1 /* $NetBSD: sched_4bsd.c,v 1.24 2008/10/07 09:48:27 rmind Exp $ */ 2 3 /*- 4 * Copyright (c) 1999, 2000, 2004, 2006, 2007, 2008 The NetBSD Foundation, Inc. 5 * All rights reserved. 6 * 7 * This code is derived from software contributed to The NetBSD Foundation 8 * by Jason R. Thorpe of the Numerical Aerospace Simulation Facility, 9 * NASA Ames Research Center, by Charles M. Hannum, Andrew Doran, and 10 * Daniel Sieger. 11 * 12 * Redistribution and use in source and binary forms, with or without 13 * modification, are permitted provided that the following conditions 14 * are met: 15 * 1. Redistributions of source code must retain the above copyright 16 * notice, this list of conditions and the following disclaimer. 17 * 2. Redistributions in binary form must reproduce the above copyright 18 * notice, this list of conditions and the following disclaimer in the 19 * documentation and/or other materials provided with the distribution. 20 * 21 * THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS 22 * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED 23 * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR 24 * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE FOUNDATION OR CONTRIBUTORS 25 * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR 26 * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF 27 * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS 28 * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN 29 * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) 30 * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE 31 * POSSIBILITY OF SUCH DAMAGE. 32 */ 33 34 /*- 35 * Copyright (c) 1982, 1986, 1990, 1991, 1993 36 * The Regents of the University of California. All rights reserved. 37 * (c) UNIX System Laboratories, Inc. 38 * All or some portions of this file are derived from material licensed 39 * to the University of California by American Telephone and Telegraph 40 * Co. or Unix System Laboratories, Inc. and are reproduced herein with 41 * the permission of UNIX System Laboratories, Inc. 42 * 43 * Redistribution and use in source and binary forms, with or without 44 * modification, are permitted provided that the following conditions 45 * are met: 46 * 1. Redistributions of source code must retain the above copyright 47 * notice, this list of conditions and the following disclaimer. 48 * 2. Redistributions in binary form must reproduce the above copyright 49 * notice, this list of conditions and the following disclaimer in the 50 * documentation and/or other materials provided with the distribution. 51 * 3. Neither the name of the University nor the names of its contributors 52 * may be used to endorse or promote products derived from this software 53 * without specific prior written permission. 54 * 55 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 56 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 57 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 58 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 59 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 60 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 61 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 62 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 63 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 64 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 65 * SUCH DAMAGE. 66 * 67 * @(#)kern_synch.c 8.9 (Berkeley) 5/19/95 68 */ 69 70 #include <sys/cdefs.h> 71 __KERNEL_RCSID(0, "$NetBSD: sched_4bsd.c,v 1.24 2008/10/07 09:48:27 rmind Exp $"); 72 73 #include "opt_ddb.h" 74 #include "opt_lockdebug.h" 75 #include "opt_perfctrs.h" 76 77 #include <sys/param.h> 78 #include <sys/systm.h> 79 #include <sys/callout.h> 80 #include <sys/cpu.h> 81 #include <sys/proc.h> 82 #include <sys/kernel.h> 83 #include <sys/signalvar.h> 84 #include <sys/resourcevar.h> 85 #include <sys/sched.h> 86 #include <sys/sysctl.h> 87 #include <sys/kauth.h> 88 #include <sys/lockdebug.h> 89 #include <sys/kmem.h> 90 #include <sys/intr.h> 91 92 #include <uvm/uvm_extern.h> 93 94 static void updatepri(struct lwp *); 95 static void resetpriority(struct lwp *); 96 97 extern unsigned int sched_pstats_ticks; /* defined in kern_synch.c */ 98 99 /* Number of hardclock ticks per sched_tick() */ 100 static int rrticks; 101 102 /* 103 * Force switch among equal priority processes every 100ms. 104 * Called from hardclock every hz/10 == rrticks hardclock ticks. 