

struct sme_sensor_event {
	int		state;
	int		event;
};

static const struct sme_sensor_state sme_sensor_drive_state[] = {
	{ ENVSYS_DRIVE_EMPTY, 		"drive state is unknown" },
	{ ENVSYS_DRIVE_READY, 		"drive is ready" },
	{ ENVSYS_DRIVE_POWERUP,		"drive is powering up" },
	{ ENVSYS_DRIVE_ONLINE, 		"drive is online" },
	{ ENVSYS_DRIVE_IDLE, 		"drive is idle" },
	{ ENVSYS_DRIVE_ACTIVE, 		"drive is active" },
	{ ENVSYS_DRIVE_REBUILD, 	"drive is rebuilding" },
	{ ENVSYS_DRIVE_POWERDOWN,	"drive is powering down" },
	{ ENVSYS_DRIVE_FAIL, 		"drive failed" },
	{ ENVSYS_DRIVE_PFAIL,		"drive degraded" },
	{ -1, 				"unknown" }
};

static const struct sme_sensor_event sme_sensor_event[] = {
	{ ENVSYS_SVALID, 	PENVSYS_EVENT_NORMAL },
	{ ENVSYS_SCRITICAL, 	PENVSYS_EVENT_CRITICAL },
	{ ENVSYS_SCRITOVER, 	PENVSYS_EVENT_CRITOVER },
	{ ENVSYS_SCRITUNDER, 	PENVSYS_EVENT_CRITUNDER },
	{ ENVSYS_SWARNOVER, 	PENVSYS_EVENT_WARNOVER },
	{ ENVSYS_SWARNUNDER,	PENVSYS_EVENT_WARNUNDER },
	{ -1, 			-1 }
};

static struct workqueue *seewq;
static struct callout seeco;
static bool sme_events_initialized = false;
kmutex_t sme_list_mtx, sme_event_mtx, sme_event_init_mtx;
kcondvar_t sme_event_cv;

/* 10 seconds of timeout for the callout */
static int sme_events_timeout = 10;
static int sme_events_timeout_sysctl(SYSCTLFN_PROTO);
#define SME_EVTIMO 	(sme_events_timeout * hz)

/*
 * sysctl(9) stuff to handle the refresh value in the callout
 * function.
 */
static int
sme_events_timeout_sysctl(SYSCTLFN_ARGS)
{
	struct sysctlnode node;
	int timo, error;

	node = *rnode;
	timo = sme_events_timeout;
	node.sysctl_data = &timo;

	error = sysctl_lookup(SYSCTLFN_CALL(&node));
	if (error || newp == NULL)
		return error;

	/* min 1s */
	if (timo < 1)
		return EINVAL;

	sme_events_timeout = timo;
	return 0;
}

SYSCTL_SETUP(sysctl_kern_envsys_timeout_setup, "sysctl kern.envsys subtree")
{
	const struct sysctlnode *node, *envsys_node;

	sysctl_createv(clog, 0, NULL, &node,
			CTLFLAG_PERMANENT,
			CTLTYPE_NODE, "kern", NULL,
			NULL, 0, NULL, 0,
			CTL_KERN, CTL_EOL);

	sysctl_createv(clog, 0, &node, &envsys_node,
			0,
			CTLTYPE_NODE, "envsys", NULL,
			NULL, 0, NULL, 0,
			CTL_CREATE, CTL_EOL);

	sysctl_createv(clog, 0, &envsys_node, &node,
			CTLFLAG_READWRITE,
			CTLTYPE_INT, "refresh_value",
			SYSCTL_DESCR("wait time in seconds to refresh "
			    "sensors being monitored"),
			sme_events_timeout_sysctl, 0, &sme_events_timeout, 0,
			CTL_CREATE, CTL_EOL);
}


/*
 * sme_event_register:
 *
 * 	+ Registers a sysmon envsys event.
 * 	+ Creates a new sysmon envsys event.
 */
int
sme_event_register(sme_event_t *see)
{
	sme_event_t *lsee;
	int error = 0;

