ts.c revision 7c478bd95313f5f23a4c958a745db2134aa03244
/*
* CDDL HEADER START
*
* The contents of this file are subject to the terms of the
* Common Development and Distribution License, Version 1.0 only
* (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 2004 Sun Microsystems, Inc. All rights reserved.
* Use is subject to license terms.
*/
/* Copyright (c) 1984, 1986, 1987, 1988, 1989 AT&T */
/* All Rights Reserved */
#pragma ident "%Z%%M% %I% %E% SMI" /* from SVr4.0 1.23 */
#include <sys/types.h>
#include <sys/param.h>
#include <sys/sysmacros.h>
#include <sys/cred.h>
#include <sys/proc.h>
#include <sys/session.h>
#include <sys/strsubr.h>
#include <sys/signal.h>
#include <sys/user.h>
#include <sys/priocntl.h>
#include <sys/class.h>
#include <sys/disp.h>
#include <sys/procset.h>
#include <sys/debug.h>
#include <sys/ts.h>
#include <sys/tspriocntl.h>
#include <sys/iapriocntl.h>
#include <sys/kmem.h>
#include <sys/errno.h>
#include <sys/cpuvar.h>
#include <sys/systm.h> /* for lbolt */
#include <sys/vtrace.h>
#include <sys/vmsystm.h>
#include <sys/schedctl.h>
#include <sys/tnf_probe.h>
#include <sys/atomic.h>
#include <sys/policy.h>
#include <sys/sdt.h>
#include <sys/cpupart.h>
#include <vm/rm.h>
#include <vm/seg_kmem.h>
#include <sys/modctl.h>
static pri_t ts_init(id_t, int, classfuncs_t **);
static struct sclass csw = {
"TS",
ts_init,
0
};
static struct modlsched modlsched = {
&mod_schedops, "time sharing sched class", &csw
};
static struct modlinkage modlinkage = {
MODREV_1, (void *)&modlsched, NULL
};
int
_init()
{
return (mod_install(&modlinkage));
}
int
_fini()
{
return (EBUSY); /* don't remove TS for now */
}
int
_info(struct modinfo *modinfop)
{
return (mod_info(&modlinkage, modinfop));
}
/*
* Class specific code for the time-sharing class
*/
/*
* Extern declarations for variables defined in the ts master file
*/
#define TSMAXUPRI 60
pri_t ts_maxupri = TSMAXUPRI; /* max time-sharing user priority */
pri_t ts_maxumdpri; /* maximum user mode ts priority */
pri_t ia_maxupri = IAMAXUPRI; /* max interactive user priority */
pri_t ia_boost = IA_BOOST; /* boost value for interactive */
tsdpent_t *ts_dptbl; /* time-sharing disp parameter table */
pri_t *ts_kmdpris; /* array of global pris used by ts procs when */
/* sleeping or running in kernel after sleep */
static id_t ia_cid;
int ts_sleep_promote = 1;
#define tsmedumdpri (ts_maxumdpri >> 1)
#define TS_NEWUMDPRI(tspp) \
{ \
pri_t pri; \
pri = (tspp)->ts_cpupri + (tspp)->ts_upri + (tspp)->ts_boost; \
if (pri > ts_maxumdpri) \
(tspp)->ts_umdpri = ts_maxumdpri; \
else if (pri < 0) \
(tspp)->ts_umdpri = 0; \
else \
(tspp)->ts_umdpri = pri; \
ASSERT((tspp)->ts_umdpri >= 0 && (tspp)->ts_umdpri <= ts_maxumdpri); \
}
/*
* The tsproc_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. Each list has a dummy "head" which is
* never removed, so the list is never empty. ts_update traverses these
* lists to update the priorities of threads that have been waiting on
* the run queue.
*/
#define TS_LISTS 16 /* number of lists, must be power of 2 */
/* hash function, argument is a thread pointer */
#define TS_LIST_HASH(tp) (((uintptr_t)(tp) >> 9) & (TS_LISTS - 1))
/* iterate to the next list */
#define TS_LIST_NEXT(i) (((i) + 1) & (TS_LISTS - 1))
/*
* Insert thread into the appropriate tsproc list.
*/
#define TS_LIST_INSERT(tspp) \
{ \
int index = TS_LIST_HASH(tspp->ts_tp); \
kmutex_t *lockp = &ts_list_lock[index]; \
tsproc_t *headp = &ts_plisthead[index]; \
mutex_enter(lockp); \
tspp->ts_next = headp->ts_next; \
tspp->ts_prev = headp; \
headp->ts_next->ts_prev = tspp; \
headp->ts_next = tspp; \
mutex_exit(lockp); \
}
/*
* Remove thread from tsproc list.
*/
#define TS_LIST_DELETE(tspp) \
{ \
int index = TS_LIST_HASH(tspp->ts_tp); \
kmutex_t *lockp = &ts_list_lock[index]; \
mutex_enter(lockp); \
tspp->ts_prev->ts_next = tspp->ts_next; \
tspp->ts_next->ts_prev = tspp->ts_prev; \
mutex_exit(lockp); \
}
static int ts_admin(caddr_t, cred_t *);
static int ts_enterclass(kthread_t *, id_t, void *, cred_t *, void *);
static int ts_fork(kthread_t *, kthread_t *, void *);
static int ts_getclinfo(void *);
static int ts_getclpri(pcpri_t *);
static int ts_parmsin(void *);
static int ts_parmsout(void *, pc_vaparms_t *);
static int ts_vaparmsin(void *, pc_vaparms_t *);
static int ts_vaparmsout(void *, pc_vaparms_t *);
static int ts_parmsset(kthread_t *, void *, id_t, cred_t *);
static int ts_donice(kthread_t *, cred_t *, int, int *);
static void ts_exitclass(void *);
static int ts_canexit(kthread_t *, cred_t *);
static void ts_forkret(kthread_t *, kthread_t *);
static void ts_nullsys();
static void ts_parmsget(kthread_t *, void *);
static void ts_preempt(kthread_t *);
static void ts_setrun(kthread_t *);
static void ts_sleep(kthread_t *);
static pri_t ts_swapin(kthread_t *, int);
static pri_t ts_swapout(kthread_t *, int);
static void ts_tick(kthread_t *);
static void ts_trapret(kthread_t *);
static void ts_update(void *);
static int ts_update_list(int);
static void ts_wakeup(kthread_t *);
static pri_t ts_globpri(kthread_t *);
static void ts_yield(kthread_t *);
extern tsdpent_t *ts_getdptbl(void);
extern pri_t *ts_getkmdpris(void);
extern pri_t td_getmaxumdpri(void);
static int ts_alloc(void **, int);
static void ts_free(void *);
pri_t ia_init(id_t, int, classfuncs_t **);
static int ia_getclinfo(void *);
static int ia_parmsin(void *);
static int ia_vaparmsin(void *, pc_vaparms_t *);
static int ia_vaparmsout(void *, pc_vaparms_t *);
static int ia_parmsset(kthread_t *, void *, id_t, cred_t *);
static void ia_parmsget(kthread_t *, void *);
static void ia_set_process_group(pid_t, pid_t, pid_t);
static void ts_change_priority(kthread_t *, tsproc_t *);
extern pri_t ts_maxkmdpri; /* maximum kernel mode ts priority */
static pri_t ts_maxglobpri; /* maximum global priority used by ts class */
static kmutex_t ts_dptblock; /* protects time sharing dispatch table */
static kmutex_t ts_list_lock[TS_LISTS]; /* protects tsproc lists */
static tsproc_t ts_plisthead[TS_LISTS]; /* dummy tsproc at head of lists */
static gid_t IA_gid = 0;
static struct classfuncs ts_classfuncs = {
/* class functions */
ts_admin,
ts_getclinfo,
ts_parmsin,
ts_parmsout,
ts_vaparmsin,
ts_vaparmsout,
ts_getclpri,
ts_alloc,
ts_free,
/* thread functions */
ts_enterclass,
ts_exitclass,
ts_canexit,
ts_fork,
ts_forkret,
ts_parmsget,
ts_parmsset,
ts_nullsys, /* stop */
ts_nullsys, /* exit */
ts_nullsys, /* active */
ts_nullsys, /* inactive */
ts_swapin,
ts_swapout,
ts_trapret,
ts_preempt,
ts_setrun,
ts_sleep,
ts_tick,
ts_wakeup,
ts_donice,
ts_globpri,
ts_nullsys, /* set_process_group */
ts_yield,
};
/*
* ia_classfuncs is used for interactive class threads; IA threads are stored
* on the same class list as TS threads, and most of the class functions are
* identical, but a few have different enough functionality to require their
* own functions.
