1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731
|
/*
* Copyright (c) 1996, 1998, 1999 University of Utah and the Flux Group.
* All rights reserved.
*
* This file is part of the Flux OSKit. The OSKit is free software, also known
* as "open source;" you can redistribute it and/or modify it under the terms
* of the GNU General Public License (GPL), version 2, as published by the Free
* Software Foundation (FSF). To explore alternate licensing terms, contact
* the University of Utah at csl-dist@cs.utah.edu or +1-801-585-3271.
*
* The OSKit is distributed in the hope that it will be useful, but WITHOUT ANY
* WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS
* FOR A PARTICULAR PURPOSE. See the GPL for more details. You should have
* received a copy of the GPL along with the OSKit; see the file COPYING. If
* not, write to the FSF, 59 Temple Place #330, Boston, MA 02111-1307, USA.
*/
#ifdef CPU_INHERIT
/*
* Root scheduler for CPU inheritance scheduling system. Implements (or
* attempts to) the POSIX style fixed priority scheduler for SCHED_RR and
* SCHED_FIFO. Certain elements of this are a bit confused since its not
* always clear why the thread being scheduled stopped running, so adhering
* to the POSIX rules is difficult (translation: not done).
*/
#include <stdio.h>
#include <stdlib.h>
#include <errno.h>
#include <sys/types.h>
#include <oskit/threads/pthread.h>
#include <oskit/threads/cpuinherit.h>
#include <oskit/queue.h>
#include <malloc.h>
#include <assert.h>
#include "hash.h"
#ifdef MEASURE
#include <oskit/machine/proc_reg.h>
struct fixedpri_stats {
int msgs;
int msg_cycles;
};
static struct fixedpri_stats stats;
void dump_fixedpri_stats();
#undef CPUDEBUG
#define CPUDEBUG(cond, args...)
#endif
/*
* Basically the same scheduler as the builtin version.
*/
#define MAXPRI (PRIORITY_MAX + 1)
#define SCHED_RR_INTERVAL 1
/*
* The scheduler structures are per-scheduler instantiation.
*
* The RunQ is a multilevel queue of doubly linked lists. Use a bitmask
* to indicate whichrunq is non-empty, with the least significant bit
* being the highest priority (cause of ffs).
*
* TIDs are mapped back to the state structure using a key value.
*/
typedef struct fixedpri_sched {
queue_head_t runq[MAXPRI];
int runq_count;
oskit_u32_t runq_mask;
short ready;
short preemptible;
pthread_t maintid; /* *Is* the root scheduler */
pthread_t tid; /* Back pointer to thread */
hash_table_t *hashtable; /* Hash TID to pstate */
#ifdef DEBUG
struct fixedpri_state *threads[128];
#endif
} fixedpri_sched_t;
/*
* Per-thread state information structure.
*/
typedef struct fixedpri_state {
queue_chain_t runq; /* Queueing element */
pthread_t tid; /* TID of thread */
int policy; /* FIFO or RR */
int priority;
int ticks;
int base_priority;
#ifdef DEBUG
int lastcode;
#endif
} fixedpri_state_t;
#define NULL_FPSTATE ((fixedpri_state_t *) 0)
extern int ffs();
void *fixedpri_schedloop(void *arg);
static fixedpri_sched_t *allocate_fixedpri_scheduler(void);
static void fixedpri_canceled(void *arg);
static int fixedpri_pedantic = 1;
/*
* Map TID to pstate structure with a hash table.
*/
static void
fixedpri_setstate(fixedpri_sched_t *psched,
pthread_t tid, fixedpri_state_t *pstate)
{
if (tidhash_add(psched->hashtable, (void *) pstate, tid))
panic("fixedpri_setstate: "
"hashtable add failed: tid:%d pstate:0x%x",
tid, (int) pstate);
}
static fixedpri_state_t *
fixedpri_getstate(fixedpri_sched_t *psched, pthread_t tid)
{
fixedpri_state_t *pstate;
if ((pstate = (fixedpri_state_t *)
tidhash_lookup(psched->hashtable, tid)) == NULL)
panic("fixedpri_getstate: "
"hashtable lookup failed: tid:%d", tid);
return pstate;
}
static void
fixedpri_remstate(fixedpri_sched_t *psched, pthread_t tid)
{
tidhash_rem(psched->hashtable, tid);
}
/*
* Special hook to bootstrap the root scheduler. Return the entrypoint
* of the scheduler, and the data structure to pass as the first argument.
* The caller will create a new thread with those values and run it.
* The caller passes in the TID of the main thread. Create the state
* structure for it as well.
*/
void
bootstrap_root_scheduler(pthread_t tid, int preemptible,
void *(**function)(void *), void **argument)
{
fixedpri_state_t *pstate;
fixedpri_sched_t *psched;
/*
* Allocate the scheduler data structure.
