1 /* 2 * Copyright (c) 1982, 1986, 1990 Regents of the University of California. 3 * All rights reserved. The Berkeley software License Agreement 4 * specifies the terms and conditions for redistribution. 5 * 6 * @(#)kern_synch.c 7.12 (Berkeley) 12/01/90 7 */ 8 9 #include "machine/pte.h" 10 #include "machine/psl.h" 11 #include "machine/mtpr.h" 12 13 #include "param.h" 14 #include "systm.h" 15 #include "user.h" 16 #include "proc.h" 17 #include "vm.h" 18 #include "kernel.h" 19 #include "buf.h" 20 21 /* 22 * Force switch among equal priority processes every 100ms. 23 */ 24 roundrobin() 25 { 26 27 runrun++; 28 aston(); 29 timeout(roundrobin, (caddr_t)0, hz / 10); 30 } 31 32 /* 33 * constants for digital decay and forget 34 * 90% of (p_cpu) usage in 5*loadav time 35 * 95% of (p_pctcpu) usage in 60 seconds (load insensitive) 36 * Note that, as ps(1) mentions, this can let percentages 37 * total over 100% (I've seen 137.9% for 3 processes). 38 * 39 * Note that hardclock updates p_cpu and p_cpticks independently. 40 * 41 * We wish to decay away 90% of p_cpu in (5 * loadavg) seconds. 42 * That is, the system wants to compute a value of decay such 43 * that the following for loop: 44 * for (i = 0; i < (5 * loadavg); i++) 45 * p_cpu *= decay; 46 * will compute 47 * p_cpu *= 0.1; 48 * for all values of loadavg: 49 * 50 * Mathematically this loop can be expressed by saying: 51 * decay ** (5 * loadavg) ~= .1 52 * 53 * The system computes decay as: 54 * decay = (2 * loadavg) / (2 * loadavg + 1) 55 * 56 * We wish to prove that the system's computation of decay 57 * will always fulfill the equation: 58 * decay ** (5 * loadavg) ~= .1 59 * 60 * If we compute b as: 61 * b = 2 * loadavg 62 * then 63 * decay = b / (b + 1) 64 * 65 * We now need to prove two things: 66 * 1) Given factor ** (5 * loadavg) ~= .1, prove factor == b/(b+1) 67 * 2) Given b/(b+1) ** power ~= .1, prove power == (5 * loadavg) 68 * 69 * Facts: 70 * For x close to zero, exp(x) =~ 1 + x, since 71 * exp(x) = 0! + x**1/1! + x**2/2! + ... . 72 * therefore exp(-1/b) =~ 1 - (1/b) = (b-1)/b. 73 * For x close to zero, ln(1+x) =~ x, since 74 * ln(1+x) = x - x**2/2 + x**3/3 - ... -1 < x < 1 75 * therefore ln(b/(b+1)) = ln(1 - 1/(b+1)) =~ -1/(b+1). 76 * ln(.1) =~ -2.30 77 * 78 * Proof of (1): 79 * Solve (factor)**(power) =~ .1 given power (5*loadav): 80 * solving for factor, 81 * ln(factor) =~ (-2.30/5*loadav), or 82 * factor =~ exp(-1/((5/2.30)*loadav) =~ exp(-1/(2*loadav)) = 83 * exp(-1/b) =~ (b-1)/b =~ b/(b+1). QED 84 * 85 * Proof of (2): 86 * Solve (factor)**(power) =~ .1 given factor == (b/(b+1)): 87 * solving for power, 88 * power*ln(b/(b+1)) =~ -2.30, or 89 * power =~ 2.3 * (b + 1) = 4.6*loadav + 2.3 =~ 5*loadav. QED 90 * 91 * Actual power values for the implemented algorithm are as follows: 92 * loadav: 1 2 3 4 93 * power: 5.68 10.32 14.94 19.55 94 */ 95 96 /* calculations for digital decay to forget 90% of usage in 5*loadav sec */ 97 #define get_b(loadav) (2 * (loadav)) 98 #define get_pcpu(b, cpu) (((b) * ((cpu) & 0377)) / ((b) + FSCALE)) 99 100 /* decay 95% of `p_pctcpu' in 60 seconds; see CCPU_SHIFT before changing */ 101 fixpt_t ccpu = 0.95122942450071400909 * FSCALE; /* exp(-1/20) */ 102 103 /* 104 * If `ccpu' is not equal to `exp(-1/20)' and you still want to use the 105 * faster/more-accurate formula, you'll have to estimate CCPU_SHIFT below 106 * and possibly adjust FSHIFT in "param.h" so that (FSHIFT >= CCPU_SHIFT). 107 * 108 * To estimate CCPU_SHIFT for exp(-1/20), the following formula was used: 109 * 1 - exp(-1/20) ~= 0.0487 ~= 0.0488 == 1 (fixed pt, *11* bits). 