*/
static struct classfuncs ia_classfuncs = {
/* class functions */
ts_admin,
ia_getclinfo,
ia_parmsin,
ts_parmsout,
ia_vaparmsin,
ia_vaparmsout,
ts_getclpri,
ts_alloc,
ts_free,
/* thread functions */
ts_enterclass,
ts_exitclass,
ts_canexit,
ts_fork,
ts_forkret,
ia_parmsget,
ia_parmsset,
ts_nullsys, /* stop */
ts_nullsys, /* exit */
ts_nullsys, /* active */
ts_nullsys, /* inactive */
ts_swapin,
ts_swapout,
ts_trapret,
ts_preempt,
ts_setrun,
ts_sleep,
ts_tick,
ts_wakeup,
ts_donice,
ts_globpri,
ia_set_process_group,
ts_yield,
};
/*
* Time sharing class initialization. Called by dispinit() at boot time.
* We can ignore the clparmsz argument since we know that the smallest
* possible parameter buffer is big enough for us.
*/
/* ARGSUSED */
static pri_t
ts_init(id_t cid, int clparmsz, classfuncs_t **clfuncspp)
{
int i;
extern pri_t ts_getmaxumdpri(void);
ts_dptbl = ts_getdptbl();
ts_kmdpris = ts_getkmdpris();
ts_maxumdpri = ts_getmaxumdpri();
ts_maxglobpri = MAX(ts_kmdpris[0], ts_dptbl[ts_maxumdpri].ts_globpri);
/*
* Initialize the tsproc lists.
*/
for (i = 0; i < TS_LISTS; i++) {
ts_plisthead[i].ts_next = ts_plisthead[i].ts_prev =
&ts_plisthead[i];
}
/*
* We're required to return a pointer to our classfuncs
* structure and the highest global priority value we use.
*/
*clfuncspp = &ts_classfuncs;
return (ts_maxglobpri);
}
/*
* Interactive class scheduler initialization
*/
/* ARGSUSED */
pri_t
ia_init(id_t cid, int clparmsz, classfuncs_t **clfuncspp)
{
/*
* We're required to return a pointer to our classfuncs
* structure and the highest global priority value we use.
*/
ia_cid = cid;
*clfuncspp = &ia_classfuncs;
return (ts_maxglobpri);
}
/*
* Get or reset the ts_dptbl values per the user's request.
*/
static int
ts_admin(caddr_t uaddr, cred_t *reqpcredp)
{
tsadmin_t tsadmin;
tsdpent_t *tmpdpp;
int userdpsz;
int i;
size_t tsdpsz;
if (get_udatamodel() == DATAMODEL_NATIVE) {
if (copyin(uaddr, &tsadmin, sizeof (tsadmin_t)))
return (EFAULT);
}
#ifdef _SYSCALL32_IMPL
else {
/* get tsadmin struct from ILP32 caller */
tsadmin32_t tsadmin32;
if (copyin(uaddr, &tsadmin32, sizeof (tsadmin32_t)))
return (EFAULT);
tsadmin.ts_dpents =
(struct tsdpent *)(uintptr_t)tsadmin32.ts_dpents;
tsadmin.ts_ndpents = tsadmin32.ts_ndpents;
tsadmin.ts_cmd = tsadmin32.ts_cmd;
}
#endif /* _SYSCALL32_IMPL */
tsdpsz = (ts_maxumdpri + 1) * sizeof (tsdpent_t);
switch (tsadmin.ts_cmd) {
case TS_GETDPSIZE:
tsadmin.ts_ndpents = ts_maxumdpri + 1;
if (get_udatamodel() == DATAMODEL_NATIVE) {
if (copyout(&tsadmin, uaddr, sizeof (tsadmin_t)))
return (EFAULT);
}
#ifdef _SYSCALL32_IMPL
else {
/* return tsadmin struct to ILP32 caller */
tsadmin32_t tsadmin32;
tsadmin32.ts_dpents =
(caddr32_t)(uintptr_t)tsadmin.ts_dpents;
tsadmin32.ts_ndpents = tsadmin.ts_ndpents;
tsadmin32.ts_cmd = tsadmin.ts_cmd;
if (copyout(&tsadmin32, uaddr, sizeof (tsadmin32_t)))
return (EFAULT);
}
#endif /* _SYSCALL32_IMPL */
break;
case TS_GETDPTBL:
userdpsz = MIN(tsadmin.ts_ndpents * sizeof (tsdpent_t),
tsdpsz);
if (copyout(ts_dptbl, tsadmin.ts_dpents, userdpsz))
return (EFAULT);
tsadmin.ts_ndpents = userdpsz / sizeof (tsdpent_t);
if (get_udatamodel() == DATAMODEL_NATIVE) {
if (copyout(&tsadmin, uaddr, sizeof (tsadmin_t)))
return (EFAULT);
}
#ifdef _SYSCALL32_IMPL
else {
/* return tsadmin struct to ILP32 callers */
tsadmin32_t tsadmin32;
tsadmin32.ts_dpents =
(caddr32_t)(uintptr_t)tsadmin.ts_dpents;
tsadmin32.ts_ndpents = tsadmin.ts_ndpents;
tsadmin32.ts_cmd = tsadmin.ts_cmd;
if (copyout(&tsadmin32, uaddr, sizeof (tsadmin32_t)))
return (EFAULT);
}
#endif /* _SYSCALL32_IMPL */
break;
case TS_SETDPTBL:
/*
* We require that the requesting process has sufficient
* priveleges. We also require that the table supplied by
* the user exactly match the current ts_dptbl in size.
*/
if (secpolicy_dispadm(reqpcredp) != 0)
return (EPERM);
if (tsadmin.ts_ndpents * sizeof (tsdpent_t) != tsdpsz) {
return (EINVAL);
}
/*
* We read the user supplied table into a temporary buffer
* where it is validated before being copied over the
* ts_dptbl.
*/
tmpdpp = kmem_alloc(tsdpsz, KM_SLEEP);
if (copyin((caddr_t)tsadmin.ts_dpents, (caddr_t)tmpdpp,
tsdpsz)) {
kmem_free(tmpdpp, tsdpsz);
return (EFAULT);
}
for (i = 0; i < tsadmin.ts_ndpents; i++) {
/*
* Validate the user supplied values. All we are doing
* here is verifying that the values are within their
* allowable ranges and will not panic the system. We
* make no attempt to ensure that the resulting
* configuration makes sense or results in reasonable
* performance.
*/
if (tmpdpp[i].ts_quantum <= 0) {
kmem_free(tmpdpp, tsdpsz);
return (EINVAL);
}
if (tmpdpp[i].ts_tqexp > ts_maxumdpri ||
tmpdpp[i].ts_tqexp < 0) {
kmem_free(tmpdpp, tsdpsz);
return (EINVAL);
}
if (tmpdpp[i].ts_slpret > ts_maxumdpri ||
tmpdpp[i].ts_slpret < 0) {
kmem_free(tmpdpp, tsdpsz);
return (EINVAL);
}
if (tmpdpp[i].ts_maxwait < 0) {
kmem_free(tmpdpp, tsdpsz);
return (EINVAL);
}
if (tmpdpp[i].ts_lwait > ts_maxumdpri ||
tmpdpp[i].ts_lwait < 0) {
kmem_free(tmpdpp, tsdpsz);
return (EINVAL);
}
}
/*
* Copy the user supplied values over the current ts_dptbl
* values. The ts_globpri member is read-only so we don't
* overwrite it.
*/
mutex_enter(&ts_dptblock);
for (i = 0; i < tsadmin.ts_ndpents; i++) {
ts_dptbl[i].ts_quantum = tmpdpp[i].ts_quantum;
ts_dptbl[i].ts_tqexp = tmpdpp[i].ts_tqexp;
ts_dptbl[i].ts_slpret = tmpdpp[i].ts_slpret;
ts_dptbl[i].ts_maxwait = tmpdpp[i].ts_maxwait;
ts_dptbl[i].ts_lwait = tmpdpp[i].ts_lwait;
}
mutex_exit(&ts_dptblock);
kmem_free(tmpdpp, tsdpsz);
break;
default:
return (EINVAL);
}
return (0);
}
/*
* Allocate a time-sharing class specific thread structure and
* initialize it with the parameters supplied. Also move the thread
* to specified time-sharing priority.
*/
static int
ts_enterclass(kthread_t *t, id_t cid, void *parmsp,
cred_t *reqpcredp, void *bufp)
{
tsparms_t *tsparmsp = (tsparms_t *)parmsp;
tsproc_t *tspp;
pri_t reqtsuprilim;
pri_t reqtsupri;
static uint32_t tspexists = 0; /* set on first occurrence of */
/* a time-sharing process */
tspp = (tsproc_t *)bufp;
ASSERT(tspp != NULL);
/*
* Initialize the tsproc structure.
*/
tspp->ts_cpupri = tsmedumdpri;
if (cid == ia_cid) {
/*
* Check to make sure caller is either privileged or the
* window system. When the window system is converted
* to using privileges, the second check can go away.
*/
if (reqpcredp != NULL && !groupmember(IA_gid, reqpcredp) &&
secpolicy_setpriority(reqpcredp) != 0)
return (EPERM);
/*
* Belongs to IA "class", so set appropriate flags.
* Mark as 'on' so it will not be a swap victim
* while forking.
*/
tspp->ts_flags = TSIA | TSIASET;
tspp->ts_boost = ia_boost;
} else {
tspp->ts_flags = 0;
tspp->ts_boost = 0;
}
if (tsparmsp == NULL) {
/*
* Use default values.
