xref: /onnv/onnv-gate/usr/src/uts/common/disp/fss.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 2008 Sun Microsystems, Inc.  All rights reserved.
 * Use is subject to license terms.
 */

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

#include <sys/types.h>
#include <sys/param.h>
#include <sys/sysmacros.h>
#include <sys/cred.h>
#include <sys/proc.h>
#include <sys/strsubr.h>
#include <sys/priocntl.h>
#include <sys/class.h>
#include <sys/disp.h>
#include <sys/procset.h>
#include <sys/debug.h>
#include <sys/kmem.h>
#include <sys/errno.h>
#include <sys/systm.h>
#include <sys/schedctl.h>
#include <sys/vmsystm.h>
#include <sys/atomic.h>
#include <sys/project.h>
#include <sys/modctl.h>
#include <sys/fss.h>
#include <sys/fsspriocntl.h>
#include <sys/cpupart.h>
#include <sys/zone.h>
#include <vm/rm.h>
#include <vm/seg_kmem.h>
#include <sys/tnf_probe.h>
#include <sys/policy.h>
#include <sys/sdt.h>
#include <sys/cpucaps.h>

/*
 * FSS Data Structures:
 *
 *                 fsszone
 *                  -----           -----
 *  -----          |     |         |     |
 * |     |-------->|     |<------->|     |<---->...
 * |     |          -----           -----
 * |     |          ^    ^            ^
 * |     |---       |     \            \
 *  -----    |      |      \            \
 * fsspset   |      |       \            \
 *           |      |        \            \
 *           |    -----       -----       -----
 *            -->|     |<--->|     |<--->|     |
 *               |     |     |     |     |     |
 *                -----       -----       -----
 *               fssproj
 *
 *
 * That is, fsspsets contain a list of fsszone's that are currently active in
 * the pset, and a list of fssproj's, corresponding to projects with runnable
 * threads on the pset.  fssproj's in turn point to the fsszone which they
 * are a member of.
 *
 * An fssproj_t is removed when there are no threads in it.
 *
 * An fsszone_t is removed when there are no projects with threads in it.
 *
 * Projects in a zone compete with each other for cpu time, receiving cpu
 * allocation within a zone proportional to fssproj->fssp_shares
 * (project.cpu-shares); at a higher level zones compete with each other,
 * receiving allocation in a pset proportional to fsszone->fssz_shares
 * (zone.cpu-shares).  See fss_decay_usage() for the precise formula.
 */

static pri_t fss_init(id_t, int, classfuncs_t **);

static struct sclass fss = {
	"FSS",
	fss_init,
	0
};

extern struct mod_ops mod_schedops;

/*
 * Module linkage information for the kernel.
 */
static struct modlsched modlsched = {
	&mod_schedops, "fair share scheduling class", &fss
};

static struct modlinkage modlinkage = {
	MODREV_1, (void *)&modlsched, NULL
};

#define	FSS_MAXUPRI	60

/*
 * The fssproc_t structures are kept in an array of circular doubly linked
 * lists.  A hash on the thread pointer is used to determine which list each
 * thread should be placed in.  Each list has a dummy "head" which is never
 * removed, so the list is never empty.  fss_update traverses these lists to
 * update the priorities of threads that have been waiting on the run queue.
 */
#define	FSS_LISTS		16 /* number of lists, must be power of 2 */
#define	FSS_LIST_HASH(t)	(((uintptr_t)(t) >> 9) & (FSS_LISTS - 1))
#define	FSS_LIST_NEXT(i)	(((i) + 1) & (FSS_LISTS - 1))

#define	FSS_LIST_INSERT(fssproc)				\
{								\
	int index = FSS_LIST_HASH(fssproc->fss_tp);		\
	kmutex_t *lockp = &fss_listlock[index];			\
	fssproc_t *headp = &fss_listhead[index];		\
	mutex_enter(lockp);					\
	fssproc->fss_next = headp->fss_next;			\
	fssproc->fss_prev = headp;				\
	headp->fss_next->fss_prev = fssproc;			\
	headp->fss_next = fssproc;				\
	mutex_exit(lockp);					\
}

#define	FSS_LIST_DELETE(fssproc)				\
{								\
	int index = FSS_LIST_HASH(fssproc->fss_tp);		\
	kmutex_t *lockp = &fss_listlock[index];			\
	mutex_enter(lockp);					\
	fssproc->fss_prev->fss_next = fssproc->fss_next;	\
	fssproc->fss_next->fss_prev = fssproc->fss_prev;	\
	mutex_exit(lockp);					\
}