110 * 111 * If you dont want to bother with the faster/more-accurate formula, you 112 * can set CCPU_SHIFT to (FSHIFT + 1) which will use a slower/less-accurate 113 * (more general) method of calculating the %age of CPU used by a process. 114 */ 115 #define CCPU_SHIFT 11 116 117 /* 118 * Recompute process priorities, once a second 119 */ 120 schedcpu() 121 { 122 register fixpt_t b = get_b(averunnable[0]); 123 register struct proc *p; 124 register int s, a; 125 126 wakeup((caddr_t)&lbolt); 127 for (p = allproc; p != NULL; p = p->p_nxt) { 128 if (p->p_time != 127) 129 p->p_time++; 130 if (p->p_stat==SSLEEP || p->p_stat==SSTOP) 131 if (p->p_slptime != 127) 132 p->p_slptime++; 133 p->p_pctcpu = (p->p_pctcpu * ccpu) >> FSHIFT; 134 /* 135 * If the process has slept the entire second, 136 * stop recalculating its priority until it wakes up. 137 */ 138 if (p->p_slptime > 1) 139 continue; 140 /* 141 * p_pctcpu is only for ps. 142 */ 143 #if (FSHIFT >= CCPU_SHIFT) 144 p->p_pctcpu += (hz == 100)? 145 ((fixpt_t) p->p_cpticks) << (FSHIFT - CCPU_SHIFT): 146 100 * (((fixpt_t) p->p_cpticks) 147 << (FSHIFT - CCPU_SHIFT)) / hz; 148 #else 149 p->p_pctcpu += ((FSCALE - ccpu) * 150 (p->p_cpticks * FSCALE / hz)) >> FSHIFT; 151 #endif 152 p->p_cpticks = 0; 153 a = (int) get_pcpu(b, p->p_cpu) + p->p_nice; 154 if (a < 0) 155 a = 0; 156 if (a > 255) 157 a = 255; 158 p->p_cpu = a; 159 (void) setpri(p); 160 s = splhigh(); /* prevent state changes */ 161 if (p->p_pri >= PUSER) { 162 #define PPQ (128 / NQS) 163 if ((p != u.u_procp || noproc) && 164 p->p_stat == SRUN && 165 (p->p_flag & SLOAD) && 166 (p->p_pri / PPQ) != (p->p_usrpri / PPQ)) { 167 remrq(p); 168 p->p_pri = p->p_usrpri; 169 setrq(p); 170 } else 171 p->p_pri = p->p_usrpri; 172 } 173 splx(s); 174 } 175 vmmeter(); 176 if (runin!=0) { 177 runin = 0; 178 wakeup((caddr_t)&runin); 179 } 180 if (bclnlist != NULL) 181 wakeup((caddr_t)&proc[2]); 182 timeout(schedcpu, (caddr_t)0, hz); 183 } 184 185 /* 186 * Recalculate the priority of a process after it has slept for a while. 187 */ 188 updatepri(p) 189 register struct proc *p; 190 { 191 register int a = p->p_cpu & 0377; 192 register fixpt_t b = get_b(averunnable[0]); 193 194 p->p_slptime--; /* the first time was done in schedcpu */ 195 while (a && --p->p_slptime) 196 a = (int) get_pcpu(b, a) /* + p->p_nice */; 197 p->p_slptime = 0; 198 if (a < 0) 199 a = 0; 200 if (a > 255) 201 a = 255; 202 p->p_cpu = a; 203 (void) setpri(p); 204 } 205 206 #define SQSIZE 0100 /* Must be power of 2 */ 207 #define HASH(x) (( (int) x >> 5) & (SQSIZE-1)) 208 struct slpque { 209 struct proc *sq_head; 210 struct proc **sq_tailp; 211 } slpque[SQSIZE]; 212 213 /* 214 * During autoconfiguration or after a panic, a sleep will simply 215 * lower the priority briefly to allow interrupts, then return. 216 * The priority to be used (safepri) is machine-dependent, thus this 217 * value is initialized and maintained in the machine-dependent layers. 218 * This priority will typically be 0, or the lowest priority 219 * that is safe for use on the interrupt stack; it can be made 220 * higher to block network software interrupts after panics. 221 */ 222 int safepri; 223 224 /* 225 * General sleep call. 226 * Suspends current process until a wakeup is made on chan. 227 * The process will then be made runnable with priority pri. 228 * Sleeps at most timo/hz seconds (0 means no timeout). 229 * If pri includes PCATCH flag, signals are checked 230 * before and after sleeping, else signals are not checked. 