*/
tspp->ts_uprilim = tspp->ts_upri = 0;
tspp->ts_nice = NZERO;
} else {
/*
* Use supplied values.
*/
if (tsparmsp->ts_uprilim == TS_NOCHANGE)
reqtsuprilim = 0;
else {
if (tsparmsp->ts_uprilim > 0 &&
secpolicy_setpriority(reqpcredp) != 0)
return (EPERM);
reqtsuprilim = tsparmsp->ts_uprilim;
}
if (tsparmsp->ts_upri == TS_NOCHANGE) {
reqtsupri = reqtsuprilim;
} else {
if (tsparmsp->ts_upri > 0 &&
secpolicy_setpriority(reqpcredp) != 0)
return (EPERM);
/*
* Set the user priority to the requested value
* or the upri limit, whichever is lower.
*/
reqtsupri = tsparmsp->ts_upri;
if (reqtsupri > reqtsuprilim)
reqtsupri = reqtsuprilim;
}
tspp->ts_uprilim = reqtsuprilim;
tspp->ts_upri = reqtsupri;
tspp->ts_nice = NZERO - (NZERO * reqtsupri)
/ ts_maxupri;
}
TS_NEWUMDPRI(tspp);
tspp->ts_dispwait = 0;
tspp->ts_timeleft = ts_dptbl[tspp->ts_cpupri].ts_quantum;
tspp->ts_tp = t;
/*
* Reset priority. Process goes to a "user mode" priority
* here regardless of whether or not it has slept since
* entering the kernel.
*/
thread_lock(t); /* get dispatcher lock on thread */
t->t_clfuncs = &(sclass[cid].cl_funcs->thread);
t->t_cid = cid;
t->t_cldata = (void *)tspp;
t->t_schedflag &= ~TS_RUNQMATCH;
ts_change_priority(t, tspp);
thread_unlock(t);
/*
* Link new structure into tsproc list.
*/
TS_LIST_INSERT(tspp);
/*
* If this is the first time-sharing thread to occur since
* boot we set up the initial call to ts_update() here.
* Use an atomic compare-and-swap since that's easier and
* faster than a mutex (but check with an ordinary load first
* since most of the time this will already be done).
*/
if (tspexists == 0 && cas32(&tspexists, 0, 1) == 0)
(void) timeout(ts_update, NULL, hz);
return (0);
}
/*
* Free tsproc structure of thread.
*/
static void
ts_exitclass(void *procp)
{
tsproc_t *tspp = (tsproc_t *)procp;
/* Remove tsproc_t structure from list */
TS_LIST_DELETE(tspp);
kmem_free(tspp, sizeof (tsproc_t));
}
/* ARGSUSED */
static int
ts_canexit(kthread_t *t, cred_t *cred)
{
/*
* A thread can always leave a TS/IA class
*/
return (0);
}
static int
ts_fork(kthread_t *t, kthread_t *ct, void *bufp)
{
tsproc_t *ptspp; /* ptr to parent's tsproc structure */
tsproc_t *ctspp; /* ptr to child's tsproc structure */
ASSERT(MUTEX_HELD(&ttoproc(t)->p_lock));
ctspp = (tsproc_t *)bufp;
ASSERT(ctspp != NULL);
ptspp = (tsproc_t *)t->t_cldata;
/*
* Initialize child's tsproc structure.
*/
thread_lock(t);
ctspp->ts_timeleft = ts_dptbl[ptspp->ts_cpupri].ts_quantum;
ctspp->ts_cpupri = ptspp->ts_cpupri;
ctspp->ts_boost = ptspp->ts_boost;
ctspp->ts_uprilim = ptspp->ts_uprilim;
ctspp->ts_upri = ptspp->ts_upri;
TS_NEWUMDPRI(ctspp);
ctspp->ts_nice = ptspp->ts_nice;
ctspp->ts_dispwait = 0;
ctspp->ts_flags = ptspp->ts_flags & ~(TSKPRI | TSBACKQ);
ctspp->ts_tp = ct;
thread_unlock(t);
/*
* Link new structure into tsproc list.
*/
ct->t_cldata = (void *)ctspp;
TS_LIST_INSERT(ctspp);
return (0);
}
/*
* Child is placed at back of dispatcher queue and parent gives
* up processor so that the child runs first after the fork.
* This allows the child immediately execing to break the multiple
* use of copy on write pages with no disk home. The parent will
* get to steal them back rather than uselessly copying them.
*/
static void
ts_forkret(kthread_t *t, kthread_t *ct)
{
proc_t *pp = ttoproc(t);
proc_t *cp = ttoproc(ct);
tsproc_t *tspp;
ASSERT(t == curthread);
ASSERT(MUTEX_HELD(&pidlock));
/*
* Grab the child's p_lock before dropping pidlock to ensure
* the process does not disappear before we set it running.
*/
mutex_enter(&cp->p_lock);
mutex_exit(&pidlock);
continuelwps(cp);
mutex_exit(&cp->p_lock);
mutex_enter(&pp->p_lock);
continuelwps(pp);
mutex_exit(&pp->p_lock);
thread_lock(t);
tspp = (tsproc_t *)(t->t_cldata);
tspp->ts_cpupri = ts_dptbl[tspp->ts_cpupri].ts_tqexp;
TS_NEWUMDPRI(tspp);
tspp->ts_timeleft = ts_dptbl[tspp->ts_cpupri].ts_quantum;
tspp->ts_dispwait = 0;
t->t_pri = ts_dptbl[tspp->ts_umdpri].ts_globpri;
ASSERT(t->t_pri >= 0 && t->t_pri <= ts_maxglobpri);
tspp->ts_flags &= ~TSKPRI;
THREAD_TRANSITION(t);
ts_setrun(t);
thread_unlock(t);
swtch();
}
/*
* Get information about the time-sharing class into the buffer
* pointed to by tsinfop. The maximum configured user priority
* is the only information we supply. ts_getclinfo() is called
* for TS threads, and ia_getclinfo() is called for IA threads.
*/
static int
ts_getclinfo(void *infop)
{
tsinfo_t *tsinfop = (tsinfo_t *)infop;
tsinfop->ts_maxupri = ts_maxupri;
return (0);
}
static int
ia_getclinfo(void *infop)
{
iainfo_t *iainfop = (iainfo_t *)infop;
iainfop->ia_maxupri = ia_maxupri;
return (0);
}
/*
* Return the global scheduling priority ranges for the timesharing
* class in pcpri_t structure.
*/
static int
ts_getclpri(pcpri_t *pcprip)
{
pcprip->pc_clpmax = ts_dptbl[ts_maxumdpri].ts_globpri;
pcprip->pc_clpmin = ts_dptbl[0].ts_globpri;
return (0);
}
static void
ts_nullsys()
{}
/*
* Get the time-sharing parameters of the thread pointed to by
* tsprocp into the buffer pointed to by tsparmsp. ts_parmsget()
* is called for TS threads, and ia_parmsget() is called for IA
* threads.
*/
static void
ts_parmsget(kthread_t *t, void *parmsp)
{
tsproc_t *tspp = (tsproc_t *)t->t_cldata;
tsparms_t *tsparmsp = (tsparms_t *)parmsp;
tsparmsp->ts_uprilim = tspp->ts_uprilim;
tsparmsp->ts_upri = tspp->ts_upri;
}
static void
ia_parmsget(kthread_t *t, void *parmsp)
{
tsproc_t *tspp = (tsproc_t *)t->t_cldata;
iaparms_t *iaparmsp = (iaparms_t *)parmsp;
iaparmsp->ia_uprilim = tspp->ts_uprilim;
iaparmsp->ia_upri = tspp->ts_upri;
if (tspp->ts_flags & TSIASET)
iaparmsp->ia_mode = IA_SET_INTERACTIVE;
else
iaparmsp->ia_mode = IA_INTERACTIVE_OFF;
iaparmsp->ia_nice = tspp->ts_nice;
}
/*
* Check the validity of the time-sharing parameters in the buffer
* pointed to by tsparmsp.
* ts_parmsin() is called for TS threads, and ia_parmsin() is called
* for IA threads.
*/
static int
ts_parmsin(void *parmsp)
{
tsparms_t *tsparmsp = (tsparms_t *)parmsp;
/*
* Check validity of parameters.
*/
if ((tsparmsp->ts_uprilim > ts_maxupri ||
tsparmsp->ts_uprilim < -ts_maxupri) &&
tsparmsp->ts_uprilim != TS_NOCHANGE)
return (EINVAL);
if ((tsparmsp->ts_upri > ts_maxupri ||
tsparmsp->ts_upri < -ts_maxupri) &&
tsparmsp->ts_upri != TS_NOCHANGE)
return (EINVAL);
return (0);
}
static int
ia_parmsin(void *parmsp)
{
iaparms_t *iaparmsp = (iaparms_t *)parmsp;
if ((iaparmsp->ia_uprilim > ia_maxupri ||
iaparmsp->ia_uprilim < -ia_maxupri) &&
iaparmsp->ia_uprilim != IA_NOCHANGE) {
return (EINVAL);
}
if ((iaparmsp->ia_upri > ia_maxupri ||
iaparmsp->ia_upri < -ia_maxupri) &&
iaparmsp->ia_upri != IA_NOCHANGE) {
return (EINVAL);
}
return (0);
}
/*
* Check the validity of the time-sharing parameters in the pc_vaparms_t
* structure vaparmsp and put them in the buffer pointed to by tsparmsp.
