BSD 4_4_Lite1 release

[unix-history] / usr / src / sys / kern / kern_synch.c

diff --git a/usr/src/sys/kern/kern_synch.c b/usr/src/sys/kern/kern_synch.c
index d33c40b..1c2a578 100644 (file)
--- a/usr/src/sys/kern/kern_synch.c
+++ b/usr/src/sys/kern/kern_synch.c
@@ -1,221 +1,474 @@
-/*
- * Copyright (c) 1982, 1986 Regents of the University of California.
- * All rights reserved.  The Berkeley software License Agreement
- * specifies the terms and conditions for redistribution.
+/*-
+ * Copyright (c) 1982, 1986, 1990, 1991, 1993
+ *     The Regents of the University of California.  All rights reserved.
+ * (c) UNIX System Laboratories, Inc.
+ * All or some portions of this file are derived from material licensed
+ * to the University of California by American Telephone and Telegraph
+ * Co. or Unix System Laboratories, Inc. and are reproduced herein with
+ * the permission of UNIX System Laboratories, Inc.
+ *
+ * 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 the University of
+ *     California, Berkeley and its contributors.
+ * 4. Neither the name of the University nor the names of its contributors
+ *    may be used to endorse or promote products derived from this software
+ *    without specific prior written permission.

  *
  *

- *     @(#)kern_synch.c        7.2 (Berkeley) %G%
+ * THIS SOFTWARE IS PROVIDED BY THE REGENTS 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 REGENTS 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.
+ *
+ *     @(#)kern_synch.c        8.6 (Berkeley) 1/21/94

  */
 
  */
 

-#include "../machine/pte.h"
-#include "../machine/psl.h"
-#include "../machine/mtpr.h"
-
-#include "param.h"
-#include "systm.h"
-#include "dir.h"
-#include "user.h"
-#include "proc.h"
-#include "file.h"
-#include "inode.h"
-#include "vm.h"
-#include "kernel.h"
-#include "buf.h"
+#include <sys/param.h>
+#include <sys/systm.h>
+#include <sys/proc.h>
+#include <sys/kernel.h>
+#include <sys/buf.h>
+#include <sys/signalvar.h>
+#include <sys/resourcevar.h>
+#include <sys/vmmeter.h>
+#ifdef KTRACE
+#include <sys/ktrace.h>
+#endif
+
+#include <machine/cpu.h>
+
+u_char curpriority;            /* usrpri of curproc */
+int    lbolt;                  /* once a second sleep address */

 
 /*
  * Force switch among equal priority processes every 100ms.
  */
 
 /*
  * Force switch among equal priority processes every 100ms.
  */

-roundrobin()
+/* ARGSUSED */
+void
+roundrobin(arg)
+       void *arg;

 {
 
 {
 

-       runrun++;
-       aston();
-       timeout(roundrobin, (caddr_t)0, hz / 10);
+       need_resched();
+       timeout(roundrobin, NULL, hz / 10);