105 * 106 * There's no need to lock anywhere in this routine, as it's 107 * CPU-local and runs at IPL_SCHED (called from clock interrupt). 108 */ 109 /* ARGSUSED */ 110 void 111 sched_tick(struct cpu_info *ci) 112 { 113 struct schedstate_percpu *spc = &ci->ci_schedstate; 114 lwp_t *l; 115 116 spc->spc_ticks = rrticks; 117 118 if (CURCPU_IDLE_P()) { 119 cpu_need_resched(ci, 0); 120 return; 121 } 122 l = ci->ci_data.cpu_onproc; 123 if (l == NULL) { 124 return; 125 } 126 switch (l->l_class) { 127 case SCHED_FIFO: 128 /* No timeslicing for FIFO jobs. */ 129 break; 130 case SCHED_RR: 131 /* Force it into mi_switch() to look for other jobs to run. */ 132 cpu_need_resched(ci, RESCHED_KPREEMPT); 133 break; 134 default: 135 if (spc->spc_flags & SPCF_SHOULDYIELD) { 136 /* 137 * Process is stuck in kernel somewhere, probably 138 * due to buggy or inefficient code. Force a 139 * kernel preemption. 140 */ 141 cpu_need_resched(ci, RESCHED_KPREEMPT); 142 } else if (spc->spc_flags & SPCF_SEENRR) { 143 /* 144 * The process has already been through a roundrobin 145 * without switching and may be hogging the CPU. 146 * Indicate that the process should yield. 147 */ 148 spc->spc_flags |= SPCF_SHOULDYIELD; 149 cpu_need_resched(ci, 0); 150 } else { 151 spc->spc_flags |= SPCF_SEENRR; 152 } 153 break; 154 } 155 } 156 157 /* 158 * Why PRIO_MAX - 2? From setpriority(2): 159 * 160 * prio is a value in the range -20 to 20. The default priority is 161 * 0; lower priorities cause more favorable scheduling. A value of 162 * 19 or 20 will schedule a process only when nothing at priority <= 163 * 0 is runnable. 164 * 165 * This gives estcpu influence over 18 priority levels, and leaves nice 166 * with 40 levels. One way to think about it is that nice has 20 levels 167 * either side of estcpu's 18. 168 */ 169 #define ESTCPU_SHIFT 11 170 #define ESTCPU_MAX ((PRIO_MAX - 2) << ESTCPU_SHIFT) 171 #define ESTCPU_ACCUM (1 << (ESTCPU_SHIFT - 1)) 172 #define ESTCPULIM(e) min((e), ESTCPU_MAX) 173 174 /* 175 * Constants for digital decay and forget: 176 * 90% of (l_estcpu) usage in 5 * loadav time 177 * 95% of (l_pctcpu) usage in 60 seconds (load insensitive) 178 * Note that, as ps(1) mentions, this can let percentages 179 * total over 100% (I've seen 137.9% for 3 processes). 180 * 181 * Note that hardclock updates l_estcpu and l_cpticks independently. 182 * 183 * We wish to decay away 90% of l_estcpu in (5 * loadavg) seconds. 184 * That is, the system wants to compute a value of decay such 185 * that the following for loop: 186 * for (i = 0; i < (5 * loadavg); i++) 187 * l_estcpu *= decay; 188 * will compute 189 * l_estcpu *= 0.1; 190 * for all values of loadavg: 191 * 192 * Mathematically this loop can be expressed by saying: 193 * decay ** (5 * loadavg) ~= .1 194 * 195 * The system computes decay as: 196 * decay = (2 * loadavg) / (2 * loadavg + 1) 197 * 198 * We wish to prove that the system's computation of decay 199 * will always fulfill the equation: 200 * decay ** (5 * loadavg) ~= .1 201 * 202 * If we compute b as: 203 * b = 2 * loadavg 204 * then 205 * decay = b / (b + 1) 206 * 207 * We now need to prove two things: 208 * 1) Given factor ** (5 * loadavg) ~= .1, prove factor == b/(b+1) 209 * 2) Given b/(b+1) ** power ~= .1, prove power == (5 * loadavg) 210 * 211 * Facts: 212 * For x close to zero, exp(x) =~ 1 + x, since 213 * exp(x) = 0! + x**1/1! + x**2/2! + ... . 