#define	FSS_TICK_COST	1000	/* tick cost for threads with nice level = 0 */

/*
 * Decay rate percentages are based on n/128 rather than n/100 so  that
 * calculations can avoid having to do an integer divide by 100 (divide
 * by FSS_DECAY_BASE == 128 optimizes to an arithmetic shift).
 *
 * FSS_DECAY_MIN	=  83/128 ~= 65%
 * FSS_DECAY_MAX	= 108/128 ~= 85%
 * FSS_DECAY_USG	=  96/128 ~= 75%
 */
#define	FSS_DECAY_MIN	83	/* fsspri decay pct for threads w/ nice -20 */
#define	FSS_DECAY_MAX	108	/* fsspri decay pct for threads w/ nice +19 */
#define	FSS_DECAY_USG	96	/* fssusage decay pct for projects */
#define	FSS_DECAY_BASE	128	/* base for decay percentages above */

#define	FSS_NICE_MIN	0
#define	FSS_NICE_MAX	(2 * NZERO - 1)
#define	FSS_NICE_RANGE	(FSS_NICE_MAX - FSS_NICE_MIN + 1)

static int	fss_nice_tick[FSS_NICE_RANGE];
static int	fss_nice_decay[FSS_NICE_RANGE];

static pri_t	fss_maxupri = FSS_MAXUPRI; /* maximum FSS user priority */
static pri_t	fss_maxumdpri; /* maximum user mode fss priority */
static pri_t	fss_maxglobpri;	/* maximum global priority used by fss class */
static pri_t	fss_minglobpri;	/* minimum global priority */

static fssproc_t fss_listhead[FSS_LISTS];
static kmutex_t	fss_listlock[FSS_LISTS];

static fsspset_t *fsspsets;
static kmutex_t fsspsets_lock;	/* protects fsspsets */

static id_t	fss_cid;

static time_t	fss_minrun = 2;	/* t_pri becomes 59 within 2 secs */
static time_t	fss_minslp = 2;	/* min time on sleep queue for hardswap */
static int	fss_quantum = 11;

static void	fss_newpri(fssproc_t *);
static void	fss_update(void *);
static int	fss_update_list(int);
static void	fss_change_priority(kthread_t *, fssproc_t *);

static int	fss_admin(caddr_t, cred_t *);
static int	fss_getclinfo(void *);
static int	fss_parmsin(void *);
static int	fss_parmsout(void *, pc_vaparms_t *);
static int	fss_vaparmsin(void *, pc_vaparms_t *);
static int	fss_vaparmsout(void *, pc_vaparms_t *);
static int	fss_getclpri(pcpri_t *);
static int	fss_alloc(void **, int);
static void	fss_free(void *);

static int	fss_enterclass(kthread_t *, id_t, void *, cred_t *, void *);
static void	fss_exitclass(void *);
static int	fss_canexit(kthread_t *, cred_t *);
static int	fss_fork(kthread_t *, kthread_t *, void *);
static void	fss_forkret(kthread_t *, kthread_t *);
static void	fss_parmsget(kthread_t *, void *);
static int	fss_parmsset(kthread_t *, void *, id_t, cred_t *);
static void	fss_stop(kthread_t *, int, int);
static void	fss_exit(kthread_t *);
static void	fss_active(kthread_t *);
static void	fss_inactive(kthread_t *);
static pri_t	fss_swapin(kthread_t *, int);
static pri_t	fss_swapout(kthread_t *, int);
static void	fss_trapret(kthread_t *);
static void	fss_preempt(kthread_t *);
static void	fss_setrun(kthread_t *);
static void	fss_sleep(kthread_t *);
static void	fss_tick(kthread_t *);
static void	fss_wakeup(kthread_t *);
static int	fss_donice(kthread_t *, cred_t *, int, int *);
static int	fss_doprio(kthread_t *, cred_t *, int, int *);
static pri_t	fss_globpri(kthread_t *);
static void	fss_yield(kthread_t *);
static void	fss_nullsys();

static struct classfuncs fss_classfuncs = {
	/* class functions */
	fss_admin,
	fss_getclinfo,
	fss_parmsin,
	fss_parmsout,
	fss_vaparmsin,
	fss_vaparmsout,
	fss_getclpri,
	fss_alloc,
	fss_free,

	/* thread functions */
	fss_enterclass,
	fss_exitclass,
	fss_canexit,
	fss_fork,
	fss_forkret,
	fss_parmsget,
	fss_parmsset,
	fss_stop,
	fss_exit,
	fss_active,
	fss_inactive,
	fss_swapin,
	fss_swapout,
	fss_trapret,
	fss_preempt,
	fss_setrun,
	fss_sleep,
	fss_tick,
	fss_wakeup,
	fss_donice,
	fss_globpri,
	fss_nullsys,	/* set_process_group */
	fss_yield,
	fss_doprio,
};

int
_init()
{
	return (mod_install(&modlinkage));
}

int
_fini()
{
	return (EBUSY);
}

int
_info(struct modinfo *modinfop)
{
	return (mod_info(&modlinkage, modinfop));
}