231 * Returns 0 if awakened, EWOULDBLOCK if the timeout expires. 232 * If PCATCH is set and a signal needs to be delivered, 233 * ERESTART is returned if the current system call should be restarted 234 * if possible, and EINTR is returned if the system call should 235 * be interrupted by the signal (return EINTR). 236 */ 237 tsleep(chan, pri, wmesg, timo) 238 caddr_t chan; 239 int pri; 240 char *wmesg; 241 int timo; 242 { 243 register struct proc *rp; 244 register struct slpque *qp; 245 register s; 246 int sig, catch = pri & PCATCH; 247 extern int cold; 248 int endtsleep(); 249 250 rp = u.u_procp; 251 s = splhigh(); 252 if (cold || panicstr) { 253 /* 254 * After a panic, or during autoconfiguration, 255 * just give interrupts a chance, then just return; 256 * don't run any other procs or panic below, 257 * in case this is the idle process and already asleep. 258 */ 259 splx(safepri); 260 splx(s); 261 return (0); 262 } 263 #ifdef DIAGNOSTIC 264 if (chan == 0 || rp->p_stat != SRUN || rp->p_rlink) 265 panic("tsleep"); 266 #endif 267 rp->p_wchan = chan; 268 rp->p_wmesg = wmesg; 269 rp->p_slptime = 0; 270 rp->p_pri = pri & PRIMASK; 271 qp = &slpque[HASH(chan)]; 272 if (qp->sq_head == 0) 273 qp->sq_head = rp; 274 else 275 *qp->sq_tailp = rp; 276 *(qp->sq_tailp = &rp->p_link) = 0; 277 if (timo) 278 timeout(endtsleep, (caddr_t)rp, timo); 279 /* 280 * If we stop in CURSIG/issig(), a wakeup or a SIGCONT 281 * (or both) could occur while we were stopped. 282 * A SIGCONT would cause us to be marked as SSLEEP 283 * without resuming us, thus we must be ready for sleep 284 * when CURSIG is called. If the wakeup happens while we're 285 * stopped, rp->p_wchan will be 0 upon return from CURSIG. 286 */ 287 if (catch) { 288 rp->p_flag |= SSINTR; 289 if (sig = CURSIG(rp)) { 290 if (rp->p_wchan) 291 unsleep(rp); 292 rp->p_stat = SRUN; 293 goto resume; 294 } 295 if (rp->p_wchan == 0) { 296 catch = 0; 297 goto resume; 298 } 299 } 300 rp->p_stat = SSLEEP; 301 (void) spl0(); 302 u.u_ru.ru_nvcsw++; 303 swtch(); 304 resume: 305 curpri = rp->p_usrpri; 306 splx(s); 307 rp->p_flag &= ~SSINTR; 308 if (rp->p_flag & STIMO) { 309 rp->p_flag &= ~STIMO; 310 if (catch == 0 || sig == 0) 311 return (EWOULDBLOCK); 312 } else if (timo) 313 untimeout(endtsleep, (caddr_t)rp); 314 if (catch && (sig != 0 || (sig = CURSIG(rp)))) { 315 if (u.u_sigintr & sigmask(sig)) 316 return (EINTR); 317 return (ERESTART); 318 } 319 return (0); 320 } 321 322 /* 323 * Implement timeout for tsleep. 324 * If process hasn't been awakened (wchan non-zero), 325 * set timeout flag and undo the sleep. If proc 326 * is stopped, just unsleep so it will remain stopped. 327 */ 328 endtsleep(p) 329 register struct proc *p; 330 { 331 int s = splhigh(); 332 333 if (p->p_wchan) { 334 if (p->p_stat == SSLEEP) 335 setrun(p); 336 else 337 unsleep(p); 338 p->p_flag |= STIMO; 339 } 340 splx(s); 341 } 342 343 /* 344 * Short-term, non-interruptable sleep. 345 */ 346 sleep(chan, pri) 347 caddr_t chan; 348 int pri; 349 { 350 register struct proc *rp; 351 register struct slpque *qp; 352 register s; 353 extern int cold; 354 355 #ifdef DIAGNOSTIC 356 if (pri > PZERO) { 357 printf("sleep called with pri %d > PZERO, wchan: %x\n", 358 pri, chan); 359 panic("old sleep"); 360 } 361 #endif 362 rp = u.u_procp; 363 s = splhigh(); 364 if (cold || panicstr) { 365 /* 366 * After a panic, or during autoconfiguration, 367 * just give interrupts a chance, then just return; 368 * don't run any other procs or panic below, 369 * in case this is the idle process and already asleep. 