* pc_vaparms_t contains (key, value) pairs of parameter.
* ts_vaparmsin() is called for TS threads, and ia_vaparmsin() is called
* for IA threads. ts_vaparmsin() is the variable parameter version of
* ts_parmsin() and ia_vaparmsin() is the variable parameter version of
* ia_parmsin().
*/
static int
ts_vaparmsin(void *parmsp, pc_vaparms_t *vaparmsp)
{
tsparms_t *tsparmsp = (tsparms_t *)parmsp;
int priflag = 0;
int limflag = 0;
uint_t cnt;
pc_vaparm_t *vpp = &vaparmsp->pc_parms[0];
/*
* TS_NOCHANGE (-32768) is outside of the range of values for
* ts_uprilim and ts_upri. If the structure tsparms_t is changed,
* TS_NOCHANGE should be replaced by a flag word (in the same manner
* as in rt.c).
*/
tsparmsp->ts_uprilim = TS_NOCHANGE;
tsparmsp->ts_upri = TS_NOCHANGE;
/*
* Get the varargs parameter and check validity of parameters.
*/
if (vaparmsp->pc_vaparmscnt > PC_VAPARMCNT)
return (EINVAL);
for (cnt = 0; cnt < vaparmsp->pc_vaparmscnt; cnt++, vpp++) {
switch (vpp->pc_key) {
case TS_KY_UPRILIM:
if (limflag++)
return (EINVAL);
tsparmsp->ts_uprilim = (pri_t)vpp->pc_parm;
if (tsparmsp->ts_uprilim > ts_maxupri ||
tsparmsp->ts_uprilim < -ts_maxupri)
return (EINVAL);
break;
case TS_KY_UPRI:
if (priflag++)
return (EINVAL);
tsparmsp->ts_upri = (pri_t)vpp->pc_parm;
if (tsparmsp->ts_upri > ts_maxupri ||
tsparmsp->ts_upri < -ts_maxupri)
return (EINVAL);
break;
default:
return (EINVAL);
}
}
if (vaparmsp->pc_vaparmscnt == 0) {
/*
* Use default parameters.
*/
tsparmsp->ts_upri = tsparmsp->ts_uprilim = 0;
}
return (0);
}
static int
ia_vaparmsin(void *parmsp, pc_vaparms_t *vaparmsp)
{
iaparms_t *iaparmsp = (iaparms_t *)parmsp;
int priflag = 0;
int limflag = 0;
int mflag = 0;
uint_t cnt;
pc_vaparm_t *vpp = &vaparmsp->pc_parms[0];
/*
* IA_NOCHANGE (-32768) is outside of the range of values for
* ia_uprilim, ia_upri and ia_mode. If the structure iaparms_t is
* changed, IA_NOCHANGE should be replaced by a flag word (in the
* same manner as in rt.c).
*/
iaparmsp->ia_uprilim = IA_NOCHANGE;
iaparmsp->ia_upri = IA_NOCHANGE;
iaparmsp->ia_mode = IA_NOCHANGE;
/*
* Get the varargs parameter and check validity of parameters.
*/
if (vaparmsp->pc_vaparmscnt > PC_VAPARMCNT)
return (EINVAL);
for (cnt = 0; cnt < vaparmsp->pc_vaparmscnt; cnt++, vpp++) {
switch (vpp->pc_key) {
case IA_KY_UPRILIM:
if (limflag++)
return (EINVAL);
iaparmsp->ia_uprilim = (pri_t)vpp->pc_parm;
if (iaparmsp->ia_uprilim > ia_maxupri ||
iaparmsp->ia_uprilim < -ia_maxupri)
return (EINVAL);
break;
case IA_KY_UPRI:
if (priflag++)
return (EINVAL);
iaparmsp->ia_upri = (pri_t)vpp->pc_parm;
if (iaparmsp->ia_upri > ia_maxupri ||
iaparmsp->ia_upri < -ia_maxupri)
return (EINVAL);
break;
case IA_KY_MODE:
if (mflag++)
return (EINVAL);
iaparmsp->ia_mode = (int)vpp->pc_parm;
if (iaparmsp->ia_mode != IA_SET_INTERACTIVE &&
iaparmsp->ia_mode != IA_INTERACTIVE_OFF)
return (EINVAL);
break;
default:
return (EINVAL);
}
}
if (vaparmsp->pc_vaparmscnt == 0) {
/*
* Use default parameters.
*/
iaparmsp->ia_upri = iaparmsp->ia_uprilim = 0;
iaparmsp->ia_mode = IA_SET_INTERACTIVE;
}
return (0);
}
/*
* Nothing to do here but return success.
*/
/* ARGSUSED */
static int
ts_parmsout(void *parmsp, pc_vaparms_t *vaparmsp)
{
return (0);
}
/*
* Copy all selected time-sharing class parameters to the user.
* The parameters are specified by a key.
*/
static int
ts_vaparmsout(void *prmsp, pc_vaparms_t *vaparmsp)
{
tsparms_t *tsprmsp = (tsparms_t *)prmsp;
int priflag = 0;
int limflag = 0;
uint_t cnt;
pc_vaparm_t *vpp = &vaparmsp->pc_parms[0];
ASSERT(MUTEX_NOT_HELD(&curproc->p_lock));
if (vaparmsp->pc_vaparmscnt > PC_VAPARMCNT)
return (EINVAL);
for (cnt = 0; cnt < vaparmsp->pc_vaparmscnt; cnt++, vpp++) {
switch (vpp->pc_key) {
case TS_KY_UPRILIM:
if (limflag++)
return (EINVAL);
if (copyout(&tsprmsp->ts_uprilim,
(caddr_t)(uintptr_t)vpp->pc_parm, sizeof (pri_t)))
return (EFAULT);
break;
case TS_KY_UPRI:
if (priflag++)
return (EINVAL);
if (copyout(&tsprmsp->ts_upri,
(caddr_t)(uintptr_t)vpp->pc_parm, sizeof (pri_t)))
return (EFAULT);
break;
default:
return (EINVAL);
}
}
return (0);
}
/*
* Copy all selected interactive class parameters to the user.
* The parameters are specified by a key.
*/
static int
ia_vaparmsout(void *prmsp, pc_vaparms_t *vaparmsp)
{
iaparms_t *iaprmsp = (iaparms_t *)prmsp;
int priflag = 0;
int limflag = 0;
int mflag = 0;
uint_t cnt;
pc_vaparm_t *vpp = &vaparmsp->pc_parms[0];
ASSERT(MUTEX_NOT_HELD(&curproc->p_lock));
if (vaparmsp->pc_vaparmscnt > PC_VAPARMCNT)
return (EINVAL);
for (cnt = 0; cnt < vaparmsp->pc_vaparmscnt; cnt++, vpp++) {
switch (vpp->pc_key) {
case IA_KY_UPRILIM:
if (limflag++)
return (EINVAL);
if (copyout(&iaprmsp->ia_uprilim,
(caddr_t)(uintptr_t)vpp->pc_parm, sizeof (pri_t)))
return (EFAULT);
break;
case IA_KY_UPRI:
if (priflag++)
return (EINVAL);
if (copyout(&iaprmsp->ia_upri,
(caddr_t)(uintptr_t)vpp->pc_parm, sizeof (pri_t)))
return (EFAULT);
break;
case IA_KY_MODE:
if (mflag++)
return (EINVAL);
if (copyout(&iaprmsp->ia_mode,
(caddr_t)(uintptr_t)vpp->pc_parm, sizeof (int)))
return (EFAULT);
break;
default:
return (EINVAL);
}
}
return (0);
}
/*
* Set the scheduling parameters of the thread pointed to by tsprocp
* to those specified in the buffer pointed to by tsparmsp.
* ts_parmsset() is called for TS threads, and ia_parmsset() is
* called for IA threads.
*/
/* ARGSUSED */
static int
ts_parmsset(kthread_t *tx, void *parmsp, id_t reqpcid, cred_t *reqpcredp)
{
char nice;
pri_t reqtsuprilim;
pri_t reqtsupri;
tsparms_t *tsparmsp = (tsparms_t *)parmsp;
tsproc_t *tspp = (tsproc_t *)tx->t_cldata;
ASSERT(MUTEX_HELD(&(ttoproc(tx))->p_lock));
if (tsparmsp->ts_uprilim == TS_NOCHANGE)
reqtsuprilim = tspp->ts_uprilim;
else
reqtsuprilim = tsparmsp->ts_uprilim;
if (tsparmsp->ts_upri == TS_NOCHANGE)
reqtsupri = tspp->ts_upri;
else
reqtsupri = tsparmsp->ts_upri;
/*
* Make sure the user priority doesn't exceed the upri limit.