/*ARGSUSED*/
static int
fss_project_walker(kproject_t *kpj, void *buf)
{
	return (0);
}

void *
fss_allocbuf(int op, int type)
{
	fssbuf_t *fssbuf;
	void **fsslist;
	int cnt;
	int i;
	size_t size;

	ASSERT(op == FSS_NPSET_BUF || op == FSS_NPROJ_BUF || op == FSS_ONE_BUF);
	ASSERT(type == FSS_ALLOC_PROJ || type == FSS_ALLOC_ZONE);
	ASSERT(MUTEX_HELD(&cpu_lock));

	fssbuf = kmem_zalloc(sizeof (fssbuf_t), KM_SLEEP);
	switch (op) {
	case FSS_NPSET_BUF:
		cnt = cpupart_list(NULL, 0, CP_NONEMPTY);
		break;
	case FSS_NPROJ_BUF:
		cnt = project_walk_all(ALL_ZONES, fss_project_walker, NULL);
		break;
	case FSS_ONE_BUF:
		cnt = 1;
		break;
	}

	switch (type) {
	case FSS_ALLOC_PROJ:
		size = sizeof (fssproj_t);
		break;
	case FSS_ALLOC_ZONE:
		size = sizeof (fsszone_t);
		break;
	}
	fsslist = kmem_zalloc(cnt * sizeof (void *), KM_SLEEP);
	fssbuf->fssb_size = cnt;
	fssbuf->fssb_list = fsslist;
	for (i = 0; i < cnt; i++)
		fsslist[i] = kmem_zalloc(size, KM_SLEEP);
	return (fssbuf);
}

void
fss_freebuf(fssbuf_t *fssbuf, int type)
{
	void **fsslist;
	int i;
	size_t size;

	ASSERT(fssbuf != NULL);
	ASSERT(type == FSS_ALLOC_PROJ || type == FSS_ALLOC_ZONE);
	fsslist = fssbuf->fssb_list;

	switch (type) {
	case FSS_ALLOC_PROJ:
		size = sizeof (fssproj_t);
		break;
	case FSS_ALLOC_ZONE:
		size = sizeof (fsszone_t);
		break;
	}

	for (i = 0; i < fssbuf->fssb_size; i++) {
		if (fsslist[i] != NULL)
			kmem_free(fsslist[i], size);
	}
	kmem_free(fsslist, sizeof (void *) * fssbuf->fssb_size);
	kmem_free(fssbuf, sizeof (fssbuf_t));
}

static fsspset_t *
fss_find_fsspset(cpupart_t *cpupart)
{
	int i;
	fsspset_t *fsspset = NULL;
	int found = 0;

	ASSERT(cpupart != NULL);
	ASSERT(MUTEX_HELD(&fsspsets_lock));

	/*
	 * Search for the cpupart pointer in the array of fsspsets.
	 */
	for (i = 0; i < max_ncpus; i++) {
		fsspset = &fsspsets[i];
		if (fsspset->fssps_cpupart == cpupart) {
			ASSERT(fsspset->fssps_nproj > 0);
			found = 1;
			break;
		}
	}
	if (found == 0) {
		/*
		 * If we didn't find anything, then use the first
		 * available slot in the fsspsets array.
		 */
		for (i = 0; i < max_ncpus; i++) {
			fsspset = &fsspsets[i];
			if (fsspset->fssps_cpupart == NULL) {
				ASSERT(fsspset->fssps_nproj == 0);
				found = 1;
				break;
			}
		}
		fsspset->fssps_cpupart = cpupart;
	}
	ASSERT(found == 1);
	return (fsspset);
}

static void
fss_del_fsspset(fsspset_t *fsspset)
{
	ASSERT(MUTEX_HELD(&fsspsets_lock));
	ASSERT(MUTEX_HELD(&fsspset->fssps_lock));
	ASSERT(fsspset->fssps_nproj == 0);
	ASSERT(fsspset->fssps_list == NULL);
	ASSERT(fsspset->fssps_zones == NULL);
	fsspset->fssps_cpupart = NULL;
	fsspset->fssps_maxfsspri = 0;
	fsspset->fssps_shares = 0;
}

/*
 * The following routine returns a pointer to the fsszone structure which
 * belongs to zone "zone" and cpu partition fsspset, if such structure exists.
 */
static fsszone_t *
fss_find_fsszone(fsspset_t *fsspset, zone_t *zone)
{
	fsszone_t *fsszone;