	KASSERT(see != NULL);

	mutex_enter(&sme_event_mtx);
	/*
	 * Ensure that we don't add events for the same sensor
	 * and with the same type.
	 */
	LIST_FOREACH(lsee, &sme_events_list, see_list) {
		if (strcmp(lsee->pes.pes_sensname,
		    see->pes.pes_sensname) == 0) {
			if (lsee->type == see->type) {
				DPRINTF(("%s: dev=%s sensor=%s type=%d "
				    "(already exists)\n", __func__,
				    see->pes.pes_dvname,
				    see->pes.pes_sensname, see->type));
				mutex_exit(&sme_event_mtx);
				return EEXIST;
			}
		}
	}

	DPRINTF(("%s: dev=%s sensor=%s snum=%d type=%d "
	    "critval=%" PRIu32 "\n", __func__,
	    see->pes.pes_dvname, see->pes.pes_sensname,
	    see->snum, see->type, see->critval));

	LIST_INSERT_HEAD(&sme_events_list, see, see_list);
	/*
	 * Initialize the events framework if it wasn't initialized
	 * before.
	 */
	mutex_enter(&sme_event_init_mtx);
	if (sme_events_initialized == false)
		error = sme_events_init();
	mutex_exit(&sme_event_init_mtx);

	mutex_exit(&sme_event_mtx);
	return error;
}

/*
 * sme_event_unregister_all:
 *
 * 	+ Unregisters all sysmon envsys events associated with a
 * 	  sysmon envsys device.
 */
void
sme_event_unregister_all(struct sysmon_envsys *sme)
{
	sme_event_t *see;
	int evcounter = 0;

	KASSERT(sme != NULL);

	mutex_enter(&sme_event_mtx);
	LIST_FOREACH(see, &sme_events_list, see_list) {
		if (strcmp(see->pes.pes_dvname, sme->sme_name) == 0)
			evcounter++;
	}

	DPRINTF(("%s: total events %d (%s)\n", __func__,
	    evcounter, sme->sme_name));

	while ((see = LIST_FIRST(&sme_events_list)) != NULL) {
		if (evcounter == 0)
			break;

		if (strcmp(see->pes.pes_dvname, sme->sme_name) == 0) {
			DPRINTF(("%s: event %s %d removed (%s)\n", __func__,
			    see->pes.pes_sensname, see->type, sme->sme_name));

			while (see->see_flags & SME_EVENT_WORKING)
				cv_wait(&sme_event_cv, &sme_event_mtx);

			LIST_REMOVE(see, see_list);
			kmem_free(see, sizeof(*see));
			evcounter--;
		}
	}

	if (LIST_EMPTY(&sme_events_list)) {
		mutex_exit(&sme_event_mtx);
		sme_events_destroy();
		return;
	}

	mutex_exit(&sme_event_mtx);
}

/*
 * sme_event_unregister:
 *
 * 	+ Unregisters a sysmon envsys event.
 */
int
sme_event_unregister(const char *sensor, int type)
{
	sme_event_t *see;
	bool found = false;

	KASSERT(sensor != NULL);

	mutex_enter(&sme_event_mtx);
	LIST_FOREACH(see, &sme_events_list, see_list) {
		if (strcmp(see->pes.pes_sensname, sensor) == 0) {
			if (see->type == type) {
				found = true;
				break;
			}
		}
	}

	if (!found) {
		mutex_exit(&sme_event_mtx);
		return EINVAL;
	}

	while (see->see_flags & SME_EVENT_WORKING)
		cv_wait(&sme_event_cv, &sme_event_mtx);