 }
 
 }
 

-/* fraction for digital decay to forget 90% of usage in 5*loadav sec */
-#define        filter(loadav) ((2 * (loadav)) / (2 * (loadav) + 1))
+/*
+ * Constants for digital decay and forget:
+ *     90% of (p_estcpu) usage in 5 * loadav time
+ *     95% of (p_pctcpu) usage in 60 seconds (load insensitive)
+ *          Note that, as ps(1) mentions, this can let percentages
+ *          total over 100% (I've seen 137.9% for 3 processes).
+ *
+ * Note that hardclock updates p_estcpu and p_cpticks independently.
+ *
+ * We wish to decay away 90% of p_estcpu in (5 * loadavg) seconds.
+ * That is, the system wants to compute a value of decay such
+ * that the following for loop:
+ *     for (i = 0; i < (5 * loadavg); i++)
+ *             p_estcpu *= decay;
+ * will compute
+ *     p_estcpu *= 0.1;
+ * for all values of loadavg:
+ *
+ * Mathematically this loop can be expressed by saying:
+ *     decay ** (5 * loadavg) ~= .1
+ *
+ * The system computes decay as:
+ *     decay = (2 * loadavg) / (2 * loadavg + 1)
+ *
+ * We wish to prove that the system's computation of decay
+ * will always fulfill the equation:
+ *     decay ** (5 * loadavg) ~= .1
+ *
+ * If we compute b as:
+ *     b = 2 * loadavg
+ * then
+ *     decay = b / (b + 1)
+ *
+ * We now need to prove two things:
+ *     1) Given factor ** (5 * loadavg) ~= .1, prove factor == b/(b+1)
+ *     2) Given b/(b+1) ** power ~= .1, prove power == (5 * loadavg)
+ *     
+ * Facts:
+ *         For x close to zero, exp(x) =~ 1 + x, since
+ *              exp(x) = 0! + x**1/1! + x**2/2! + ... .
+ *              therefore exp(-1/b) =~ 1 - (1/b) = (b-1)/b.
+ *         For x close to zero, ln(1+x) =~ x, since
+ *              ln(1+x) = x - x**2/2 + x**3/3 - ...     -1 < x < 1
+ *              therefore ln(b/(b+1)) = ln(1 - 1/(b+1)) =~ -1/(b+1).
+ *         ln(.1) =~ -2.30
+ *
+ * Proof of (1):
+ *    Solve (factor)**(power) =~ .1 given power (5*loadav):
+ *     solving for factor,
+ *      ln(factor) =~ (-2.30/5*loadav), or
+ *      factor =~ exp(-1/((5/2.30)*loadav)) =~ exp(-1/(2*loadav)) =
+ *          exp(-1/b) =~ (b-1)/b =~ b/(b+1).                    QED
+ *
+ * Proof of (2):
+ *    Solve (factor)**(power) =~ .1 given factor == (b/(b+1)):
+ *     solving for power,
+ *      power*ln(b/(b+1)) =~ -2.30, or
+ *      power =~ 2.3 * (b + 1) = 4.6*loadav + 2.3 =~ 5*loadav.  QED
+ *
+ * Actual power values for the implemented algorithm are as follows:
+ *      loadav: 1       2       3       4
+ *      power:  5.68    10.32   14.94   19.55
+ */
+
+/* calculations for digital decay to forget 90% of usage in 5*loadav sec */
+#define        loadfactor(loadav)      (2 * (loadav))
+#define        decay_cpu(loadfac, cpu) (((loadfac) * (cpu)) / ((loadfac) + FSCALE))

 
 

-double ccpu = 0.95122942450071400909;          /* exp(-1/20) */
+/* decay 95% of `p_pctcpu' in 60 seconds; see CCPU_SHIFT before changing */
+fixpt_t        ccpu = 0.95122942450071400909 * FSCALE;         /* exp(-1/20) */

 
 /*
 
 /*

- * Recompute process priorities, once a second
+ * If `ccpu' is not equal to `exp(-1/20)' and you still want to use the
+ * faster/more-accurate formula, you'll have to estimate CCPU_SHIFT below
+ * and possibly adjust FSHIFT in "param.h" so that (FSHIFT >= CCPU_SHIFT).
+ *
+ * To estimate CCPU_SHIFT for exp(-1/20), the following formula was used:
+ *     1 - exp(-1/20) ~= 0.0487 ~= 0.0488 == 1 (fixed pt, *11* bits).
+ *
+ * If you dont want to bother with the faster/more-accurate formula, you
+ * can set CCPU_SHIFT to (FSHIFT + 1) which will use a slower/less-accurate
+ * (more general) method of calculating the %age of CPU used by a process.

  */
  */

-schedcpu()
+#define        CCPU_SHIFT      11
+
+/*
+ * Recompute process priorities, every hz ticks.
+ */
+/* ARGSUSED */
+void
+schedcpu(arg)
+       void *arg;

 {
 {

-       register double ccpu1 = (1.0 - ccpu) / (double)hz;
+       register fixpt_t loadfac = loadfactor(averunnable.ldavg[0]);

        register struct proc *p;
        register struct proc *p;

-       register int s, a;
-       float scale = filter(avenrun[0]);
+       register int s;
+       register unsigned int newcpu;

 
        wakeup((caddr_t)&lbolt);
 
        wakeup((caddr_t)&lbolt);