*/
if (reqtsupri > reqtsuprilim)
reqtsupri = reqtsuprilim;
/*
* Basic permissions enforced by generic kernel code
* for all classes require that a thread attempting
* to change the scheduling parameters of a target
* thread be privileged or have a real or effective
* UID matching that of the target thread. We are not
* called unless these basic permission checks have
* already passed. The time-sharing class requires in
* addition that the calling thread be privileged if it
* is attempting to raise the upri limit above its current
* value This may have been checked previously but if our
* caller passed us a non-NULL credential pointer we assume
* it hasn't and we check it here.
*/
if (reqpcredp != NULL &&
reqtsuprilim > tspp->ts_uprilim &&
secpolicy_setpriority(reqpcredp) != 0)
return (EPERM);
/*
* Set ts_nice to the nice value corresponding to the user
* priority we are setting. Note that setting the nice field
* of the parameter struct won't affect upri or nice.
*/
nice = NZERO - (reqtsupri * NZERO) / ts_maxupri;
if (nice >= 2 * NZERO)
nice = 2 * NZERO - 1;
thread_lock(tx);
tspp->ts_uprilim = reqtsuprilim;
tspp->ts_upri = reqtsupri;
TS_NEWUMDPRI(tspp);
tspp->ts_nice = nice;
if ((tspp->ts_flags & TSKPRI) != 0) {
thread_unlock(tx);
return (0);
}
tspp->ts_dispwait = 0;
ts_change_priority(tx, tspp);
thread_unlock(tx);
return (0);
}
static int
ia_parmsset(kthread_t *tx, void *parmsp, id_t reqpcid, cred_t *reqpcredp)
{
tsproc_t *tspp = (tsproc_t *)tx->t_cldata;
iaparms_t *iaparmsp = (iaparms_t *)parmsp;
proc_t *p;
pid_t pid, pgid, sid;
pid_t on, off;
struct stdata *stp;
int sess_held;
/*
* Handle user priority changes
*/
if (iaparmsp->ia_mode == IA_NOCHANGE)
return (ts_parmsset(tx, parmsp, reqpcid, reqpcredp));
/*
* Check permissions for changing modes.
*/
if (reqpcredp != NULL && !groupmember(IA_gid, reqpcredp) &&
secpolicy_setpriority(reqpcredp) != 0) {
/*
* Silently fail in case this is just a priocntl
* call with upri and uprilim set to IA_NOCHANGE.
*/
return (0);
}
ASSERT(MUTEX_HELD(&pidlock));
if ((p = ttoproc(tx)) == NULL) {
return (0);
}
ASSERT(MUTEX_HELD(&p->p_lock));
if (p->p_stat == SIDL) {
return (0);
}
pid = p->p_pid;
sid = p->p_sessp->s_sid;
pgid = p->p_pgrp;
if (iaparmsp->ia_mode == IA_SET_INTERACTIVE) {
/*
* session leaders must be turned on now so all processes
* in the group controlling the tty will be turned on or off.
* if the ia_mode is off for the session leader,
* ia_set_process_group will return without setting the
* processes in the group controlling the tty on.
*/
thread_lock(tx);
tspp->ts_flags |= TSIASET;
thread_unlock(tx);
}
TTY_HOLD(p->p_sessp);
sess_held = 1;
if ((pid == sid) && (p->p_sessp->s_vp != NULL) &&
((stp = p->p_sessp->s_vp->v_stream) != NULL)) {
if ((stp->sd_pgidp != NULL) && (stp->sd_sidp != NULL)) {
pgid = stp->sd_pgidp->pid_id;
sess_held = 0;
TTY_RELE(p->p_sessp);
if (iaparmsp->ia_mode ==
IA_SET_INTERACTIVE) {
off = 0;
on = pgid;
} else {
off = pgid;
on = 0;
}
TRACE_3(TR_FAC_IA, TR_ACTIVE_CHAIN,
"active chain:pid %d gid %d %p",
pid, pgid, p);
ia_set_process_group(sid, off, on);
}
}
if (sess_held)
TTY_RELE(p->p_sessp);
thread_lock(tx);
if (iaparmsp->ia_mode == IA_SET_INTERACTIVE) {
tspp->ts_flags |= TSIASET;
tspp->ts_boost = ia_boost;
} else {
tspp->ts_flags &= ~TSIASET;
tspp->ts_boost = -ia_boost;
}
thread_unlock(tx);
return (ts_parmsset(tx, parmsp, reqpcid, reqpcredp));
}
/*
* Return the global scheduling priority that would be assigned
* to a thread entering the time-sharing class with the ts_upri.
*/
static pri_t
ts_globpri(kthread_t *t)
{
tsproc_t *tspp;
pri_t tspri;
ASSERT(MUTEX_HELD(&ttoproc(t)->p_lock));
tspp = (tsproc_t *)t->t_cldata;
tspri = tsmedumdpri + tspp->ts_upri;
if (tspri > ts_maxumdpri)
tspri = ts_maxumdpri;
else if (tspri < 0)
tspri = 0;
return (ts_dptbl[tspri].ts_globpri);
}
/*
* Arrange for thread to be placed in appropriate location
* on dispatcher queue.
*
* This is called with the current thread in TS_ONPROC and locked.
*/
static void
ts_preempt(kthread_t *t)
{
tsproc_t *tspp = (tsproc_t *)(t->t_cldata);
klwp_t *lwp;
#ifdef KSLICE
extern int kslice;
#endif
pri_t oldpri = t->t_pri;
ASSERT(t == curthread);
ASSERT(THREAD_LOCK_HELD(curthread));
/*
* If preempted in the kernel, make sure the thread has
* a kernel priority if needed.
*/
lwp = curthread->t_lwp;
if (!(tspp->ts_flags & TSKPRI) && lwp != NULL && t->t_kpri_req) {
tspp->ts_flags |= TSKPRI;
THREAD_CHANGE_PRI(t, ts_kmdpris[0]);
ASSERT(t->t_pri >= 0 && t->t_pri <= ts_maxglobpri);
t->t_trapret = 1; /* so ts_trapret will run */
aston(t);
}
/*
* If preempted in user-land mark the thread
* as swappable because I know it isn't holding any locks.
*/
ASSERT(t->t_schedflag & TS_DONT_SWAP);
if (lwp != NULL && lwp->lwp_state == LWP_USER)
t->t_schedflag &= ~TS_DONT_SWAP;
/*
* Check to see if we're doing "preemption control" here. If
* we are, and if the user has requested that this thread not
* be preempted, and if preemptions haven't been put off for
* too long, let the preemption happen here but try to make
* sure the thread is rescheduled as soon as possible. We do
* this by putting it on the front of the highest priority run
* queue in the TS class. If the preemption has been put off
* for too long, clear the "nopreempt" bit and let the thread
* be preempted.
*/
if (t->t_schedctl && schedctl_get_nopreempt(t)) {
if (tspp->ts_timeleft > -SC_MAX_TICKS) {
DTRACE_SCHED1(schedctl__nopreempt, kthread_t *, t);
if (!(tspp->ts_flags & TSKPRI)) {
THREAD_CHANGE_PRI(t, ts_maxumdpri);
t->t_schedflag |= TS_DONT_SWAP;
schedctl_set_yield(t, 1);
}
setfrontdq(t);
goto done;
} else {
schedctl_set_nopreempt(t, 0);
DTRACE_SCHED1(schedctl__preempt, kthread_t *, t);
TNF_PROBE_2(schedctl_preempt, "schedctl TS ts_preempt",
/* CSTYLED */, tnf_pid, pid, ttoproc(t)->p_pid,
tnf_lwpid, lwpid, t->t_tid);
/*
* Fall through and be preempted below.
*/
}
}
if ((tspp->ts_flags & (TSBACKQ|TSKPRI)) == TSBACKQ) {
tspp->ts_timeleft = ts_dptbl[tspp->ts_cpupri].ts_quantum;
tspp->ts_dispwait = 0;
tspp->ts_flags &= ~TSBACKQ;
setbackdq(t);
} else if ((tspp->ts_flags & (TSBACKQ|TSKPRI)) == (TSBACKQ|TSKPRI)) {
tspp->ts_flags &= ~TSBACKQ;
setbackdq(t);
} else {
#ifdef KSLICE
if (kslice)
setbackdq(t);
else
#endif
setfrontdq(t);
}
done:
TRACE_2(TR_FAC_DISP, TR_PREEMPT,
"preempt:tid %p old pri %d", t, oldpri);
}
static void
ts_setrun(kthread_t *t)
{
tsproc_t *tspp = (tsproc_t *)(t->t_cldata);
ASSERT(THREAD_LOCK_HELD(t)); /* t should be in transition */
if (tspp->ts_dispwait > ts_dptbl[tspp->ts_umdpri].ts_maxwait) {
tspp->ts_cpupri = ts_dptbl[tspp->ts_cpupri].ts_slpret;
TS_NEWUMDPRI(tspp);
tspp->ts_timeleft = ts_dptbl[tspp->ts_cpupri].ts_quantum;
tspp->ts_dispwait = 0;
if ((tspp->ts_flags & TSKPRI) == 0) {
THREAD_CHANGE_PRI(t,
ts_dptbl[tspp->ts_umdpri].ts_globpri);
ASSERT(t->t_pri >= 0 && t->t_pri <= ts_maxglobpri);
}
}
tspp->ts_flags &= ~TSBACKQ;
if (tspp->ts_flags & TSIA) {
if (tspp->ts_flags & TSIASET)
setfrontdq(t);
else
setbackdq(t);
} else {
if (t->t_disp_time != lbolt)
setbackdq(t);
else
setfrontdq(t);
}
}
/*
* Prepare thread for sleep. We reset the thread priority so it will
* run at the kernel priority level when it wakes up.
