/* $OpenBSD: sched_bsd.c,v 1.74 2023/02/04 19:33:03 cheloha Exp $ */ /* $NetBSD: kern_synch.c,v 1.37 1996/04/22 01:38:37 christos Exp $ */ /*- * 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. 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. * * 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 #include #include #include #include #include #include #include #include #include #include #ifdef KTRACE #include #endif int lbolt; /* once a second sleep address */ int rrticks_init; /* # of hardclock ticks per roundrobin() */ #ifdef MULTIPROCESSOR struct __mp_lock sched_lock; #endif void schedcpu(void *); uint32_t decay_aftersleep(uint32_t, uint32_t); /* * Force switch among equal priority processes every 100ms. */ void roundrobin(struct cpu_info *ci) { struct schedstate_percpu *spc = &ci->ci_schedstate; spc->spc_rrticks = rrticks_init; if (ci->ci_curproc != NULL) { if (spc->spc_schedflags & SPCF_SEENRR) { /* * The process has already been through a roundrobin * without switching and may be hogging the CPU. * Indicate that the process should yield. */ atomic_setbits_int(&spc->spc_schedflags, SPCF_SHOULDYIELD); } else { atomic_setbits_int(&spc->spc_schedflags, SPCF_SEENRR); } } if (spc->spc_nrun) need_resched(ci); } /* * 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)) /* decay 95% of `p_pctcpu' in 60 seconds; see CCPU_SHIFT before changing */ fixpt_t ccpu = 0.95122942450071400909 * FSCALE; /* exp(-1/20) */ /* * 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 don't 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. */ #define CCPU_SHIFT 11 /* * Recompute process priorities, every second. */ void schedcpu(void *arg) { struct timeout *to = (struct timeout *)arg; fixpt_t loadfac = loadfactor(averunnable.ldavg[0]); struct proc *p; int s; unsigned int newcpu; LIST_FOREACH(p, &allproc, p_list) { /* * Idle threads are never placed on the runqueue, * therefore computing their priority is pointless. */ if (p->p_cpu != NULL && p->p_cpu->ci_schedstate.spc_idleproc == p) continue; /* * Increment sleep time (if sleeping). We ignore overflow. */ 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 (p->p_slptime > 1) continue; SCHED_LOCK(s); /* * p_pctcpu is only for diagnostic tools such as ps. */ #if (FSHIFT >= CCPU_SHIFT) p->p_pctcpu += (stathz == 100)? ((fixpt_t) p->p_cpticks) << (FSHIFT - CCPU_SHIFT): 100 * (((fixpt_t) p->p_cpticks) << (FSHIFT - CCPU_SHIFT)) / stathz; #else p->p_pctcpu += ((FSCALE - ccpu) * (p->p_cpticks * FSCALE / stathz)) >> FSHIFT; #endif p->p_cpticks = 0; newcpu = (u_int) decay_cpu(loadfac, p->p_estcpu); setpriority(p, newcpu, p->p_p->ps_nice); if (p->p_stat == SRUN && (p->p_runpri / SCHED_PPQ) != (p->p_usrpri / SCHED_PPQ)) { remrunqueue(p); setrunqueue(p->p_cpu, p, p->p_usrpri); } SCHED_UNLOCK(s); } uvm_meter(); wakeup(&lbolt); timeout_add_sec(to, 1); } /* * 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. */ uint32_t decay_aftersleep(uint32_t estcpu, uint32_t slptime) { fixpt_t loadfac = loadfactor(averunnable.ldavg[0]); uint32_t newcpu; if (slptime > 5 * loadfac) newcpu = 0; else { newcpu = estcpu; slptime--; /* the first time was done in schedcpu */ while (newcpu && --slptime) newcpu = decay_cpu(loadfac, newcpu); } return (newcpu); } /* * General yield call. Puts the current process back on its run queue and * performs a voluntary context switch. */ void yield(void) { struct proc *p = curproc; int s; SCHED_LOCK(s); setrunqueue(p->p_cpu, p, p->p_usrpri); p->p_ru.ru_nvcsw++; mi_switch(); SCHED_UNLOCK(s); } /* * General preemption call. Puts the current process back on its run queue * and performs an involuntary context switch. If a process is supplied, * we switch to that process. Otherwise, we use the normal process selection * criteria. */ void preempt(void) { struct proc *p = curproc; int s; SCHED_LOCK(s); setrunqueue(p->p_cpu, p, p->p_usrpri); p->p_ru.ru_nivcsw++; mi_switch(); SCHED_UNLOCK(s); } void mi_switch(void) { struct schedstate_percpu *spc = &curcpu()->ci_schedstate; struct proc *p = curproc; struct proc *nextproc; struct process *pr = p->p_p; struct timespec ts; #ifdef MULTIPROCESSOR int hold_count; int sched_count; #endif assertwaitok(); KASSERT(p->p_stat != SONPROC); SCHED_ASSERT_LOCKED(); #ifdef MULTIPROCESSOR /* * Release the kernel_lock, as we are about to yield the CPU. */ sched_count = __mp_release_all_but_one(&sched_lock); if (_kernel_lock_held()) hold_count = __mp_release_all(&kernel_lock); else hold_count = 0; #endif /* * Compute the amount of time during which the current * process was running, and add that to its total so far. */ nanouptime(&ts); if (timespeccmp(&ts, &spc->spc_runtime, <)) { #if 0 printf("uptime is not monotonic! " "ts=%lld.%09lu, runtime=%lld.%09lu\n", (long long)tv.tv_sec, tv.tv_nsec, (long long)spc->spc_runtime.tv_sec, spc->spc_runtime.tv_nsec); #endif } else { timespecsub(&ts, &spc->spc_runtime, &ts); timespecadd(&p->p_rtime, &ts, &p->p_rtime); } /* add the time counts for this thread to the process's total */ tuagg_unlocked(pr, p); /* * Process is about to yield the CPU; clear the appropriate * scheduling flags. */ atomic_clearbits_int(&spc->spc_schedflags, SPCF_SWITCHCLEAR); nextproc = sched_chooseproc(); if (p != nextproc) { uvmexp.swtch++; TRACEPOINT(sched, off__cpu, nextproc->p_tid + THREAD_PID_OFFSET, nextproc->p_p->ps_pid); cpu_switchto(p, nextproc); TRACEPOINT(sched, on__cpu, NULL); } else { TRACEPOINT(sched, remain__cpu, NULL); p->p_stat = SONPROC; } clear_resched(curcpu()); SCHED_ASSERT_LOCKED(); /* * To preserve lock ordering, we need to release the sched lock * and grab it after we grab the big lock. * In the future, when the sched lock isn't recursive, we'll * just release it here. */ #ifdef MULTIPROCESSOR __mp_unlock(&sched_lock); #endif SCHED_ASSERT_UNLOCKED(); smr_idle(); /* * We're running again; record our new start time. We might * be running on a new CPU now, so don't use the cache'd * schedstate_percpu pointer. */ KASSERT(p->p_cpu == curcpu()); nanouptime(&p->p_cpu->ci_schedstate.spc_runtime); #ifdef MULTIPROCESSOR /* * Reacquire the kernel_lock now. We do this after we've * released the scheduler lock to avoid deadlock, and before * we reacquire the interlock and the scheduler lock. */ if (hold_count) __mp_acquire_count(&kernel_lock, hold_count); __mp_acquire_count(&sched_lock, sched_count + 1); #endif } /* * Change process state to be runnable, * placing it on the run queue. */ void setrunnable(struct proc *p) { struct process *pr = p->p_p; u_char prio; SCHED_ASSERT_LOCKED(); switch (p->p_stat) { case 0: case SRUN: case SONPROC: case SDEAD: case SIDL: default: panic("setrunnable"); case SSTOP: /* * If we're being traced (possibly because someone attached us * while we were stopped), check for a signal from the debugger. */ if ((pr->ps_flags & PS_TRACED) != 0 && pr->ps_xsig != 0) atomic_setbits_int(&p->p_siglist, sigmask(pr->ps_xsig)); prio = p->p_usrpri; unsleep(p); break; case SSLEEP: prio = p->p_slppri; unsleep(p); /* e.g. when sending signals */ break; } setrunqueue(NULL, p, prio); if (p->p_slptime > 1) { uint32_t newcpu; newcpu = decay_aftersleep(p->p_estcpu, p->p_slptime); setpriority(p, newcpu, pr->ps_nice); } p->p_slptime = 0; } /* * Compute the priority of a process. */ void setpriority(struct proc *p, uint32_t newcpu, uint8_t nice) { unsigned int newprio; newprio = min((PUSER + newcpu + NICE_WEIGHT * (nice - NZERO)), MAXPRI); SCHED_ASSERT_LOCKED(); p->p_estcpu = newcpu; p->p_usrpri = newprio; } /* * We adjust the priority of the current process. The priority of a process * gets worse as it accumulates CPU time. The cpu usage estimator (p_estcpu) * is increased here. The formula for computing priorities (in kern_synch.c) * will compute a different value each time p_estcpu increases. This can * cause a switch, but unless the priority crosses a PPQ boundary the actual * queue will not change. The cpu usage estimator ramps up quite quickly * when the process is running (linearly), and decays away exponentially, at * a rate which is proportionally slower when the