-       for (p = allproc; p != NULL; p = p->p_nxt) {
-               if (p->p_time != 127)
-                       p->p_time++;
-               if (p->p_stat==SSLEEP || p->p_stat==SSTOP)
-                       if (p->p_slptime != 127)
-                               p->p_slptime++;
+       for (p = (struct proc *)allproc; p != NULL; p = p->p_next) {
+               /*
+                * Increment time in/out of memory and sleep time
+                * (if sleeping).  We ignore overflow; with 16-bit int's
+                * (remember them?) overflow takes 45 days.
+                */
+               p->p_swtime++;
+               if (p->p_stat == SSLEEP || p->p_stat == SSTOP)
+                       p->p_slptime++;
+               p->p_pctcpu = (p->p_pctcpu * ccpu) >> FSHIFT;

                /*
                 * If the process has slept the entire second,
                 * stop recalculating its priority until it wakes up.
                 */
                /*
                 * If the process has slept the entire second,
                 * stop recalculating its priority until it wakes up.
                 */

-               if (p->p_slptime > 1) {
-                       p->p_pctcpu *= ccpu;
+               if (p->p_slptime > 1)

                        continue;
                        continue;

-               }
+               s = splstatclock();     /* prevent state changes */

                /*
                 * p_pctcpu is only for ps.
                 */
                /*
                 * p_pctcpu is only for ps.
                 */

-               p->p_pctcpu = ccpu * p->p_pctcpu + ccpu1 * p->p_cpticks;
+#if    (FSHIFT >= CCPU_SHIFT)
+               p->p_pctcpu += (hz == 100)?
+                       ((fixpt_t) p->p_cpticks) << (FSHIFT - CCPU_SHIFT):
+                       100 * (((fixpt_t) p->p_cpticks)
+                               << (FSHIFT - CCPU_SHIFT)) / hz;
+#else
+               p->p_pctcpu += ((FSCALE - ccpu) *
+                       (p->p_cpticks * FSCALE / hz)) >> FSHIFT;
+#endif

                p->p_cpticks = 0;
                p->p_cpticks = 0;

-               a = (int) (scale * (p->p_cpu & 0377)) + p->p_nice;
-               if (a < 0)
-                       a = 0;
-               if (a > 255)
-                       a = 255;
-               p->p_cpu = a;
-               (void) setpri(p);
-               s = splhigh();  /* prevent state changes */
-               if (p->p_pri >= PUSER) {
-#define        PPQ     (128 / NQS)
-                       if ((p != u.u_procp || noproc) &&
+               newcpu = (u_int) decay_cpu(loadfac, p->p_estcpu) + p->p_nice;
+               p->p_estcpu = min(newcpu, UCHAR_MAX);
+               resetpriority(p);
+               if (p->p_priority >= PUSER) {
+#define        PPQ     (128 / NQS)             /* priorities per queue */
+                       if ((p != curproc) &&

                            p->p_stat == SRUN &&
                            p->p_stat == SRUN &&

-                           (p->p_flag & SLOAD) &&
-                           (p->p_pri / PPQ) != (p->p_usrpri / PPQ)) {
+                           (p->p_flag & P_INMEM) &&
+                           (p->p_priority / PPQ) != (p->p_usrpri / PPQ)) {

                                remrq(p);
                                remrq(p);

-                               p->p_pri = p->p_usrpri;
-                               setrq(p);
+                               p->p_priority = p->p_usrpri;
+                               setrunqueue(p);

                        } else
                        } else

-                               p->p_pri = p->p_usrpri;
+                               p->p_priority = p->p_usrpri;

                }
                splx(s);
        }
        vmmeter();
                }
                splx(s);
        }
        vmmeter();

-       if (runin!=0) {
-               runin = 0;
-               wakeup((caddr_t)&runin);
-       }

        if (bclnlist != NULL)
        if (bclnlist != NULL)

-               wakeup((caddr_t)&proc[2]);
-       timeout(schedcpu, (caddr_t)0, hz);
+               wakeup((caddr_t)pageproc);
+       timeout(schedcpu, (void *)0, hz);

 }
 
 /*
  * Recalculate the priority of a process after it has slept for a while.
 }
 
 /*
  * Recalculate the priority of a process after it has slept for a while.

+ * For all load averages >= 1 and max p_estcpu of 255, sleeping for at
+ * least six times the loadfactor will decay p_estcpu to zero.