	DPRINTF(("%s: removing dev=%s sensor=%s type=%d\n",
	    __func__, see->pes.pes_dvname, sensor, type));
	LIST_REMOVE(see, see_list);
	/*
	 * So the events list is empty, we'll do the following:
	 *
	 * 	- stop and destroy the callout.
	 * 	- destroy the workqueue.
	 */
	if (LIST_EMPTY(&sme_events_list)) {
		mutex_exit(&sme_event_mtx);
		sme_events_destroy();
		goto out;
	}
	mutex_exit(&sme_event_mtx);

out:
	kmem_free(see, sizeof(*see));
	return 0;
}

/*
 * sme_event_drvadd:
 *
 * 	+ Adds a new sysmon envsys event for a driver if a sensor
 * 	  has set any accepted monitoring flag.
 */
void
sme_event_drvadd(void *arg)
{
	sme_event_drv_t *sed_t = arg;
	int error = 0;

	KASSERT(sed_t != NULL);

#define SEE_REGEVENT(a, b, c)						\
do {									\
	if (sed_t->edata->flags & (a)) {				\
		char str[32] = "monitoring-state-";			\
									\
		error = sme_event_add(sed_t->sdict,			\
				      sed_t->edata,			\
				      sed_t->sme->sme_name,		\
				      NULL,				\
				      0,				\
				      (b),				\
				      sed_t->powertype);		\
		if (error && error != EEXIST)				\
			printf("%s: failed to add event! "		\
			    "error=%d sensor=%s event=%s\n",		\
			    __func__, error, sed_t->edata->desc, (c));	\
		else {							\
			mutex_enter(&sme_list_mtx);			\
			(void)strlcat(str, (c), sizeof(str));		\
			prop_dictionary_set_bool(sed_t->sdict,		\
						 str,			\
						 true);			\
			mutex_exit(&sme_list_mtx);			\
		}							\
	}								\
} while (/* CONSTCOND */ 0)

	SEE_REGEVENT(ENVSYS_FMONCRITICAL,
		     PENVSYS_EVENT_CRITICAL,
		     "critical");

	SEE_REGEVENT(ENVSYS_FMONCRITUNDER,
		     PENVSYS_EVENT_CRITUNDER,
		     "critunder");

	SEE_REGEVENT(ENVSYS_FMONCRITOVER,
		     PENVSYS_EVENT_CRITOVER,
		     "critover");

	SEE_REGEVENT(ENVSYS_FMONWARNUNDER,
		     PENVSYS_EVENT_WARNUNDER,
		     "warnunder");

	SEE_REGEVENT(ENVSYS_FMONWARNOVER,
		     PENVSYS_EVENT_WARNOVER,
		     "warnover");

	SEE_REGEVENT(ENVSYS_FMONDRVSTATE,
		     PENVSYS_EVENT_DRIVE_STCHANGED,
		     "drvstchanged");

	/* we are done, free memory now */
	kmem_free(sed_t, sizeof(*sed_t));
}

/*
 * sme_event_add:
 *
 * 	+ Initializes or updates a sysmon envsys event.
 */
int
sme_event_add(prop_dictionary_t sdict, envsys_data_t *edata,
	      const char *drvn, const char *objkey,
	      int32_t critval, int crittype, int powertype)
{
	sme_event_t *see = NULL;
	prop_object_t obj;
	int error = 0;

	KASSERT(sdict != NULL || edata != NULL);

	mutex_enter(&sme_event_mtx);
	/* critical condition set via userland */
	if (objkey && critval) {
		obj = prop_dictionary_get(sdict, objkey);
		if (obj != NULL) {
			/*
			 * object is already in dictionary, update
			 * the critical value.
			 */
	 		LIST_FOREACH(see, &sme_events_list, see_list) {
				if (strcmp(edata->desc,
				    see->pes.pes_sensname) == 0)
					if (crittype == see->type)
						break;
		 	}
			if (see->critval != critval) {
				see->critval = critval;
				DPRINTF(("%s: sensor=%s type=%d "
				    "(critval updated)\n", __func__,
				    edata->desc, see->type));
			}
			goto out;
		}
	}

	if (LIST_EMPTY(&sme_events_list))
		goto register_event;