214 * therefore exp(-1/b) =~ 1 - (1/b) = (b-1)/b. 215 * For x close to zero, ln(1+x) =~ x, since 216 * ln(1+x) = x - x**2/2 + x**3/3 - ... -1 < x < 1 217 * therefore ln(b/(b+1)) = ln(1 - 1/(b+1)) =~ -1/(b+1). 218 * ln(.1) =~ -2.30 219 * 220 * Proof of (1): 221 * Solve (factor)**(power) =~ .1 given power (5*loadav): 222 * solving for factor, 223 * ln(factor) =~ (-2.30/5*loadav), or 224 * factor =~ exp(-1/((5/2.30)*loadav)) =~ exp(-1/(2*loadav)) = 225 * exp(-1/b) =~ (b-1)/b =~ b/(b+1). QED 226 * 227 * Proof of (2): 228 * Solve (factor)**(power) =~ .1 given factor == (b/(b+1)): 229 * solving for power, 230 * power*ln(b/(b+1)) =~ -2.30, or 231 * power =~ 2.3 * (b + 1) = 4.6*loadav + 2.3 =~ 5*loadav. QED 232 * 233 * Actual power values for the implemented algorithm are as follows: 234 * loadav: 1 2 3 4 235 * power: 5.68 10.32 14.94 19.55 236 */ 237 238 /* calculations for digital decay to forget 90% of usage in 5*loadav sec */ 239 #define loadfactor(loadav) (2 * (loadav)) 240 241 static fixpt_t 242 decay_cpu(fixpt_t loadfac, fixpt_t estcpu) 243 { 244 245 if (estcpu == 0) { 246 return 0; 247 } 248 249 #if !defined(_LP64) 250 /* avoid 64bit arithmetics. */ 251 #define FIXPT_MAX ((fixpt_t)((UINTMAX_C(1) << sizeof(fixpt_t) * CHAR_BIT) - 1)) 252 if (__predict_true(loadfac <= FIXPT_MAX / ESTCPU_MAX)) { 253 return estcpu * loadfac / (loadfac + FSCALE); 254 } 255 #endif /* !defined(_LP64) */ 256 257 return (uint64_t)estcpu * loadfac / (loadfac + FSCALE); 258 } 259 260 /* 261 * For all load averages >= 1 and max l_estcpu of (255 << ESTCPU_SHIFT), 262 * sleeping for at least seven times the loadfactor will decay l_estcpu to 263 * less than (1 << ESTCPU_SHIFT). 264 * 265 * note that our ESTCPU_MAX is actually much smaller than (255 << ESTCPU_SHIFT). 266 */ 267 static fixpt_t 268 decay_cpu_batch(fixpt_t loadfac, fixpt_t estcpu, unsigned int n) 269 { 270 271 if ((n << FSHIFT) >= 7 * loadfac) { 272 return 0; 273 } 274 275 while (estcpu != 0 && n > 1) { 276 estcpu = decay_cpu(loadfac, estcpu); 277 n--; 278 } 279 280 return estcpu; 281 } 282 283 /* 284 * sched_pstats_hook: 285 * 286 * Periodically called from sched_pstats(); used to recalculate priorities. 287 */ 288 void 289 sched_pstats_hook(struct lwp *l, int batch) 290 { 291 292 /* 293 * If the LWP has slept an entire second, stop recalculating 294 * its priority until it wakes up. 295 */ 296 KASSERT(lwp_locked(l, NULL)); 297 if (l->l_slptime > 0) { 298 fixpt_t loadfac = 2 * (averunnable.ldavg[0]); 299 l->l_estcpu = decay_cpu(loadfac, l->l_estcpu); 300 resetpriority(l); 301 } 302 } 303 304 /* 305 * Recalculate the priority of a process after it has slept for a while. 306 */ 307 static void 308 updatepri(struct lwp *l) 309 { 310 fixpt_t loadfac; 311 312 KASSERT(lwp_locked(l, NULL)); 313 KASSERT(l->l_slptime > 1); 314 315 loadfac = loadfactor(averunnable.ldavg[0]); 316 317 l->l_slptime--; /* the first time was done in sched_pstats */ 318 l->l_estcpu = decay_cpu_batch(loadfac, l->l_estcpu, l->l_slptime); 319 resetpriority(l); 320 } 321 322 void 323 sched_rqinit(void) 324 { 325 326 } 327 328 void 329 sched_setrunnable(struct lwp *l) 330 { 331 332 if (l->l_slptime > 1) 333 updatepri(l); 334 } 335 336 void 337 sched_nice(struct proc *p, int n) 338 { 339 struct lwp *l; 340 341 KASSERT(mutex_owned(p->p_lock)); 342 343 p->p_nice = n; 344 LIST_FOREACH(l, &p->p_lwps, l_sibling) { 345 lwp_lock(l); 346 resetpriority(l); 347 lwp_unlock(l); 348 } 349 } 350 351 /* 352 * Recompute the priority of an LWP. Arrange to reschedule if 353 * the resulting priority is better than that of the current LWP. 354 */ 355 static void 356 resetpriority(struct lwp *l) 357 { 358 pri_t pri; 359 struct proc *p = l->l_proc; 360 361 KASSERT(lwp_locked(l, NULL)); 362 363 if (l->l_class != SCHED_OTHER) 364 return; 365 366 /* See comments above ESTCPU_SHIFT definition. */ 367 pri = (PRI_KERNEL - 1) - (l->l_estcpu >> ESTCPU_SHIFT) - p->p_nice; 368 pri = imax(pri, 0); 369 if (pri != l->l_priority) 370 lwp_changepri(l, pri); 371 } 372 373 /* 374 * We adjust the priority of the