370 */ 371 splx(safepri); 372 splx(s); 373 return; 374 } 375 #ifdef DIAGNOSTIC 376 if (chan==0 || rp->p_stat != SRUN || rp->p_rlink) 377 panic("sleep"); 378 #endif 379 rp->p_wchan = chan; 380 rp->p_wmesg = NULL; 381 rp->p_slptime = 0; 382 rp->p_pri = pri; 383 qp = &slpque[HASH(chan)]; 384 if (qp->sq_head == 0) 385 qp->sq_head = rp; 386 else 387 *qp->sq_tailp = rp; 388 *(qp->sq_tailp = &rp->p_link) = 0; 389 rp->p_stat = SSLEEP; 390 (void) spl0(); 391 u.u_ru.ru_nvcsw++; 392 swtch(); 393 curpri = rp->p_usrpri; 394 splx(s); 395 } 396 397 /* 398 * Remove a process from its wait queue 399 */ 400 unsleep(p) 401 register struct proc *p; 402 { 403 register struct slpque *qp; 404 register struct proc **hp; 405 int s; 406 407 s = splhigh(); 408 if (p->p_wchan) { 409 hp = &(qp = &slpque[HASH(p->p_wchan)])->sq_head; 410 while (*hp != p) 411 hp = &(*hp)->p_link; 412 *hp = p->p_link; 413 if (qp->sq_tailp == &p->p_link) 414 qp->sq_tailp = hp; 415 p->p_wchan = 0; 416 } 417 splx(s); 418 } 419 420 /* 421 * Wake up all processes sleeping on chan. 422 */ 423 wakeup(chan) 424 register caddr_t chan; 425 { 426 register struct slpque *qp; 427 register struct proc *p, **q; 428 int s; 429 430 s = splhigh(); 431 qp = &slpque[HASH(chan)]; 432 restart: 433 for (q = &qp->sq_head; p = *q; ) { 434 #ifdef DIAGNOSTIC 435 if (p->p_rlink || p->p_stat != SSLEEP && p->p_stat != SSTOP) 436 panic("wakeup"); 437 #endif 438 if (p->p_wchan==chan) { 439 p->p_wchan = 0; 440 *q = p->p_link; 441 if (qp->sq_tailp == &p->p_link) 442 qp->sq_tailp = q; 443 if (p->p_stat == SSLEEP) { 444 /* OPTIMIZED INLINE EXPANSION OF setrun(p) */ 445 if (p->p_slptime > 1) 446 updatepri(p); 447 p->p_stat = SRUN; 448 if (p->p_flag & SLOAD) 449 setrq(p); 450 /* 451 * Since curpri is a usrpri, 452 * p->p_pri is always better than curpri. 453 */ 454 runrun++; 455 aston(); 456 if ((p->p_flag&SLOAD) == 0) { 457 if (runout != 0) { 458 runout = 0; 459 wakeup((caddr_t)&runout); 460 } 461 wantin++; 462 } 463 /* END INLINE EXPANSION */ 464 goto restart; 465 } 466 } else 467 q = &p->p_link; 468 } 469 splx(s); 470 } 471 472 /* 473 * Initialize the (doubly-linked) run queues 474 * to be empty. 475 */ 476 rqinit() 477 { 478 register int i; 479 480 for (i = 0; i < NQS; i++) 481 qs[i].ph_link = qs[i].ph_rlink = (struct proc *)&qs[i]; 482 } 483 484 /* 485 * Set the process running; 486 * arrange for it to be swapped in if necessary. 487 */ 488 setrun(p) 489 register struct proc *p; 490 { 491 register int s; 492 493 s = splhigh(); 494 switch (p->p_stat) { 495 496 case 0: 497 case SWAIT: 498 case SRUN: 499 case SZOMB: 500 default: 501 panic("setrun"); 502 503 case SSTOP: 504 case SSLEEP: 505 unsleep(p); /* e.g. when sending signals */ 506 break; 507 508 case SIDL: 509 break; 510 } 511 p->p_stat = SRUN; 512 if (p->p_flag & SLOAD) 513 setrq(p); 514 splx(s); 515 if (p->p_slptime > 1) 516 updatepri(p); 517 if (p->p_pri < curpri) { 518 runrun++; 519 aston(); 520 } 521 if ((p->p_flag&SLOAD) == 0) { 522 if (runout != 0) { 523 runout = 0; 524 wakeup((caddr_t)&runout); 525 } 526 wantin++; 527 } 528 } 529 530 /* 531 * Set user priority. 532 * The rescheduling flag (runrun) 533 * is set if the priority is better 534 * than the currently running process. 535 */ 536 setpri(pp) 537 register struct proc *pp; 538 { 539 register int p; 540 541 p = (pp->p_cpu & 0377)/4; 542 p += PUSER + 2 * pp->p_nice; 543 if (pp->p_rssize > pp->p_maxrss && freemem < desfree) 544 p += 2*4; /* effectively, nice(4) */ 545 if (p > 127) 546 p = 127; 547 if (p < curpri) { 548 runrun++; 549 aston(); 550 } 551 pp->p_usrpri = p; 552 return (p); 553 } 554