	ASSERT(MUTEX_HELD(&fsspset->fssps_lock));

	if (fsspset->fssps_list != NULL) {
		/*
		 * There are projects/zones active on this cpu partition
		 * already.  Try to find our zone among them.
		 */
		fsszone = fsspset->fssps_zones;
		do {
			if (fsszone->fssz_zone == zone) {
				return (fsszone);
			}
			fsszone = fsszone->fssz_next;
		} while (fsszone != fsspset->fssps_zones);
	}
	return (NULL);
}

/*
 * The following routine links new fsszone structure into doubly linked list of
 * zones active on the specified cpu partition.
 */
static void
fss_insert_fsszone(fsspset_t *fsspset, zone_t *zone, fsszone_t *fsszone)
{
	ASSERT(MUTEX_HELD(&fsspset->fssps_lock));

	fsszone->fssz_zone = zone;
	fsszone->fssz_rshares = zone->zone_shares;

	if (fsspset->fssps_zones == NULL) {
		/*
		 * This will be the first fsszone for this fsspset
		 */
		fsszone->fssz_next = fsszone->fssz_prev = fsszone;
		fsspset->fssps_zones = fsszone;
	} else {
		/*
		 * Insert this fsszone to the doubly linked list.
		 */
		fsszone_t *fssz_head = fsspset->fssps_zones;

		fsszone->fssz_next = fssz_head;
		fsszone->fssz_prev = fssz_head->fssz_prev;
		fssz_head->fssz_prev->fssz_next = fsszone;
		fssz_head->fssz_prev = fsszone;
		fsspset->fssps_zones = fsszone;
	}
}

/*
 * The following routine removes a single fsszone structure from the doubly
 * linked list of zones active on the specified cpu partition.  Note that
 * global fsspsets_lock must be held in case this fsszone structure is the last
 * on the above mentioned list.  Also note that the fsszone structure is not
 * freed here, it is the responsibility of the caller to call kmem_free for it.
 */
static void
fss_remove_fsszone(fsspset_t *fsspset, fsszone_t *fsszone)
{
	ASSERT(MUTEX_HELD(&fsspset->fssps_lock));
	ASSERT(fsszone->fssz_nproj == 0);
	ASSERT(fsszone->fssz_shares == 0);
	ASSERT(fsszone->fssz_runnable == 0);

	if (fsszone->fssz_next != fsszone) {
		/*
		 * This is not the last zone in the list.
		 */
		fsszone->fssz_prev->fssz_next = fsszone->fssz_next;
		fsszone->fssz_next->fssz_prev = fsszone->fssz_prev;
		if (fsspset->fssps_zones == fsszone)
			fsspset->fssps_zones = fsszone->fssz_next;
	} else {
		/*
		 * This was the last zone active in this cpu partition.
		 */
		fsspset->fssps_zones = NULL;
	}
}

/*
 * The following routine returns a pointer to the fssproj structure
 * which belongs to project kpj and cpu partition fsspset, if such structure
 * exists.
 */
static fssproj_t *
fss_find_fssproj(fsspset_t *fsspset, kproject_t *kpj)
{
	fssproj_t *fssproj;

	ASSERT(MUTEX_HELD(&fsspset->fssps_lock));

	if (fsspset->fssps_list != NULL) {
		/*
		 * There are projects running on this cpu partition already.
		 * Try to find our project among them.
		 */
		fssproj = fsspset->fssps_list;
		do {
			if (fssproj->fssp_proj == kpj) {
				ASSERT(fssproj->fssp_pset == fsspset);
				return (fssproj);
			}
			fssproj = fssproj->fssp_next;
		} while (fssproj != fsspset->fssps_list);
	}
	return (NULL);
}

/*
 * The following routine links new fssproj structure into doubly linked list
 * of projects running on the specified cpu partition.
 */
static void
fss_insert_fssproj(fsspset_t *fsspset, kproject_t *kpj, fsszone_t *fsszone,
    fssproj_t *fssproj)
{
	ASSERT(MUTEX_HELD(&fsspset->fssps_lock));

	fssproj->fssp_pset = fsspset;
	fssproj->fssp_proj = kpj;
	fssproj->fssp_shares = kpj->kpj_shares;

	fsspset->fssps_nproj++;

	if (fsspset->fssps_list == NULL) {
		/*
		 * This will be the first fssproj for this fsspset
		 */
		fssproj->fssp_next = fssproj->fssp_prev = fssproj;
		fsspset->fssps_list = fssproj;
	} else {
		/*
		 * Insert this fssproj to the doubly linked list.
		 */
		fssproj_t *fssp_head = fsspset->fssps_list;

		fssproj->fssp_next = fssp_head;
		fssproj->fssp_prev = fssp_head->fssp_prev;
		fssp_head->fssp_prev->fssp_next = fssproj;
		fssp_head->fssp_prev = fssproj;
		fsspset->fssps_list = fssproj;
	}
	fssproj->fssp_fsszone = fsszone;
	fsszone->fssz_nproj++;
	ASSERT(fsszone->fssz_nproj != 0);
}