  */
  */

+void

 updatepri(p)
        register struct proc *p;
 {
 updatepri(p)
        register struct proc *p;
 {

-       register int a = p->p_cpu & 0377;
-       float scale = filter(avenrun[0]);
-
-       p->p_slptime--;         /* the first time was done in schedcpu */
-       while (a && --p->p_slptime)
-               a = (int) (scale * a) /* + p->p_nice */;
-       if (a < 0)
-               a = 0;
-       if (a > 255)
-               a = 255;
-       p->p_cpu = a;
-       (void) setpri(p);
+       register unsigned int newcpu = p->p_estcpu;
+       register fixpt_t loadfac = loadfactor(averunnable.ldavg[0]);
+
+       if (p->p_slptime > 5 * loadfac)
+               p->p_estcpu = 0;
+       else {
+               p->p_slptime--; /* the first time was done in schedcpu */
+               while (newcpu && --p->p_slptime)
+                       newcpu = (int) decay_cpu(loadfac, newcpu);
+               p->p_estcpu = min(newcpu, UCHAR_MAX);
+       }
+       resetpriority(p);

 }
 
 }
 

-#define SQSIZE 0100    /* Must be power of 2 */
-#define HASH(x)        (( (int) x >> 5) & (SQSIZE-1))
+/*
+ * We're only looking at 7 bits of the address; everything is
+ * aligned to 4, lots of things are aligned to greater powers
+ * of 2.  Shift right by 8, i.e. drop the bottom 256 worth.
+ */
+#define TABLESIZE      128
+#define LOOKUP(x)      (((int)(x) >> 8) & (TABLESIZE - 1))

 struct slpque {
        struct proc *sq_head;
        struct proc **sq_tailp;
 struct slpque {
        struct proc *sq_head;
        struct proc **sq_tailp;

-} slpque[SQSIZE];
+} slpque[TABLESIZE];

 
 /*
 
 /*

- * Give up the processor till a wakeup occurs
- * on chan, at which time the process
- * enters the scheduling queue at priority pri.
- * The most important effect of pri is that when
- * pri<=PZERO a signal cannot disturb the sleep;
- * if pri>PZERO signals will be processed.
- * Callers of this routine must be prepared for
- * premature return, and check that the reason for
- * sleeping has gone away.
+ * During autoconfiguration or after a panic, a sleep will simply
+ * lower the priority briefly to allow interrupts, then return.
+ * The priority to be used (safepri) is machine-dependent, thus this
+ * value is initialized and maintained in the machine-dependent layers.
+ * This priority will typically be 0, or the lowest priority
+ * that is safe for use on the interrupt stack; it can be made
+ * higher to block network software interrupts after panics.

  */
  */

-sleep(chan, pri)
-       caddr_t chan;
-       int pri;
+int safepri;
+
+/*
+ * General sleep call.  Suspends the current process until a wakeup is
+ * performed on the specified identifier.  The process will then be made
+ * runnable with the specified priority.  Sleeps at most timo/hz seconds
+ * (0 means no timeout).  If pri includes PCATCH flag, signals are checked
+ * before and after sleeping, else signals are not checked.  Returns 0 if
+ * awakened, EWOULDBLOCK if the timeout expires.  If PCATCH is set and a
+ * signal needs to be delivered, ERESTART is returned if the current system
+ * call should be restarted if possible, and EINTR is returned if the system
+ * call should be interrupted by the signal (return EINTR).
+ */
+int
+tsleep(ident, priority, wmesg, timo)
+       void *ident;
+       int priority, timo;
+       char *wmesg;

 {
 {

-       register struct proc *rp;
+       register struct proc *p = curproc;

        register struct slpque *qp;
        register s;
        register struct slpque *qp;
        register s;

+       int sig, catch = priority & PCATCH;
+       extern int cold;
+       void endtsleep __P((void *));

 
 

-       rp = u.u_procp;
+#ifdef KTRACE
+       if (KTRPOINT(p, KTR_CSW))
+               ktrcsw(p->p_tracep, 1, 0);
+#endif

        s = splhigh();
        s = splhigh();

-       if (panicstr) {
+       if (cold || panicstr) {

                /*
                /*

-                * After a panic, just give interrupts a chance,
-                * then just return; don't run any other procs 
-                * or panic below, in case this is the idle process
-                * and already asleep.
-                * The splnet should be spl0 if the network was being used
-                * by the filesystem, but for now avoid network interrupts
-                * that might cause another panic.
+                * After a panic, or during autoconfiguration,
+                * just give interrupts a chance, then just return;
+                * don't run any other procs or panic below,
+                * in case this is the idle process and already asleep.