*/
static void
ts_sleep(kthread_t *t)
{
tsproc_t *tspp = (tsproc_t *)(t->t_cldata);
int flags;
pri_t old_pri = t->t_pri;
ASSERT(t == curthread);
ASSERT(THREAD_LOCK_HELD(t));
flags = tspp->ts_flags;
if (t->t_kpri_req) {
tspp->ts_flags = flags | TSKPRI;
THREAD_CHANGE_PRI(t, ts_kmdpris[0]);
ASSERT(t->t_pri >= 0 && t->t_pri <= ts_maxglobpri);
t->t_trapret = 1; /* so ts_trapret will run */
aston(t);
} else if (tspp->ts_dispwait > ts_dptbl[tspp->ts_umdpri].ts_maxwait) {
/*
* If thread has blocked in the kernel (as opposed to
* being merely preempted), recompute the user mode priority.
*/
tspp->ts_cpupri = ts_dptbl[tspp->ts_cpupri].ts_slpret;
TS_NEWUMDPRI(tspp);
tspp->ts_timeleft = ts_dptbl[tspp->ts_cpupri].ts_quantum;
tspp->ts_dispwait = 0;
THREAD_CHANGE_PRI(curthread,
ts_dptbl[tspp->ts_umdpri].ts_globpri);
ASSERT(curthread->t_pri >= 0 &&
curthread->t_pri <= ts_maxglobpri);
tspp->ts_flags = flags & ~TSKPRI;
if (DISP_MUST_SURRENDER(curthread))
cpu_surrender(curthread);
} else if (flags & TSKPRI) {
THREAD_CHANGE_PRI(curthread,
ts_dptbl[tspp->ts_umdpri].ts_globpri);
ASSERT(curthread->t_pri >= 0 &&
curthread->t_pri <= ts_maxglobpri);
tspp->ts_flags = flags & ~TSKPRI;
if (DISP_MUST_SURRENDER(curthread))
cpu_surrender(curthread);
}
t->t_stime = lbolt; /* time stamp for the swapper */
TRACE_2(TR_FAC_DISP, TR_SLEEP,
"sleep:tid %p old pri %d", t, old_pri);
}
/*
* Return Values:
*
* -1 if the thread is loaded or is not eligible to be swapped in.
*
* effective priority of the specified thread based on swapout time
* and size of process (epri >= 0 , epri <= SHRT_MAX).
*/
/* ARGSUSED */
static pri_t
ts_swapin(kthread_t *t, int flags)
{
tsproc_t *tspp = (tsproc_t *)(t->t_cldata);
long epri = -1;
proc_t *pp = ttoproc(t);
ASSERT(THREAD_LOCK_HELD(t));
/*
* We know that pri_t is a short.
* Be sure not to overrun its range.
*/
if (t->t_state == TS_RUN && (t->t_schedflag & TS_LOAD) == 0) {
time_t swapout_time;
swapout_time = (lbolt - t->t_stime) / hz;
if (INHERITED(t) || (tspp->ts_flags & (TSKPRI | TSIASET)))
epri = (long)DISP_PRIO(t) + swapout_time;
else {
/*
* Threads which have been out for a long time,
* have high user mode priority and are associated
* with a small address space are more deserving
*/
epri = ts_dptbl[tspp->ts_umdpri].ts_globpri;
ASSERT(epri >= 0 && epri <= ts_maxumdpri);
epri += swapout_time - pp->p_swrss / nz(maxpgio)/2;
}
/*
* Scale epri so SHRT_MAX/2 represents zero priority.
*/
epri += SHRT_MAX/2;
if (epri < 0)
epri = 0;
else if (epri > SHRT_MAX)
epri = SHRT_MAX;
}
return ((pri_t)epri);
}
/*
* Return Values
* -1 if the thread isn't loaded or is not eligible to be swapped out.
*
* effective priority of the specified thread based on if the swapper
* is in softswap or hardswap mode.
*
* Softswap: Return a low effective priority for threads
* sleeping for more than maxslp secs.
*
* Hardswap: Return an effective priority such that threads
* which have been in memory for a while and are
* associated with a small address space are swapped
* in before others.
*
* (epri >= 0 , epri <= SHRT_MAX).
*/
time_t ts_minrun = 2; /* XXX - t_pri becomes 59 within 2 secs */
time_t ts_minslp = 2; /* min time on sleep queue for hardswap */
static pri_t
ts_swapout(kthread_t *t, int flags)
{
tsproc_t *tspp = (tsproc_t *)(t->t_cldata);
long epri = -1;
proc_t *pp = ttoproc(t);
time_t swapin_time;
ASSERT(THREAD_LOCK_HELD(t));
if (INHERITED(t) || (tspp->ts_flags & (TSKPRI | TSIASET)) ||
(t->t_proc_flag & TP_LWPEXIT) ||
(t->t_state & (TS_ZOMB | TS_FREE | TS_STOPPED | TS_ONPROC)) ||
!(t->t_schedflag & TS_LOAD) || !SWAP_OK(t))
return (-1);
ASSERT(t->t_state & (TS_SLEEP | TS_RUN));
/*
* We know that pri_t is a short.
* Be sure not to overrun its range.
*/
swapin_time = (lbolt - t->t_stime) / hz;
if (flags == SOFTSWAP) {
if (t->t_state == TS_SLEEP && swapin_time > maxslp) {
epri = 0;
} else {
return ((pri_t)epri);
}
} else {
pri_t pri;
if ((t->t_state == TS_SLEEP && swapin_time > ts_minslp) ||
(t->t_state == TS_RUN && swapin_time > ts_minrun)) {
pri = ts_dptbl[tspp->ts_umdpri].ts_globpri;
ASSERT(pri >= 0 && pri <= ts_maxumdpri);
epri = swapin_time -
(rm_asrss(pp->p_as) / nz(maxpgio)/2) - (long)pri;
} else {
return ((pri_t)epri);
}
}
/*
* Scale epri so SHRT_MAX/2 represents zero priority.
*/
epri += SHRT_MAX/2;
if (epri < 0)
epri = 0;
else if (epri > SHRT_MAX)
epri = SHRT_MAX;
return ((pri_t)epri);
}
/*
* Check for time slice expiration. If time slice has expired
* move thread to priority specified in tsdptbl for time slice expiration
* and set runrun to cause preemption.
*/
static void
ts_tick(kthread_t *t)
{
tsproc_t *tspp = (tsproc_t *)(t->t_cldata);
klwp_t *lwp;
pri_t oldpri = t->t_pri;
ASSERT(MUTEX_HELD(&(ttoproc(t))->p_lock));
thread_lock(t);
if ((tspp->ts_flags & TSKPRI) == 0) {
if (--tspp->ts_timeleft <= 0) {
pri_t new_pri;
/*
* If we're doing preemption control and trying to
* avoid preempting this thread, just note that
* the thread should yield soon and let it keep
* running (unless it's been a while).
*/
if (t->t_schedctl && schedctl_get_nopreempt(t)) {
if (tspp->ts_timeleft > -SC_MAX_TICKS) {
DTRACE_SCHED1(schedctl__nopreempt,
kthread_t *, t);
schedctl_set_yield(t, 1);
thread_unlock_nopreempt(t);
return;
}
TNF_PROBE_2(schedctl_failsafe,
"schedctl TS ts_tick", /* CSTYLED */,
tnf_pid, pid, ttoproc(t)->p_pid,
tnf_lwpid, lwpid, t->t_tid);
}
tspp->ts_cpupri = ts_dptbl[tspp->ts_cpupri].ts_tqexp;
TS_NEWUMDPRI(tspp);
tspp->ts_dispwait = 0;
new_pri = ts_dptbl[tspp->ts_umdpri].ts_globpri;
ASSERT(new_pri >= 0 && new_pri <= ts_maxglobpri);
/*
* 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)) {
if ((t->t_schedflag & TS_LOAD) &&
(lwp = t->t_lwp) &&
lwp->lwp_state == LWP_USER)
t->t_schedflag &= ~TS_DONT_SWAP;
tspp->ts_timeleft =
ts_dptbl[tspp->ts_cpupri].ts_quantum;
} else {
tspp->ts_flags |= TSBACKQ;
cpu_surrender(t);
}
TRACE_2(TR_FAC_DISP, TR_TICK,
"tick:tid %p old pri %d", t, oldpri);
} else if (t->t_pri < t->t_disp_queue->disp_maxrunpri) {
tspp->ts_flags |= TSBACKQ;
cpu_surrender(t);
}
}
thread_unlock_nopreempt(t); /* clock thread can't be preempted */
}
/*
* If thread is currently at a kernel mode priority (has slept)
* we assign it the appropriate user mode priority and time quantum
* here. If we are lowering the thread's priority below that of
* other runnable threads we will normally set runrun via cpu_surrender() to
* cause preemption.