/*
 * The following routine removes a single fssproj structure from the doubly
 * linked list of projects running on the specified cpu partition.  Note that
 * global fsspsets_lock must be held in case if this fssproj structure is the
 * last on the above mentioned list.  Also note that the fssproj structure is
 * not freed here, it is the responsibility of the caller to call kmem_free
 * for it.
 */
static void
fss_remove_fssproj(fsspset_t *fsspset, fssproj_t *fssproj)
{
	fsszone_t *fsszone;

	ASSERT(MUTEX_HELD(&fsspsets_lock));
	ASSERT(MUTEX_HELD(&fsspset->fssps_lock));
	ASSERT(fssproj->fssp_runnable == 0);

	fsspset->fssps_nproj--;

	fsszone = fssproj->fssp_fsszone;
	fsszone->fssz_nproj--;

	if (fssproj->fssp_next != fssproj) {
		/*
		 * This is not the last part in the list.
		 */
		fssproj->fssp_prev->fssp_next = fssproj->fssp_next;
		fssproj->fssp_next->fssp_prev = fssproj->fssp_prev;
		if (fsspset->fssps_list == fssproj)
			fsspset->fssps_list = fssproj->fssp_next;
		if (fsszone->fssz_nproj == 0)
			fss_remove_fsszone(fsspset, fsszone);
	} else {
		/*
		 * This was the last project part running
		 * at this cpu partition.
		 */
		fsspset->fssps_list = NULL;
		ASSERT(fsspset->fssps_nproj == 0);
		ASSERT(fsszone->fssz_nproj == 0);
		fss_remove_fsszone(fsspset, fsszone);
		fss_del_fsspset(fsspset);
	}
}

static void
fss_inactive(kthread_t *t)
{
	fssproc_t *fssproc;
	fssproj_t *fssproj;
	fsspset_t *fsspset;
	fsszone_t *fsszone;

	ASSERT(THREAD_LOCK_HELD(t));
	fssproc = FSSPROC(t);
	fssproj = FSSPROC2FSSPROJ(fssproc);
	if (fssproj == NULL)	/* if this thread already exited */
		return;
	fsspset = FSSPROJ2FSSPSET(fssproj);
	fsszone = fssproj->fssp_fsszone;
	disp_lock_enter_high(&fsspset->fssps_displock);
	ASSERT(fssproj->fssp_runnable > 0);
	if (--fssproj->fssp_runnable == 0) {
		fsszone->fssz_shares -= fssproj->fssp_shares;
		if (--fsszone->fssz_runnable == 0)
			fsspset->fssps_shares -= fsszone->fssz_rshares;
	}
	ASSERT(fssproc->fss_runnable == 1);
	fssproc->fss_runnable = 0;
	disp_lock_exit_high(&fsspset->fssps_displock);
}

static void
fss_active(kthread_t *t)
{
	fssproc_t *fssproc;
	fssproj_t *fssproj;
	fsspset_t *fsspset;
	fsszone_t *fsszone;

	ASSERT(THREAD_LOCK_HELD(t));
	fssproc = FSSPROC(t);
	fssproj = FSSPROC2FSSPROJ(fssproc);
	if (fssproj == NULL)	/* if this thread already exited */
		return;
	fsspset = FSSPROJ2FSSPSET(fssproj);
	fsszone = fssproj->fssp_fsszone;
	disp_lock_enter_high(&fsspset->fssps_displock);
	if (++fssproj->fssp_runnable == 1) {
		fsszone->fssz_shares += fssproj->fssp_shares;
		if (++fsszone->fssz_runnable == 1)
			fsspset->fssps_shares += fsszone->fssz_rshares;
	}
	ASSERT(fssproc->fss_runnable == 0);
	fssproc->fss_runnable = 1;
	disp_lock_exit_high(&fsspset->fssps_displock);
}

/*
 * Fair share scheduler initialization. Called by dispinit() at boot time.
 * We can ignore clparmsz argument since we know that the smallest possible
 * parameter buffer is big enough for us.
 */
/*ARGSUSED*/
static pri_t
fss_init(id_t cid, int clparmsz, classfuncs_t **clfuncspp)
{
	int i;

	ASSERT(MUTEX_HELD(&cpu_lock));

	fss_cid = cid;
	fss_maxumdpri = minclsyspri - 1;
	fss_maxglobpri = minclsyspri;
	fss_minglobpri = 0;
	fsspsets = kmem_zalloc(sizeof (fsspset_t) * max_ncpus, KM_SLEEP);