                 */
                 */

-               (void) splnet();
+               splx(safepri);

                splx(s);
                splx(s);

-               return;
+               return (0);

        }
        }

-       if (chan==0 || rp->p_stat != SRUN || rp->p_rlink)
-               panic("sleep");
-       rp->p_wchan = chan;
-       rp->p_slptime = 0;
-       rp->p_pri = pri;
-       qp = &slpque[HASH(chan)];
+#ifdef DIAGNOSTIC
+       if (ident == NULL || p->p_stat != SRUN || p->p_back)
+               panic("tsleep");
+#endif
+       p->p_wchan = ident;
+       p->p_wmesg = wmesg;
+       p->p_slptime = 0;
+       p->p_priority = priority & PRIMASK;
+       qp = &slpque[LOOKUP(ident)];

        if (qp->sq_head == 0)
        if (qp->sq_head == 0)

-               qp->sq_head = rp;
+               qp->sq_head = p;

        else
        else

-               *qp->sq_tailp = rp;
-       *(qp->sq_tailp = &rp->p_link) = 0;
-       if (pri > PZERO) {
-               /*
-                * If we stop in issig(), wakeup may already have happened
-                * when we return (rp->p_wchan will then be 0).
-                */
-               if (ISSIG(rp)) {
-                       if (rp->p_wchan)
-                               unsleep(rp);
-                       rp->p_stat = SRUN;
-                       (void) spl0();
-                       goto psig;
+               *qp->sq_tailp = p;
+       *(qp->sq_tailp = &p->p_forw) = 0;
+       if (timo)
+               timeout(endtsleep, (void *)p, timo);
+       /*
+        * We put ourselves on the sleep queue and start our timeout
+        * before calling CURSIG, as we could stop there, and a wakeup
+        * or a SIGCONT (or both) could occur while we were stopped.
+        * A SIGCONT would cause us to be marked as SSLEEP
+        * without resuming us, thus we must be ready for sleep
+        * when CURSIG is called.  If the wakeup happens while we're
+        * stopped, p->p_wchan will be 0 upon return from CURSIG.
+        */
+       if (catch) {
+               p->p_flag |= P_SINTR;
+               if (sig = CURSIG(p)) {
+                       if (p->p_wchan)
+                               unsleep(p);
+                       p->p_stat = SRUN;
+                       goto resume;
+               }
+               if (p->p_wchan == 0) {
+                       catch = 0;
+                       goto resume;
+               }
+       } else
+               sig = 0;
+       p->p_stat = SSLEEP;
+       p->p_stats->p_ru.ru_nvcsw++;
+       mi_switch();
+resume:
+       curpriority = p->p_usrpri;
+       splx(s);
+       p->p_flag &= ~P_SINTR;
+       if (p->p_flag & P_TIMEOUT) {
+               p->p_flag &= ~P_TIMEOUT;
+               if (sig == 0) {
+#ifdef KTRACE
+                       if (KTRPOINT(p, KTR_CSW))
+                               ktrcsw(p->p_tracep, 0, 0);
+#endif
+                       return (EWOULDBLOCK);