	/* check if the event is already on the list */
	LIST_FOREACH(see, &sme_events_list, see_list) {
		if (strcmp(edata->desc, see->pes.pes_sensname) == 0)
			if (crittype == see->type) {
				mutex_exit(&sme_event_mtx);
				return EEXIST;
			}
	}

	/*
	 * object is not in dictionary, create a new
	 * sme event and assign required members.
	 */
register_event:
	see = kmem_zalloc(sizeof(*see), KM_NOSLEEP);
	if (see == NULL) {
		mutex_exit(&sme_event_mtx);
		return ENOMEM;
	}

	see->critval = critval;
	see->type = crittype;
	(void)strlcpy(see->pes.pes_dvname, drvn,
	    sizeof(see->pes.pes_dvname));
	see->pes.pes_type = powertype;
	(void)strlcpy(see->pes.pes_sensname, edata->desc,
	    sizeof(see->pes.pes_sensname));
	see->snum = edata->sensor;
out:
	/* update the object in the dictionary */
	if (objkey && critval) {
		error = sme_sensor_upint32(sdict, objkey, critval);
		if (error) {
			mutex_exit(&sme_event_mtx);
			return error;
		}
	}
	mutex_exit(&sme_event_mtx);
	error = sme_event_register(see);
	if (error)
		kmem_free(see, sizeof(*see));
	return error;
}

/*
 * sme_events_init:
 *
 * 	+ Initializes the events framework.
 */
int
sme_events_init(void)
{
	int error;

	error = workqueue_create(&seewq, "envsysev",
	    sme_events_worker, NULL, 0, IPL_SOFTCLOCK, WQ_MPSAFE);
	if (error)
		goto out;

	callout_init(&seeco, 0);
	callout_setfunc(&seeco, sme_events_check, NULL);
	callout_schedule(&seeco, SME_EVTIMO);
	sme_events_initialized = true;
	DPRINTF(("%s: events framework initialized\n", __func__));

out:
	return error;
}

/*
 * sme_events_destroy:
 *
 * 	+ Destroys the events framework: the workqueue and the
 * 	  callout are stopped and destroyed because there are not
 * 	  events in the queue.
 */
void
sme_events_destroy(void)
{
	mutex_enter(&sme_event_init_mtx);
	callout_stop(&seeco);
	sme_events_initialized = false;
	DPRINTF(("%s: events framework destroyed\n", __func__));
	mutex_exit(&sme_event_init_mtx);
	callout_destroy(&seeco);
	workqueue_destroy(seewq);
}

/*
 * sme_events_check:
 *
 * 	+ Runs the work on each sysmon envsys event in our
 * 	  workqueue periodically with callout.
 */
void
sme_events_check(void *arg)
{
	sme_event_t *see;

	LIST_FOREACH(see, &sme_events_list, see_list) {
		DPRINTFOBJ(("%s: dev=%s sensor=%s type=%d\n",
		    __func__,
		    see->pes.pes_dvname,
		    see->pes.pes_sensname,
		    see->type));
		workqueue_enqueue(seewq, &see->see_wk, NULL);
	}
	callout_schedule(&seeco, SME_EVTIMO);
}

/*
 * sme_events_worker:
 *
 * 	+ workqueue thread that checks if there's a critical condition
 * 	  and sends an event if the condition was triggered.
 */
void
sme_events_worker(struct work *wk, void *arg)
{
	const struct sme_sensor_state *esds = sme_sensor_drive_state;
	const struct sme_sensor_event *sse = sme_sensor_event;
	sme_event_t *see = (void *)wk;
	struct sysmon_envsys *sme;
	envsys_data_t *edata;
	int i, error = 0;

	KASSERT(wk == &see->see_wk);

	mutex_enter(&sme_event_mtx);
	see->see_flags |= SME_EVENT_WORKING;