	/*
	 * Initialize the fssproc hash table.
	 */
	for (i = 0; i < FSS_LISTS; i++)
		fss_listhead[i].fss_next = fss_listhead[i].fss_prev =
		    &fss_listhead[i];

	*clfuncspp = &fss_classfuncs;

	/*
	 * Fill in fss_nice_tick and fss_nice_decay arrays:
	 * The cost of a tick is lower at positive nice values (so that it
	 * will not increase its project's usage as much as normal) with 50%
	 * drop at the maximum level and 50% increase at the minimum level.
	 * The fsspri decay is slower at positive nice values.  fsspri values
	 * of processes with negative nice levels must decay faster to receive
	 * time slices more frequently than normal.
	 */
	for (i = 0; i < FSS_NICE_RANGE; i++) {
		fss_nice_tick[i] = (FSS_TICK_COST * (((3 * FSS_NICE_RANGE) / 2)
		    - i)) / FSS_NICE_RANGE;
		fss_nice_decay[i] = FSS_DECAY_MIN +
		    ((FSS_DECAY_MAX - FSS_DECAY_MIN) * i) /
		    (FSS_NICE_RANGE - 1);
	}

	return (fss_maxglobpri);
}

/*
 * Calculate the new cpupri based on the usage, the number of shares and
 * the number of active threads.  Reset the tick counter for this thread.
 */
static void
fss_newpri(fssproc_t *fssproc)
{
	kthread_t *tp;
	fssproj_t *fssproj;
	fsspset_t *fsspset;
	fsszone_t *fsszone;
	fsspri_t fsspri, maxfsspri;
	pri_t invpri;
	uint32_t ticks;

	tp = fssproc->fss_tp;
	ASSERT(tp != NULL);

	if (tp->t_cid != fss_cid)
		return;

	ASSERT(THREAD_LOCK_HELD(tp));

	fssproj = FSSPROC2FSSPROJ(fssproc);
	fsszone = FSSPROJ2FSSZONE(fssproj);
	if (fssproj == NULL)
		/*
		 * No need to change priority of exited threads.
		 */
		return;

	fsspset = FSSPROJ2FSSPSET(fssproj);
	disp_lock_enter_high(&fsspset->fssps_displock);

	if (fssproj->fssp_shares == 0 || fsszone->fssz_rshares == 0) {
		/*
		 * Special case: threads with no shares.
		 */
		fssproc->fss_umdpri = fss_minglobpri;
		fssproc->fss_ticks = 0;
		disp_lock_exit_high(&fsspset->fssps_displock);
		return;
	}

	/*
	 * fsspri += shusage * nrunnable * ticks
	 */
	ticks = fssproc->fss_ticks;
	fssproc->fss_ticks = 0;
	fsspri = fssproc->fss_fsspri;
	fsspri += fssproj->fssp_shusage * fssproj->fssp_runnable * ticks;
	fssproc->fss_fsspri = fsspri;

	if (fsspri < fss_maxumdpri)
		fsspri = fss_maxumdpri;	/* so that maxfsspri is != 0 */

	/*
	 * The general priority formula:
	 *
	 *			(fsspri * umdprirange)
	 *   pri = maxumdpri - ------------------------
	 *				maxfsspri
	 *
	 * If this thread's fsspri is greater than the previous largest
	 * fsspri, then record it as the new high and priority for this
	 * thread will be one (the lowest priority assigned to a thread
	 * that has non-zero shares).
	 * Note that this formula cannot produce out of bounds priority
	 * values; if it is changed, additional checks may need  to  be
	 * added.
	 */
	maxfsspri = fsspset->fssps_maxfsspri;
	if (fsspri >= maxfsspri) {
		fsspset->fssps_maxfsspri = fsspri;
		disp_lock_exit_high(&fsspset->fssps_displock);
		fssproc->fss_umdpri = 1;
	} else {
		disp_lock_exit_high(&fsspset->fssps_displock);
		invpri = (fsspri * (fss_maxumdpri - 1)) / maxfsspri;
		fssproc->fss_umdpri = fss_maxumdpri - invpri;
	}
}

/*
 * Decays usages of all running projects and resets their tick counters.
 * Called once per second from fss_update() after updating priorities.
 */
static void
fss_decay_usage()
{
	uint32_t zone_ext_shares, zone_int_shares;
	uint32_t kpj_shares, pset_shares;
	fsspset_t *fsspset;
	fssproj_t *fssproj;
	fsszone_t *fsszone;
	fsspri_t maxfsspri;
	int psetid;

	mutex_enter(&fsspsets_lock);
	/*
	 * Go through all active processor sets and decay usages of projects
	 * running on them.
	 */
	for (psetid = 0; psetid < max_ncpus; psetid++) {
		fsspset = &fsspsets[psetid];
		mutex_enter(&fsspset->fssps_lock);

		if (fsspset->fssps_cpupart == NULL ||
		    (fssproj = fsspset->fssps_list) == NULL) {
			mutex_exit(&fsspset->fssps_lock);
			continue;
		}