                }
                }

-               if (rp->p_wchan == 0)
-                       goto out;
-               rp->p_stat = SSLEEP;
-               (void) spl0();
-               u.u_ru.ru_nvcsw++;
-               swtch();
-               if (ISSIG(rp))
-                       goto psig;
-       } else {
-               rp->p_stat = SSLEEP;
-               (void) spl0();
-               u.u_ru.ru_nvcsw++;
-               swtch();
+       } else if (timo)
+               untimeout(endtsleep, (void *)p);
+       if (catch && (sig != 0 || (sig = CURSIG(p)))) {
+#ifdef KTRACE
+               if (KTRPOINT(p, KTR_CSW))
+                       ktrcsw(p->p_tracep, 0, 0);
+#endif
+               if (p->p_sigacts->ps_sigintr & sigmask(sig))
+                       return (EINTR);
+               return (ERESTART);
+       }
+#ifdef KTRACE
+       if (KTRPOINT(p, KTR_CSW))
+               ktrcsw(p->p_tracep, 0, 0);
+#endif
+       return (0);
+}
+
+/*
+ * Implement timeout for tsleep.
+ * If process hasn't been awakened (wchan non-zero),
+ * set timeout flag and undo the sleep.  If proc
+ * is stopped, just unsleep so it will remain stopped.
+ */
+void
+endtsleep(arg)
+       void *arg;
+{
+       register struct proc *p;
+       int s;
+
+       p = (struct proc *)arg;
+       s = splhigh();
+       if (p->p_wchan) {
+               if (p->p_stat == SSLEEP)
+                       setrunnable(p);
+               else
+                       unsleep(p);
+               p->p_flag |= P_TIMEOUT;

        }
        }

-       curpri = rp->p_usrpri;
-out:

        splx(s);
        splx(s);

-       return;
+}

 
 

-       /*
-        * If priority was low (>PZERO) and
-        * there has been a signal, execute non-local goto through
-        * u.u_qsave, aborting the system call in progress (see trap.c)
-        */
-psig:
-       longjmp(&u.u_qsave);
-       /*NOTREACHED*/
+/*
+ * Short-term, non-interruptable sleep.
+ */
+void
+sleep(ident, priority)
+       void *ident;
+       int priority;
+{
+       register struct proc *p = curproc;
+       register struct slpque *qp;
+       register s;
+       extern int cold;
+
+#ifdef DIAGNOSTIC
+       if (priority > PZERO) {
+               printf("sleep called with priority %d > PZERO, wchan: %x\n",
+                   priority, ident);
+               panic("old sleep");
+       }
+#endif
+       s = splhigh();
+       if (cold || panicstr) {
+               /*
+                * After a panic, or during autoconfiguration,
+                * just give interrupts a chance, then just return;
+                * don't run any other procs or panic below,
+                * in case this is the idle process and already asleep.
+                */
+               splx(safepri);
+               splx(s);
+               return;
+       }
+#ifdef DIAGNOSTIC
+       if (ident == NULL || p->p_stat != SRUN || p->p_back)
+               panic("sleep");
+#endif
+       p->p_wchan = ident;
+       p->p_wmesg = NULL;
+       p->p_slptime = 0;
+       p->p_priority = priority;
+       qp = &slpque[LOOKUP(ident)];
+       if (qp->sq_head == 0)
+               qp->sq_head = p;
+       else
+               *qp->sq_tailp = p;
+       *(qp->sq_tailp = &p->p_forw) = 0;
+       p->p_stat = SSLEEP;
+       p->p_stats->p_ru.ru_nvcsw++;
+#ifdef KTRACE
+       if (KTRPOINT(p, KTR_CSW))
+               ktrcsw(p->p_tracep, 1, 0);
+#endif
+       mi_switch();
+#ifdef KTRACE
+       if (KTRPOINT(p, KTR_CSW))
+               ktrcsw(p->p_tracep, 0, 0);
+#endif
+       curpriority = p->p_usrpri;
+       splx(s);

 }
 
 /*
  * Remove a process from its wait queue
  */
 }
 
 /*
  * Remove a process from its wait queue
  */

+void

 unsleep(p)
        register struct proc *p;
 {
 unsleep(p)
        register struct proc *p;
 {


@@ -225,11 +478,11 @@ unsleep(p)
 
        s = splhigh();
        if (p->p_wchan) {
 
        s = splhigh();
        if (p->p_wchan) {

-               hp = &(qp = &slpque[HASH(p->p_wchan)])->sq_head;
+               hp = &(qp = &slpque[LOOKUP(p->p_wchan)])->sq_head;

                while (*hp != p)
                while (*hp != p)

-                       hp = &(*hp)->p_link;
-               *hp = p->p_link;
-               if (qp->sq_tailp == &p->p_link)
+                       hp = &(*hp)->p_forw;
+               *hp = p->p_forw;
+               if (qp->sq_tailp == &p->p_forw)

                        qp->sq_tailp = hp;
                p->p_wchan = 0;
        }
                        qp->sq_tailp = hp;
                p->p_wchan = 0;
        }


@@ -237,57 +490,112 @@ unsleep(p)
 }
 
 /*
 }
 
 /*

- * Wake up all processes sleeping on chan.
+ * Make all processes sleeping on the specified identifier runnable.