current process. The priority of a process 375 * gets worse as it accumulates CPU time. The CPU usage estimator (l_estcpu) 376 * is increased here. The formula for computing priorities (in kern_synch.c) 377 * will compute a different value each time l_estcpu increases. This can 378 * cause a switch, but unless the priority crosses a PPQ boundary the actual 379 * queue will not change. The CPU usage estimator ramps up quite quickly 380 * when the process is running (linearly), and decays away exponentially, at 381 * a rate which is proportionally slower when the system is busy. The basic 382 * principle is that the system will 90% forget that the process used a lot 383 * of CPU time in 5 * loadav seconds. This causes the system to favor 384 * processes which haven't run much recently, and to round-robin among other 385 * processes. 386 */ 387 388 void 389 sched_schedclock(struct lwp *l) 390 { 391 392 if (l->l_class != SCHED_OTHER) 393 return; 394 395 KASSERT(!CURCPU_IDLE_P()); 396 l->l_estcpu = ESTCPULIM(l->l_estcpu + ESTCPU_ACCUM); 397 lwp_lock(l); 398 resetpriority(l); 399 lwp_unlock(l); 400 } 401 402 /* 403 * sched_proc_fork: 404 * 405 * Inherit the parent's scheduler history. 406 */ 407 void 408 sched_proc_fork(struct proc *parent, struct proc *child) 409 { 410 lwp_t *pl; 411 412 KASSERT(mutex_owned(parent->p_lock)); 413 414 pl = LIST_FIRST(&parent->p_lwps); 415 child->p_estcpu_inherited = pl->l_estcpu; 416 child->p_forktime = sched_pstats_ticks; 417 } 418 419 /* 420 * sched_proc_exit: 421 * 422 * Chargeback parents for the sins of their children. 423 */ 424 void 425 sched_proc_exit(struct proc *parent, struct proc *child) 426 { 427 fixpt_t loadfac = loadfactor(averunnable.ldavg[0]); 428 fixpt_t estcpu; 429 lwp_t *pl, *cl; 430 431 /* XXX Only if parent != init?? */ 432 433 mutex_enter(parent->p_lock); 434 pl = LIST_FIRST(&parent->p_lwps); 435 cl = LIST_FIRST(&child->p_lwps); 436 estcpu = decay_cpu_batch(loadfac, child->p_estcpu_inherited, 437 sched_pstats_ticks - child->p_forktime); 438 if (cl->l_estcpu > estcpu) { 439 lwp_lock(pl); 440 pl->l_estcpu = ESTCPULIM(pl->l_estcpu + cl->l_estcpu - estcpu); 441 lwp_unlock(pl); 442 } 443 mutex_exit(parent->p_lock); 444 } 445 446 void 447 sched_wakeup(struct lwp *l) 448 { 449 450 } 451 452 void 453 sched_slept(struct lwp *l) 454 { 455 456 } 457 458 void 459 sched_lwp_fork(struct lwp *l1, struct lwp *l2) 460 { 461 462 l2->l_estcpu = l1->l_estcpu; 463 } 464 465 void 466 sched_lwp_collect(struct lwp *t) 467 { 468 lwp_t *l; 469 470 /* Absorb estcpu value of collected LWP. */ 471 l = curlwp; 472 lwp_lock(l); 473 l->l_estcpu += t->l_estcpu; 474 lwp_unlock(l); 475 } 476 477 void 478 sched_oncpu(lwp_t *l) 479 { 480 481 } 482 483 void 484 sched_newts(lwp_t *l) 485 { 486 487 } 488 489 /* 490 * Sysctl nodes and initialization. 491 */ 492 493 static int 494 sysctl_sched_rtts(SYSCTLFN_ARGS) 495 { 496 struct sysctlnode node; 497 int rttsms = hztoms(rrticks); 498 499 node = *rnode; 500 node.sysctl_data = &rttsms; 501 return sysctl_lookup(SYSCTLFN_CALL(&node)); 502 } 503 504 SYSCTL_SETUP(sysctl_sched_4bsd_setup, "sysctl sched setup") 505 { 506 const struct sysctlnode *node = NULL; 507 508 sysctl_createv(clog, 0, NULL, NULL, 509 CTLFLAG_PERMANENT, 510 CTLTYPE_NODE, "kern", NULL, 511 NULL, 0, NULL, 0, 512 CTL_KERN, CTL_EOL); 513 sysctl_createv(clog, 0, NULL, &node, 514 CTLFLAG_PERMANENT, 515 CTLTYPE_NODE, "sched", 516 SYSCTL_DESCR("Scheduler options"), 517 NULL, 0, NULL, 0, 518 CTL_KERN, CTL_CREATE, CTL_EOL); 519 520 if (node == NULL) 521 return; 522 523 rrticks = hz / 10; 524 525 sysctl_createv(NULL, 0, &node, NULL, 526 CTLFLAG_PERMANENT, 527 CTLTYPE_STRING, "name", NULL, 528 NULL, 0, __UNCONST("4.4BSD"), 0, 529 CTL_CREATE, CTL_EOL); 530 sysctl_createv(NULL, 0, &node, NULL, 531 CTLFLAG_PERMANENT, 532 CTLTYPE_INT, "rtts", 533 SYSCTL_DESCR("Round-robin time quantum (in miliseconds)"), 534 sysctl_sched_rtts, 0, NULL, 0, 535 CTL_CREATE, CTL_EOL); 536 } 537