		/*
		 * Decay maxfsspri for this cpu partition with the
		 * fastest possible decay rate.
		 */
		disp_lock_enter(&fsspset->fssps_displock);

		maxfsspri = (fsspset->fssps_maxfsspri *
		    fss_nice_decay[NZERO]) / FSS_DECAY_BASE;
		if (maxfsspri < fss_maxumdpri)
			maxfsspri = fss_maxumdpri;
		fsspset->fssps_maxfsspri = maxfsspri;

		do {
			/*
			 * Decay usage for each project running on
			 * this cpu partition.
			 */
			fssproj->fssp_usage =
			    (fssproj->fssp_usage * FSS_DECAY_USG) /
			    FSS_DECAY_BASE + fssproj->fssp_ticks;
			fssproj->fssp_ticks = 0;

			fsszone = fssproj->fssp_fsszone;
			/*
			 * Readjust the project's number of shares if it has
			 * changed since we checked it last time.
			 */
			kpj_shares = fssproj->fssp_proj->kpj_shares;
			if (fssproj->fssp_shares != kpj_shares) {
				if (fssproj->fssp_runnable != 0) {
					fsszone->fssz_shares -=
					    fssproj->fssp_shares;
					fsszone->fssz_shares += kpj_shares;
				}
				fssproj->fssp_shares = kpj_shares;
			}

			/*
			 * Readjust the zone's number of shares if it
			 * has changed since we checked it last time.
			 */
			zone_ext_shares = fsszone->fssz_zone->zone_shares;
			if (fsszone->fssz_rshares != zone_ext_shares) {
				if (fsszone->fssz_runnable != 0) {
					fsspset->fssps_shares -=
					    fsszone->fssz_rshares;
					fsspset->fssps_shares +=
					    zone_ext_shares;
				}
				fsszone->fssz_rshares = zone_ext_shares;
			}
			zone_int_shares = fsszone->fssz_shares;
			pset_shares = fsspset->fssps_shares;
			/*
			 * Calculate fssp_shusage value to be used
			 * for fsspri increments for the next second.
			 */
			if (kpj_shares == 0 || zone_ext_shares == 0) {
				fssproj->fssp_shusage = 0;
			} else if (FSSPROJ2KPROJ(fssproj) == proj0p) {
				/*
				 * Project 0 in the global zone has 50%
				 * of its zone.
				 */
				fssproj->fssp_shusage = (fssproj->fssp_usage *
				    zone_int_shares * zone_int_shares) /
				    (zone_ext_shares * zone_ext_shares);
			} else {
				/*
				 * Thread's priority is based on its project's
				 * normalized usage (shusage) value which gets
				 * calculated this way:
				 *
				 *	   pset_shares^2    zone_int_shares^2
				 * usage * ------------- * ------------------
				 *	   kpj_shares^2	    zone_ext_shares^2
				 *
				 * Where zone_int_shares is the sum of shares
				 * of all active projects within the zone (and
				 * the pset), and zone_ext_shares is the number
				 * of zone shares (ie, zone.cpu-shares).
				 *
				 * If there is only one zone active on the pset
				 * the above reduces to:
				 *
				 * 			zone_int_shares^2
				 * shusage = usage * ---------------------
				 * 			kpj_shares^2
				 *
				 * If there's only one project active in the
				 * zone this formula reduces to:
				 *
				 *			pset_shares^2
				 * shusage = usage * ----------------------
				 *			zone_ext_shares^2
				 */
				fssproj->fssp_shusage = fssproj->fssp_usage *
				    pset_shares * zone_int_shares;
				fssproj->fssp_shusage /=
				    kpj_shares * zone_ext_shares;
				fssproj->fssp_shusage *=
				    pset_shares * zone_int_shares;
				fssproj->fssp_shusage /=
				    kpj_shares * zone_ext_shares;
			}
			fssproj = fssproj->fssp_next;
		} while (fssproj != fsspset->fssps_list);

		disp_lock_exit(&fsspset->fssps_displock);
		mutex_exit(&fsspset->fssps_lock);
	}
	mutex_exit(&fsspsets_lock);
}

static void
fss_change_priority(kthread_t *t, fssproc_t *fssproc)
{
	pri_t new_pri;