  */
  */

-wakeup(chan)
-       register caddr_t chan;
+void
+wakeup(ident)
+       register void *ident;

 {
        register struct slpque *qp;
        register struct proc *p, **q;
        int s;
 
        s = splhigh();
 {
        register struct slpque *qp;
        register struct proc *p, **q;
        int s;
 
        s = splhigh();

-       qp = &slpque[HASH(chan)];
+       qp = &slpque[LOOKUP(ident)];

 restart:
        for (q = &qp->sq_head; p = *q; ) {
 restart:
        for (q = &qp->sq_head; p = *q; ) {

-               if (p->p_rlink || p->p_stat != SSLEEP && p->p_stat != SSTOP)
+#ifdef DIAGNOSTIC
+               if (p->p_back || p->p_stat != SSLEEP && p->p_stat != SSTOP)

                        panic("wakeup");
                        panic("wakeup");

-               if (p->p_wchan==chan) {
+#endif
+               if (p->p_wchan == ident) {

                        p->p_wchan = 0;
                        p->p_wchan = 0;

-                       *q = p->p_link;
-                       if (qp->sq_tailp == &p->p_link)
+                       *q = p->p_forw;
+                       if (qp->sq_tailp == &p->p_forw)

                                qp->sq_tailp = q;
                        if (p->p_stat == SSLEEP) {
                                qp->sq_tailp = q;
                        if (p->p_stat == SSLEEP) {

-                               /* OPTIMIZED INLINE EXPANSION OF setrun(p) */
+                               /* OPTIMIZED EXPANSION OF setrunnable(p); */

                                if (p->p_slptime > 1)
                                        updatepri(p);
                                p->p_slptime = 0;
                                p->p_stat = SRUN;
                                if (p->p_slptime > 1)
                                        updatepri(p);
                                p->p_slptime = 0;
                                p->p_stat = SRUN;

-                               if (p->p_flag & SLOAD)
-                                       setrq(p);
+                               if (p->p_flag & P_INMEM)
+                                       setrunqueue(p);

                                /*
                                /*

-                                * Since curpri is a usrpri,
-                                * p->p_pri is always better than curpri.
+                                * Since curpriority is a user priority,
+                                * p->p_priority is always better than
+                                * curpriority.

                                 */
                                 */

-                               runrun++;
-                               aston();
-                               if ((p->p_flag&SLOAD) == 0) {
-                                       if (runout != 0) {
-                                               runout = 0;
-                                               wakeup((caddr_t)&runout);
-                                       }
-                                       wantin++;
-                               }
+                               if ((p->p_flag & P_INMEM) == 0)
+                                       wakeup((caddr_t)&proc0);
+                               else
+                                       need_resched();

                                /* END INLINE EXPANSION */
                                goto restart;
                        }
                                /* END INLINE EXPANSION */
                                goto restart;
                        }

-                       p->p_slptime = 0;

                } else
                } else

-                       q = &p->p_link;
+                       q = &p->p_forw;

        }
        splx(s);
 }
 
        }
        splx(s);
 }
 

+/*
+ * The machine independent parts of mi_switch().
+ * Must be called at splstatclock() or higher.
+ */
+void
+mi_switch()
+{
+       register struct proc *p = curproc;      /* XXX */
+       register struct rlimit *rlim;
+       register long s, u;
+       struct timeval tv;
+
+       /*
+        * Compute the amount of time during which the current
+        * process was running, and add that to its total so far.
+        */
+       microtime(&tv);
+       u = p->p_rtime.tv_usec + (tv.tv_usec - runtime.tv_usec);
+       s = p->p_rtime.tv_sec + (tv.tv_sec - runtime.tv_sec);
+       if (u < 0) {
+               u += 1000000;
+               s--;
+       } else if (u >= 1000000) {
+               u -= 1000000;
+               s++;
+       }
+       p->p_rtime.tv_usec = u;
+       p->p_rtime.tv_sec = s;
+
+       /*
+        * Check if the process exceeds its cpu resource allocation.
+        * If over max, kill it.  In any case, if it has run for more
+        * than 10 minutes, reduce priority to give others a chance.
+        */
+       rlim = &p->p_rlimit[RLIMIT_CPU];
+       if (s >= rlim->rlim_cur) {
+               if (s >= rlim->rlim_max)
+                       psignal(p, SIGKILL);
+               else {
+                       psignal(p, SIGXCPU);
+                       if (rlim->rlim_cur < rlim->rlim_max)
+                               rlim->rlim_cur += 5;
+               }
+       }
+       if (s > 10 * 60 && p->p_ucred->cr_uid && p->p_nice == NZERO) {
+               p->p_nice = NZERO + 4;
+               resetpriority(p);
+       }
+
+       /*
+        * Pick a new current process and record its start time.
+        */
+       cnt.v_swtch++;
+       cpu_switch(p);
+       microtime(&runtime);
+}
+