system is busy. The basic * principle is that the system will 90% forget that the process used a lot * of CPU time in 5 * loadav seconds. This causes the system to favor * processes which haven't run much recently, and to round-robin among other * processes. */ void schedclock(struct proc *p) { struct cpu_info *ci = curcpu(); struct schedstate_percpu *spc = &ci->ci_schedstate; uint32_t newcpu; int s; if (p == spc->spc_idleproc || spc->spc_spinning) return; SCHED_LOCK(s); newcpu = ESTCPULIM(p->p_estcpu + 1); setpriority(p, newcpu, p->p_p->ps_nice); SCHED_UNLOCK(s); } void (*cpu_setperf)(int); #define PERFPOL_MANUAL 0 #define PERFPOL_AUTO 1 #define PERFPOL_HIGH 2 int perflevel = 100; int perfpolicy = PERFPOL_AUTO; #ifndef SMALL_KERNEL /* * The code below handles CPU throttling. */ #include void setperf_auto(void *); struct timeout setperf_to = TIMEOUT_INITIALIZER(setperf_auto, NULL); extern int hw_power; void setperf_auto(void *v) { static uint64_t *idleticks, *totalticks; static int downbeats; int i, j = 0; int speedup = 0; CPU_INFO_ITERATOR cii; struct cpu_info *ci; uint64_t idle, total, allidle = 0, alltotal = 0; if (perfpolicy != PERFPOL_AUTO) return; if (cpu_setperf == NULL) return; if (hw_power) { speedup = 1; goto faster; } if (!idleticks) if (!(idleticks = mallocarray(ncpusfound, sizeof(*idleticks), M_DEVBUF, M_NOWAIT | M_ZERO))) return; if (!totalticks) if (!(totalticks = mallocarray(ncpusfound, sizeof(*totalticks), M_DEVBUF, M_NOWAIT | M_ZERO))) { free(idleticks, M_DEVBUF, sizeof(*idleticks) * ncpusfound); return; } CPU_INFO_FOREACH(cii, ci) { if (!cpu_is_online(ci)) continue; total = 0; for (i = 0; i < CPUSTATES; i++) { total += ci->ci_schedstate.spc_cp_time[i]; } total -= totalticks[j]; idle = ci->ci_schedstate.spc_cp_time[CP_IDLE] - idleticks[j]; if (idle < total / 3) speedup = 1; alltotal += total; allidle += idle; idleticks[j] += idle; totalticks[j] += total; j++; } if (allidle < alltotal / 2) speedup = 1; if (speedup && downbeats < 5) downbeats++; if (speedup && perflevel != 100) { faster: perflevel = 100; cpu_setperf(perflevel); } else if (!speedup && perflevel != 0 && --downbeats <= 0) { perflevel = 0; cpu_setperf(perflevel); } timeout_add_msec(&setperf_to, 100); } int sysctl_hwsetperf(void *oldp, size_t *oldlenp, void *newp, size_t newlen) { int err; if (!cpu_setperf) return EOPNOTSUPP; if (perfpolicy != PERFPOL_MANUAL) return sysctl_rdint(oldp, oldlenp, newp, perflevel); err = sysctl_int_bounded(oldp, oldlenp, newp, newlen, &perflevel, 0, 100); if (err) return err; if (newp != NULL) cpu_setperf(perflevel); return 0; } int sysctl_hwperfpolicy(void *oldp, size_t *oldlenp, void *newp, size_t newlen) { char policy[32]; int err; if (!cpu_setperf) return EOPNOTSUPP; switch (perfpolicy) { case PERFPOL_MANUAL: strlcpy(policy, "manual", sizeof(policy)); break; case PERFPOL_AUTO: strlcpy(policy, "auto", sizeof(policy)); break; case PERFPOL_HIGH: strlcpy(policy, "high", sizeof(policy)); break; default: strlcpy(policy, "unknown", sizeof(policy)); break; } if (newp == NULL) return sysctl_rdstring(oldp, oldlenp, newp, policy); err = sysctl_string(oldp, oldlenp, newp, newlen, policy, sizeof(policy)); if (err) return err; if (strcmp(policy, "manual") == 0) perfpolicy = PERFPOL_MANUAL; else if (strcmp(policy, "auto") == 0) perfpolicy = PERFPOL_AUTO; else if (strcmp(policy, "high") == 0) perfpolicy = PERFPOL_HIGH; else return EINVAL; if (perfpolicy == PERFPOL_AUTO) { timeout_add_msec(&setperf_to, 200); } else if (perfpolicy == PERFPOL_HIGH) { perflevel = 100; cpu_setperf(perflevel); } return 0; } #endif void scheduler_start(void) { static struct timeout schedcpu_to; /* * We avoid polluting the global namespace by keeping the scheduler * timeouts static in this function. * We setup the timeout here and kick schedcpu once to make it do * its job. */ timeout_set(&schedcpu_to, schedcpu, &schedcpu_to); rrticks_init = hz / 10; schedcpu(&schedcpu_to); #ifndef SMALL_KERNEL if (perfpolicy == PERFPOL_AUTO) timeout_add_msec(&setperf_to, 200); #endif }