	/*
	 * We have to find the sme device by looking
	 * at the power envsys device name.
	 */
	mutex_enter(&sme_list_mtx);
	LIST_FOREACH(sme, &sysmon_envsys_list, sme_list)
		if (strcmp(sme->sme_name, see->pes.pes_dvname) == 0)
			break;
	mutex_exit(&sme_list_mtx);

	KASSERT(sme != NULL);

	/* get the sensor with the index specified in see->snum */
	edata = &sme->sme_sensor_data[see->snum];

	/*
	 * refresh the sensor that was marked with a critical
	 * event.
	 */
	if ((sme->sme_flags & SME_DISABLE_GTREDATA) == 0) {
		error = (*sme->sme_gtredata)(sme, edata);
		if (error)
			return;
	}

	DPRINTFOBJ(("%s: desc=%s sensor=%d units=%d value_cur=%d\n",
	    __func__, edata->desc, edata->sensor,
	    edata->units, edata->value_cur));

#define SME_SEND_NORMALEVENT()						\
do {									\
	see->evsent = false;						\
	sysmon_penvsys_event(&see->pes, PENVSYS_EVENT_NORMAL);		\
} while (/* CONSTCOND */ 0)

#define SME_SEND_EVENT(type)						\
do {									\
	see->evsent = true;						\
	sysmon_penvsys_event(&see->pes, (type));			\
} while (/* CONSTCOND */ 0)

	switch (see->type) {
	/*
	 * if state is the same than the one that matches sse[i].state,
	 * send the event...
	 */
	case PENVSYS_EVENT_CRITICAL:
	case PENVSYS_EVENT_CRITUNDER:
	case PENVSYS_EVENT_CRITOVER:
	case PENVSYS_EVENT_WARNUNDER:
	case PENVSYS_EVENT_WARNOVER:
		for (i = 0; sse[i].state != -1; i++)
			if (sse[i].event == see->type)
				break;

		if (see->evsent && edata->state == ENVSYS_SVALID)
			SME_SEND_NORMALEVENT();

		if (!see->evsent && edata->state == sse[i].state)
			SME_SEND_EVENT(see->type);

		break;
	/*
	 * if value_cur is lower than the limit, send the event...
	 */
	case PENVSYS_EVENT_BATT_USERCAP:
	case PENVSYS_EVENT_USER_CRITMIN:
		if (see->evsent && edata->value_cur > see->critval)
			SME_SEND_NORMALEVENT();

		if (!see->evsent && edata->value_cur < see->critval)
			SME_SEND_EVENT(see->type);

		break;
	/*
	 * if value_cur is higher than the limit, send the event...
	 */
	case PENVSYS_EVENT_USER_CRITMAX:
		if (see->evsent && edata->value_cur < see->critval)
			SME_SEND_NORMALEVENT();

		if (!see->evsent && edata->value_cur > see->critval)
			SME_SEND_EVENT(see->type);

		break;
	/*
	 * if value_cur is not ENVSYS_DRIVE_ONLINE, send the event...
	 */
	case PENVSYS_EVENT_DRIVE_STCHANGED:
		/* the state has not been changed, just ignore the event */
		if (edata->value_cur == see->evsent)
			break;

		for (i = 0; esds[i].type != -1; i++)
			if (esds[i].type == edata->value_cur)
				break;

		/* copy current state description  */
		(void)strlcpy(see->pes.pes_statedesc, esds[i].desc,
		    sizeof(see->pes.pes_statedesc));

		/* state is ok again... send a normal event */
		if (see->evsent && edata->value_cur == ENVSYS_DRIVE_ONLINE)
			SME_SEND_NORMALEVENT();

		/* something bad happened to the drive... send the event */
		if (see->evsent || edata->value_cur != ENVSYS_DRIVE_ONLINE) {
			/* save current drive state */
			see->evsent = edata->value_cur;
			sysmon_penvsys_event(&see->pes, see->type);
		}

		break;
	}
	see->see_flags &= ~SME_EVENT_WORKING;
	cv_broadcast(&sme_event_cv);
	mutex_exit(&sme_event_mtx);
}