*/
psched = allocate_fixedpri_scheduler();
psched->preemptible = preemptible;
psched->maintid = tid;
/*
* Allocate the per-thread data structure for the main thread.
*/
if ((pstate =
(fixedpri_state_t *) calloc(1, sizeof(fixedpri_state_t))) == NULL)
panic("fixedpri_thread_init: No more memory");
pstate->tid = tid;
pstate->priority = PRIORITY_NORMAL;
pstate->base_priority = PRIORITY_NORMAL;
pstate->policy = SCHED_RR;
pstate->ticks = SCHED_RR_INTERVAL;
fixedpri_setstate(psched, tid, pstate);
#ifdef DEBUG
psched->threads[(int) tid] = pstate;
#endif
*argument = psched;
*function = fixedpri_schedloop;
}
/*
* Create a fixedpri scheduler. This creates the thread and makes sure
* it gets run so that it exists and is ready to handle scheduling
* messages.
*/
int
create_fixedpri_scheduler(pthread_t *tid,
const pthread_attr_t *attr, int preemptible)
{
fixedpri_sched_t *psched;
psched = allocate_fixedpri_scheduler();
psched->preemptible = preemptible;
pthread_create(tid, attr, fixedpri_schedloop, (void *) psched);
/*
* The scheduler has to run.
*/
while (! psched->ready)
oskit_pthread_sleep(1);
/*
* Back in this thread. Just return.
*/
return 0;
}
/*
* Allocate and initialize a per-instantiation root scheduler runq.
*/
static fixedpri_sched_t *
allocate_fixedpri_scheduler(void)
{
fixedpri_sched_t *psched;
int i;
if ((psched = (fixedpri_sched_t *)
calloc(1, sizeof(fixedpri_sched_t))) == NULL)
panic("allocate_fixedpri_scheduler: No more memory");
if (tidhash_create(&psched->hashtable, 0))
panic("allocate_fixedpri_scheduler: Hash Table allocation");
for (i = 0; i < MAXPRI; i++)
queue_init(&(psched->runq[i]));
return psched;
}
/*
* Initialize the per-thread scheduler state structure. This is called
* when a new thread is created. The state structure is then stored
* in the key table so that TIDs can be mapped back to the state structure.
*/
static fixedpri_state_t *
fixedpri_thread_init(fixedpri_sched_t *psched,
pthread_t tid, int priority, int policy)
{
fixedpri_state_t *pstate;
if ((pstate =
(fixedpri_state_t *) calloc(1, sizeof(fixedpri_state_t))) == NULL)
panic("fixedpri_thread_init: No more memory");
pstate->tid = tid;
pstate->priority = priority;
pstate->base_priority = priority;
pstate->policy = policy;
pstate->ticks = SCHED_RR_INTERVAL;
fixedpri_setstate(psched, tid, pstate);
#ifdef DEBUG
psched->threads[(int) tid] = pstate;
#endif
return pstate;
}
/*
* Are there any threads on the runq?
*/
static inline int
runq_empty(fixedpri_sched_t *psched)
{
return (psched->runq_mask == 0);
}
/*
* Get the highest priority scheduled thread.
*/
static inline int
runq_maxprio(fixedpri_sched_t *psched)
{
int prio;
if (runq_empty(psched))
return -1;
else {
prio = ffs(psched->runq_mask);
return PRIORITY_MAX - (prio - 1);
}
}
/*
* Determine if a thread is on the runq. Use a separate field
* since using the flags would require locking the thread. Use the
* queue chain pointer instead, setting it to zero when a thread is
* removed from the queue.
*/
static inline int
runq_onrunq(fixedpri_sched_t *psched, fixedpri_state_t *pstate)
{
return (int) pstate->runq.next;
}
/*
* Add and remove threads from the runq. The runq lock should be locked,
* and interrupts disabled.
*/
/*
* Insert at the tail of the runq.
*/
static inline void
runq_insert_tail(fixedpri_sched_t *psched, fixedpri_state_t *pstate)
{
int prio = PRIORITY_MAX - pstate->priority;
queue_head_t *phdr = &(psched->runq[prio]);
queue_enter(phdr, pstate, fixedpri_state_t *, runq);
psched->runq_mask |= (1 << prio);
psched->runq_count += 1;
}
/*
* Insert at the head of the runq.
*/
static inline void
runq_insert_head(fixedpri_sched_t *psched, fixedpri_state_t *pstate)
{
int prio = PRIORITY_MAX - pstate->priority;
queue_head_t *phdr = &(psched->runq[prio]);
queue_enter_first(phdr, pstate, fixedpri_state_t *, runq);
psched->runq_mask |= (1 << prio);
psched->runq_count += 1;
}
/*
* Dequeue highest priority pthread.