*/
static void
ts_trapret(kthread_t *t)
{
tsproc_t *tspp = (tsproc_t *)t->t_cldata;
cpu_t *cp = CPU;
pri_t old_pri = curthread->t_pri;
ASSERT(THREAD_LOCK_HELD(t));
ASSERT(t == curthread);
ASSERT(cp->cpu_dispthread == t);
ASSERT(t->t_state == TS_ONPROC);
t->t_kpri_req = 0;
if (tspp->ts_dispwait > ts_dptbl[tspp->ts_umdpri].ts_maxwait) {
tspp->ts_cpupri = ts_dptbl[tspp->ts_cpupri].ts_slpret;
TS_NEWUMDPRI(tspp);
tspp->ts_timeleft = ts_dptbl[tspp->ts_cpupri].ts_quantum;
tspp->ts_dispwait = 0;
/*
* If thread has blocked in the kernel (as opposed to
* being merely preempted), recompute the user mode priority.
*/
THREAD_CHANGE_PRI(t, ts_dptbl[tspp->ts_umdpri].ts_globpri);
cp->cpu_dispatch_pri = DISP_PRIO(t);
ASSERT(t->t_pri >= 0 && t->t_pri <= ts_maxglobpri);
tspp->ts_flags &= ~TSKPRI;
if (DISP_MUST_SURRENDER(t))
cpu_surrender(t);
} else if (tspp->ts_flags & TSKPRI) {
/*
* If thread has blocked in the kernel (as opposed to
* being merely preempted), recompute the user mode priority.
*/
THREAD_CHANGE_PRI(t, ts_dptbl[tspp->ts_umdpri].ts_globpri);
cp->cpu_dispatch_pri = DISP_PRIO(t);
ASSERT(t->t_pri >= 0 && t->t_pri <= ts_maxglobpri);
tspp->ts_flags &= ~TSKPRI;
if (DISP_MUST_SURRENDER(t))
cpu_surrender(t);
}
/*
* Swapout lwp if the swapper is waiting for this thread to
* reach a safe point.
*/
if ((t->t_schedflag & TS_SWAPENQ) && !(tspp->ts_flags & TSIASET)) {
thread_unlock(t);
swapout_lwp(ttolwp(t));
thread_lock(t);
}
TRACE_2(TR_FAC_DISP, TR_TRAPRET,
"trapret:tid %p old pri %d", t, old_pri);
}
/*
* Update the ts_dispwait values of all time sharing threads that
* are currently runnable at a user mode priority and bump the priority
* if ts_dispwait exceeds ts_maxwait. Called once per second via
* timeout which we reset here.
*
* There are several lists of time sharing threads broken up by a hash on
* the thread pointer. Each list has its own lock. This avoids blocking
* all ts_enterclass, ts_fork, and ts_exitclass operations while ts_update
* runs. ts_update traverses each list in turn.
*
* If multiple threads have their priorities updated to the same value,
* the system implicitly favors the one that is updated first (since it
* winds up first on the run queue). To avoid this unfairness, the
* traversal of threads starts at the list indicated by a marker. When
* threads in more than one list have their priorities updated, the marker
* is moved. This changes the order the threads will be placed on the run
* queue the next time ts_update is called and preserves fairness over the
* long run. The marker doesn't need to be protected by a lock since it's
* only accessed by ts_update, which is inherently single-threaded (only
* one instance can be running at a time).
*/
static void
ts_update(void *arg)
{
int i;
int new_marker = -1;
static int ts_update_marker;
/*
* Start with the ts_update_marker list, then do the rest.
*/
i = ts_update_marker;
do {
/*
* If this is the first list after the current marker to
* have threads with priorities updated, advance the marker
* to this list for the next time ts_update runs.
*/
if (ts_update_list(i) && new_marker == -1 &&
i != ts_update_marker) {
new_marker = i;
}
} while ((i = TS_LIST_NEXT(i)) != ts_update_marker);
/* advance marker for next ts_update call */
if (new_marker != -1)
ts_update_marker = new_marker;
(void) timeout(ts_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 TS or IA class, isn't runnable,
* hasn't waited long enough, has the preemption control no-preempt bit
* set, etc.)
*/
static int
ts_update_list(int i)
{
tsproc_t *tspp;
kthread_t *tx;
int updated = 0;
mutex_enter(&ts_list_lock[i]);
for (tspp = ts_plisthead[i].ts_next; tspp != &ts_plisthead[i];
tspp = tspp->ts_next) {
tx = tspp->ts_tp;
/*
* Lock the thread and verify state.
*/
thread_lock(tx);
/*
* Skip the thread if it is no longer in the TS (or IA) class.
*/
if (tx->t_clfuncs != &ts_classfuncs.thread &&
tx->t_clfuncs != &ia_classfuncs.thread)
goto next;
tspp->ts_dispwait++;
if ((tspp->ts_flags & TSKPRI) != 0)
goto next;
if (tspp->ts_dispwait <= ts_dptbl[tspp->ts_umdpri].ts_maxwait)
goto next;
if (tx->t_schedctl && schedctl_get_nopreempt(tx))
goto next;
if (tx->t_state != TS_RUN && (tx->t_state != TS_SLEEP ||
!ts_sleep_promote)) {
/* make next syscall/trap do CL_TRAPRET */
tx->t_trapret = 1;
aston(tx);
goto next;
}
tspp->ts_cpupri = ts_dptbl[tspp->ts_cpupri].ts_lwait;
TS_NEWUMDPRI(tspp);
tspp->ts_dispwait = 0;
updated = 1;
/*
* Only dequeue it if needs to move; otherwise it should
* just round-robin here.
*/
if (tx->t_pri != ts_dptbl[tspp->ts_umdpri].ts_globpri) {
pri_t oldpri = tx->t_pri;
ts_change_priority(tx, tspp);
TRACE_2(TR_FAC_DISP, TR_UPDATE,
"update:tid %p old pri %d", tx, oldpri);
}
next:
thread_unlock(tx);
}
mutex_exit(&ts_list_lock[i]);
return (updated);
}
/*
* Processes waking up go to the back of their queue. We don't
* need to assign a time quantum here because thread is still
* at a kernel mode priority and the time slicing is not done
* for threads running in the kernel after sleeping. The proper
* time quantum will be assigned by ts_trapret before the thread
* returns to user mode.
*/
static void
ts_wakeup(kthread_t *t)
{
tsproc_t *tspp = (tsproc_t *)(t->t_cldata);
ASSERT(THREAD_LOCK_HELD(t));
t->t_stime = lbolt; /* time stamp for the swapper */
if (tspp->ts_flags & TSKPRI) {
tspp->ts_flags &= ~TSBACKQ;
if (tspp->ts_flags & TSIASET)
setfrontdq(t);
else
setbackdq(t);
} else if (t->t_kpri_req) {
/*
* Give thread a priority boost if we were asked.
*/
tspp->ts_flags |= TSKPRI;
THREAD_CHANGE_PRI(t, ts_kmdpris[0]);
setbackdq(t);
t->t_trapret = 1; /* so that ts_trapret will run */
aston(t);
} else {
if (tspp->ts_dispwait > ts_dptbl[tspp->ts_umdpri].ts_maxwait) {
tspp->ts_cpupri = ts_dptbl[tspp->ts_cpupri].ts_slpret;
TS_NEWUMDPRI(tspp);
tspp->ts_timeleft =
ts_dptbl[tspp->ts_cpupri].ts_quantum;
tspp->ts_dispwait = 0;
THREAD_CHANGE_PRI(t,
ts_dptbl[tspp->ts_umdpri].ts_globpri);
ASSERT(t->t_pri >= 0 && t->t_pri <= ts_maxglobpri);
}
tspp->ts_flags &= ~TSBACKQ;
if (tspp->ts_flags & TSIA) {
if (tspp->ts_flags & TSIASET)
setfrontdq(t);
else
setbackdq(t);
} else {
if (t->t_disp_time != lbolt)
setbackdq(t);
else
setfrontdq(t);
}
}
}
/*
* When a thread yields, put it on the back of the run queue.
*/
static void
ts_yield(kthread_t *t)
{
tsproc_t *tspp = (tsproc_t *)(t->t_cldata);
ASSERT(t == curthread);
ASSERT(THREAD_LOCK_HELD(t));
/*
* Clear the preemption control "yield" bit since the user is
* doing a yield.
*/
if (t->t_schedctl)
schedctl_set_yield(t, 0);
if (tspp->ts_timeleft <= 0) {
/*
* Time slice was artificially extended to avoid
* preemption, so pretend we're preempting it now.