 /*
  * Initialize the (doubly-linked) run queues
  * to be empty.
 /*
  * Initialize the (doubly-linked) run queues
  * to be empty.


@@ -301,24 +609,23 @@ rqinit()
 }
 
 /*
 }
 
 /*

- * Set the process running;
- * arrange for it to be swapped in if necessary.
+ * Change process state to be runnable,
+ * placing it on the run queue if it is in memory,
+ * and awakening the swapper if it isn't in memory.

  */
  */

-setrun(p)
+void
+setrunnable(p)

        register struct proc *p;
 {
        register int s;
 
        s = splhigh();
        switch (p->p_stat) {
        register struct proc *p;
 {
        register int s;
 
        s = splhigh();
        switch (p->p_stat) {

-

        case 0:
        case 0:

-       case SWAIT:

        case SRUN:
        case SZOMB:
        default:
        case SRUN:
        case SZOMB:
        default:

-               panic("setrun");
-
+               panic("setrunnable");

        case SSTOP:
        case SSLEEP:
                unsleep(p);             /* e.g. when sending signals */
        case SSTOP:
        case SSLEEP:
                unsleep(p);             /* e.g. when sending signals */


@@ -327,46 +634,33 @@ setrun(p)
        case SIDL:
                break;
        }
        case SIDL:
                break;
        }

-       if (p->p_slptime > 1)
-               updatepri(p);

        p->p_stat = SRUN;
        p->p_stat = SRUN;

-       if (p->p_flag & SLOAD)
-               setrq(p);
+       if (p->p_flag & P_INMEM)
+               setrunqueue(p);

        splx(s);
        splx(s);

-       if (p->p_pri < curpri) {
-               runrun++;
-               aston();
-       }
-       if ((p->p_flag&SLOAD) == 0) {
-               if (runout != 0) {
-                       runout = 0;
-                       wakeup((caddr_t)&runout);
-               }
-               wantin++;
-       }
+       if (p->p_slptime > 1)
+               updatepri(p);
+       p->p_slptime = 0;
+       if ((p->p_flag & P_INMEM) == 0)
+               wakeup((caddr_t)&proc0);
+       else if (p->p_priority < curpriority)
+               need_resched();

 }
 
 /*
 }
 
 /*

- * Set user priority.
- * The rescheduling flag (runrun)
- * is set if the priority is better
- * than the currently running process.
+ * Compute the priority of a process when running in user mode.
+ * Arrange to reschedule if the resulting priority is better
+ * than that of the current process.

  */
  */

-setpri(pp)
-       register struct proc *pp;
+void
+resetpriority(p)
+       register struct proc *p;

 {
 {

-       register int p;
-
-       p = (pp->p_cpu & 0377)/4;
-       p += PUSER + 2 * pp->p_nice;
-       if (pp->p_rssize > pp->p_maxrss && freemem < desfree)
-               p += 2*4;       /* effectively, nice(4) */
-       if (p > 127)
-               p = 127;
-       if (p < curpri) {
-               runrun++;
-               aston();
-       }
-       pp->p_usrpri = p;
-       return (p);
+       register unsigned int newpriority;
+
+       newpriority = PUSER + p->p_estcpu / 4 + 2 * p->p_nice;
+       newpriority = min(newpriority, MAXPRI);
+       p->p_usrpri = newpriority;
+       if (newpriority < curpriority)
+               need_resched();

 }
 }