	ASSERT(THREAD_LOCK_HELD(t));
	new_pri = fssproc->fss_umdpri;
	ASSERT(new_pri >= 0 && new_pri <= fss_maxglobpri);

	t->t_cpri = fssproc->fss_upri;
	fssproc->fss_flags &= ~FSSRESTORE;
	if (t == curthread || t->t_state == TS_ONPROC) {
		/*
		 * curthread is always onproc
		 */
		cpu_t *cp = t->t_disp_queue->disp_cpu;
		THREAD_CHANGE_PRI(t, new_pri);
		if (t == cp->cpu_dispthread)
			cp->cpu_dispatch_pri = DISP_PRIO(t);
		if (DISP_MUST_SURRENDER(t)) {
			fssproc->fss_flags |= FSSBACKQ;
			cpu_surrender(t);
		} else {
			fssproc->fss_timeleft = fss_quantum;
		}
	} else {
		/*
		 * When the priority of a thread is changed, it may be
		 * necessary to adjust its position on a sleep queue or
		 * dispatch queue.  The function thread_change_pri accomplishes
		 * this.
		 */
		if (thread_change_pri(t, new_pri, 0)) {
			/*
			 * The thread was on a run queue.
			 */
			fssproc->fss_timeleft = fss_quantum;
		} else {
			fssproc->fss_flags |= FSSBACKQ;
		}
	}
}

/*
 * Update priorities of all fair-sharing threads that are currently runnable
 * at a user mode priority based on the number of shares and current usage.
 * Called once per second via timeout which we reset here.
 *
 * There are several lists of fair-sharing threads broken up by a hash on the
 * thread pointer.  Each list has its own lock.  This avoids blocking all
 * fss_enterclass, fss_fork, and fss_exitclass operations while fss_update runs.
 * fss_update traverses each list in turn.
 */
static void
fss_update(void *arg)
{
	int i;
	int new_marker = -1;
	static int fss_update_marker;

	/*
	 * Decay and update usages for all projects.
	 */
	fss_decay_usage();

	/*
	 * Start with the fss_update_marker list, then do the rest.
	 */
	i = fss_update_marker;

	/*
	 * Go around all threads, set new priorities and decay
	 * per-thread CPU usages.
	 */
	do {
		/*
		 * If this is the first list after the current marker to have
		 * threads with priorities updates, advance the marker to this
		 * list for the next time fss_update runs.
		 */
		if (fss_update_list(i) &&
		    new_marker == -1 && i != fss_update_marker)
			new_marker = i;
	} while ((i = FSS_LIST_NEXT(i)) != fss_update_marker);

	/*
	 * Advance marker for the next fss_update call
	 */
	if (new_marker != -1)
		fss_update_marker = new_marker;

	(void) timeout(fss_update, arg, hz);
}

/*
 * Updates priority for a list of threads.  Returns 1 if the priority of one
 * of the threads was actually updated, 0 if none were for various reasons
 * (thread is no longer in the FSS class, is not runnable, has the preemption
 * control no-preempt bit set, etc.)
 */
static int
fss_update_list(int i)
{
	fssproc_t *fssproc;
	fssproj_t *fssproj;
	fsspri_t fsspri;
	kthread_t *t;
	int updated = 0;

	mutex_enter(&fss_listlock[i]);
	for (fssproc = fss_listhead[i].fss_next; fssproc != &fss_listhead[i];
	    fssproc = fssproc->fss_next) {
		t = fssproc->fss_tp;
		/*
		 * Lock the thread and verify the state.
		 */
		thread_lock(t);
		/*
		 * Skip the thread if it is no longer in the FSS class or
		 * is running with kernel mode priority.
		 */
		if (t->t_cid != fss_cid)
			goto next;
		if ((fssproc->fss_flags & FSSKPRI) != 0)
			goto next;

		fssproj = FSSPROC2FSSPROJ(fssproc);
		if (fssproj == NULL)
			goto next;
		if (fssproj->fssp_shares != 0) {
			/*
			 * Decay fsspri value.
			 */
			fsspri = fssproc->fss_fsspri;
			fsspri = (fsspri * fss_nice_decay[fssproc->fss_nice]) /
			    FSS_DECAY_BASE;
			fssproc->fss_fsspri = fsspri;
		}

		if (t->t_schedctl && schedctl_get_nopreempt(t))
			goto next;
		if (t->t_state != TS_RUN && t->t_state != TS_WAIT) {
			/*
			 * Make next syscall/trap call fss_trapret
			 */
			t->t_trapret = 1;
			aston(t);
			goto next;
		}
		fss_newpri(fssproc);
		updated = 1;

		/*
		 * Only dequeue the thread if it needs to be moved; otherwise
		 * it should just round-robin here.
		 */
		if (t->t_pri != fssproc->fss_umdpri)
			fss_change_priority(t, fssproc);
next:
		thread_unlock(t);
	}
	mutex_exit(&fss_listlock[i]);
	return (updated);
}

/*ARGSUSED*/
static int
fss_admin(caddr_t uaddr