*/
DTRACE_SCHED1(schedctl__yield, int, -tspp->ts_timeleft);
tspp->ts_cpupri = ts_dptbl[tspp->ts_cpupri].ts_tqexp;
TS_NEWUMDPRI(tspp);
tspp->ts_timeleft = ts_dptbl[tspp->ts_cpupri].ts_quantum;
tspp->ts_dispwait = 0;
THREAD_CHANGE_PRI(t, ts_dptbl[tspp->ts_umdpri].ts_globpri);
ASSERT(t->t_pri >= 0 && t->t_pri <= ts_maxglobpri);
}
tspp->ts_flags &= ~TSBACKQ;
setbackdq(t);
}
/*
* Increment the nice value of the specified thread by incr and
* return the new value in *retvalp.
*/
static int
ts_donice(kthread_t *t, cred_t *cr, int incr, int *retvalp)
{
int newnice;
tsproc_t *tspp = (tsproc_t *)(t->t_cldata);
tsparms_t tsparms;
ASSERT(MUTEX_HELD(&(ttoproc(t))->p_lock));
/* If there's no change to priority, just return current setting */
if (incr == 0) {
if (retvalp) {
*retvalp = tspp->ts_nice - NZERO;
}
return (0);
}
if ((incr < 0 || incr > 2 * NZERO) &&
secpolicy_setpriority(cr) != 0)
return (EPERM);
/*
* Specifying a nice increment greater than the upper limit of
* 2 * NZERO - 1 will result in the thread's nice value being
* set to the upper limit. We check for this before computing
* the new value because otherwise we could get overflow
* if a privileged process specified some ridiculous increment.
*/
if (incr > 2 * NZERO - 1)
incr = 2 * NZERO - 1;
newnice = tspp->ts_nice + incr;
if (newnice >= 2 * NZERO)
newnice = 2 * NZERO - 1;
else if (newnice < 0)
newnice = 0;
tsparms.ts_uprilim = tsparms.ts_upri =
-((newnice - NZERO) * ts_maxupri) / NZERO;
/*
* Reset the uprilim and upri values of the thread.
* Call ts_parmsset even if thread is interactive since we're
* not changing mode.
*/
(void) ts_parmsset(t, (void *)&tsparms, (id_t)0, (cred_t *)NULL);
/*
* Although ts_parmsset already reset ts_nice it may
* not have been set to precisely the value calculated above
* because ts_parmsset determines the nice value from the
* user priority and we may have truncated during the integer
* conversion from nice value to user priority and back.
* We reset ts_nice to the value we calculated above.
*/
tspp->ts_nice = (char)newnice;
if (retvalp)
*retvalp = newnice - NZERO;
return (0);
}
/*
* ia_set_process_group marks foreground processes as interactive
* and background processes as non-interactive iff the session
* leader is interactive. This routine is called from two places:
* strioctl:SPGRP when a new process group gets
* control of the tty.
* ia_parmsset-when the process in question is a session leader.
* ia_set_process_group assumes that pidlock is held by the caller,
* either strioctl or priocntlsys. If the caller is priocntlsys
* (via ia_parmsset) then the p_lock of the session leader is held
* and the code needs to be careful about acquiring other p_locks.
*/
static void
ia_set_process_group(pid_t sid, pid_t bg_pgid, pid_t fg_pgid)
{
proc_t *leader, *fg, *bg;
tsproc_t *tspp;
kthread_t *tx;
int plocked = 0;
ASSERT(MUTEX_HELD(&pidlock));
/*
* see if the session leader is interactive AND
* if it is currently "on" AND controlling a tty
* iff it is then make the processes in the foreground
* group interactive and the processes in the background
* group non-interactive.
*/
if ((leader = (proc_t *)prfind(sid)) == NULL) {
return;
}
if (leader->p_stat == SIDL) {
return;
}
if ((tx = proctot(leader)) == NULL) {
return;
}
/*
* XXX do all the threads in the leader
*/
if (tx->t_cid != ia_cid) {
return;
}
tspp = tx->t_cldata;
/*
* session leaders that are not interactive need not have
* any processing done for them. They are typically shells
* that do not have focus and are changing the process group
* attatched to the tty, e.g. a process that is exiting
*/
TTY_HOLD(leader->p_sessp);
if (!(tspp->ts_flags & TSIASET) ||
(leader->p_sessp->s_vp == NULL) ||
(leader->p_sessp->s_vp->v_stream == NULL)) {
TTY_RELE(leader->p_sessp);
return;
}
TTY_RELE(leader->p_sessp);
/*
* If we're already holding the leader's p_lock, we should use
* mutex_tryenter instead of mutex_enter to avoid deadlocks from
* lock ordering violations.
*/
if (mutex_owned(&leader->p_lock))
plocked = 1;
if (fg_pgid == 0)
goto skip;
/*
* now look for all processes in the foreground group and
* make them interactive
*/
for (fg = (proc_t *)pgfind(fg_pgid); fg != NULL; fg = fg->p_pglink) {
/*
* if the process is SIDL it's begin forked, ignore it
*/
if (fg->p_stat == SIDL) {
continue;
}
/*
* sesssion leaders must be turned on/off explicitly
* not implicitly as happens to other members of
* the process group.
*/
if (fg->p_pid == fg->p_sessp->s_sid) {
continue;
}
TRACE_1(TR_FAC_IA, TR_GROUP_ON,
"group on:proc %p", fg);
if (plocked) {
if (mutex_tryenter(&fg->p_lock) == 0)
continue;
} else {
mutex_enter(&fg->p_lock);
}
if ((tx = proctot(fg)) == NULL) {
mutex_exit(&fg->p_lock);
continue;
}
do {
thread_lock(tx);
/*
* if this thread is not interactive continue
*/
if (tx->t_cid != ia_cid) {
thread_unlock(tx);
continue;
}
tspp = tx->t_cldata;
tspp->ts_flags |= TSIASET;
tspp->ts_boost = ia_boost;
TS_NEWUMDPRI(tspp);
if ((tspp->ts_flags & TSKPRI) != 0) {
thread_unlock(tx);
continue;
}
tspp->ts_dispwait = 0;
ts_change_priority(tx, tspp);
thread_unlock(tx);
} while ((tx = tx->t_forw) != fg->p_tlist);
mutex_exit(&fg->p_lock);
}
skip:
if (bg_pgid == 0)
return;
for (bg = (proc_t *)pgfind(bg_pgid); bg != NULL; bg = bg->p_pglink) {
if (bg->p_stat == SIDL) {
continue;
}
/*
* sesssion leaders must be turned off explicitly
* not implicitly as happens to other members of
* the process group.
*/
if (bg->p_pid == bg->p_sessp->s_sid) {
continue;
}
TRACE_1(TR_FAC_IA, TR_GROUP_OFF,
"group off:proc %p", bg);
if (plocked) {
if (mutex_tryenter(&bg->p_lock) == 0)
continue;
} else {
mutex_enter(&bg->p_lock);
}
if ((tx = proctot(bg)) == NULL) {
mutex_exit(&bg->p_lock);
continue;
}
do {
thread_lock(tx);
/*
* if this thread is not interactive continue
*/
if (tx->t_cid != ia_cid) {
thread_unlock(tx);
continue;
}
tspp = tx->t_cldata;
tspp->ts_flags &= ~TSIASET;
tspp->ts_boost = -ia_boost;
TS_NEWUMDPRI(tspp);
if ((tspp->ts_flags & TSKPRI) != 0) {
thread_unlock(tx);
continue;
}
tspp->ts_dispwait = 0;
ts_change_priority(tx, tspp);
thread_unlock(tx);
} while ((tx = tx->t_forw) != bg->p_tlist);
mutex_exit(&bg->p_lock);
}
}
static void
ts_change_priority(kthread_t *t, tsproc_t *tspp)
{
pri_t new_pri;
ASSERT(THREAD_LOCK_HELD(t));
new_pri = ts_dptbl[tspp->ts_umdpri].ts_globpri;
ASSERT(new_pri >= 0 && new_pri <= ts_maxglobpri);
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)) {
tspp->ts_flags |= TSBACKQ;
cpu_surrender(t);
} else {
tspp->ts_timeleft =
ts_dptbl[tspp->ts_cpupri].ts_quantum;
}
} else {
int frontq;
frontq = (tspp->ts_flags & TSIASET) != 0;
/*
* 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, frontq)) {
/*
* The thread was on a run queue. Reset
* its CPU timeleft from the quantum
* associated with the new priority.
*/
tspp->ts_timeleft =
ts_dptbl[tspp->ts_cpupri].ts_quantum;
} else {
tspp->ts_flags |= TSBACKQ;
}
}
}
static int
ts_alloc(void **p, int flag)
{
void *bufp;
bufp = kmem_alloc(sizeof (tsproc_t), flag);
if (bufp == NULL) {
return (ENOMEM);
} else {
*p = bufp;
return (0);
}
}
static void
ts_free(void *bufp)
{
if (bufp)
kmem_free(bufp, sizeof (tsproc_t));
}