*/
static fixedpri_state_t *
runq_dequeue(fixedpri_sched_t *psched)
{
int prio = ffs(psched->runq_mask) - 1;
queue_head_t *phdr = &(psched->runq[prio]);
fixedpri_state_t *pnext;
queue_remove_first(phdr, pnext, fixedpri_state_t *, runq);
pnext->runq.next = (queue_entry_t) 0;
psched->runq_count--;
if (queue_empty(phdr))
psched->runq_mask &= ~(1 << prio);
return pnext;
}
/*
* Debug the runq.
*/
static void
runq_check(fixedpri_sched_t *psched)
{
int count, i;
fixedpri_state_t *pstate;
count = 0;
for (i = 0; i < MAXPRI; i++) {
if (! queue_empty(&(psched->runq[i]))) {
queue_iterate(&(psched->runq[i]),
pstate, fixedpri_state_t *, runq) {
count++;
}
}
}
if (count != psched->runq_count)
panic("Bad runq(%d): 0x%x\n",
pthread_self(), psched);
}
/*
* Remove an arbitrary thread from the runq.
*/
static inline void
runq_remove(fixedpri_sched_t *psched, fixedpri_state_t *pstate)
{
int prio = PRIORITY_MAX - pstate->priority;
queue_head_t *phdr = &(psched->runq[prio]);
queue_remove(phdr, pstate, fixedpri_state_t *, runq);
pstate->runq.next = (queue_entry_t) 0;
psched->runq_count--;
if (queue_empty(phdr))
psched->runq_mask &= ~(1 << prio);
}
/*
* Debug
*/
static void
runq_debug(fixedpri_sched_t *psched)
{
int i;
fixedpri_state_t *pstate;
printf("fixedpri(%d): Count:%d Mask:0x%x\n", (int) psched->tid,
psched->runq_count, psched->runq_mask);
for (i = 0; i < MAXPRI; i++) {
if (! queue_empty(&(psched->runq[i]))) {
queue_iterate(&(psched->runq[i]),
pstate, fixedpri_state_t *, runq) {
printf("0x%x(%d) Pri %d Pol %d T %d BP %d\n",
(int) pstate, (int) pstate->tid,
pstate->priority, pstate->policy,
pstate->ticks, pstate->base_priority);
}
}
}
}
/*
* This is the scheduler loop.
*/
void *
fixedpri_schedloop(void *arg)
{
fixedpri_sched_t *psched = (fixedpri_sched_t *) arg;
schedmsg_t msg;
fixedpri_state_t *current_thread = 0, *pstate = 0;
sched_wakecond_t wakeup_cond = 0;
int pedantic, rc;
oskit_s32_t timeout;
pthread_t tid;
#ifdef MEASURE
unsigned long before, after;
atexit(dump_fixedpri_stats);
#endif
psched->tid = pthread_self();
/*
* Must tell the main scheduling code ...
*/
pthread_sched_become_scheduler();
/*
* Cancelation cleanup handler to cleanup resources at exit.
*/
pthread_cleanup_push(fixedpri_canceled, (void *) psched);
/*
* Preemption means donate with non-zero timeout.
*/
if (psched->preemptible)
timeout = PTHREAD_TICK;
else
timeout = 0;
/*
* Root scheduler! Schedule the mainthread so it gets run.
*/
if (psched->maintid) {
pstate = fixedpri_getstate(psched, psched->maintid);
runq_insert_tail(psched, pstate);
pedantic = 1;
}
else
pedantic = fixedpri_pedantic;
psched->ready = 1;
CPUDEBUG(TRUE,
"fixedpri_schedloop(%d): ready - psched:0x%x\n",
psched->tid, (int) psched);
while (1) {
pthread_testcancel();
CPUDEBUG(FIXEDPRI,
"fixedpri_schedloop(%d): qcount(%d) qmask(%x)\n",
psched->tid,
psched->runq_count, (int) psched->runq_mask);
/*
* Consume any pending messages until there are no more.
*/
if (! pthread_sched_message_recv(&msg, 0))
goto consume;
if (CPUDEBUG_ISSET(FIXEDPRI))
runq_check(psched);
/*
* Find a thread to run.
*/
if (runq_empty(psched))
current_thread = NULL_FPSTATE;
else
current_thread = runq_dequeue(psched);
/*
* If we found a thread, switch into it and wait for
* a message. Otherwise, wait for messages to arrive
* that indicate something has changed.
*/
if (current_thread) {
CPUDEBUG(FIXEDPRI,
"fixedpri_schedloop(%d): pstate 0x%x(%d)\n",
psched->tid, (int) current_thread,
current_thread->tid);
if (pedantic || !runq_empty(psched))
wakeup_cond = WAKEUP_ON_BLOCK;
else
wakeup_cond = WAKEUP_ON_SWITCH;
/*
* Donate and check the return condition.
*/
rc =
pthread_sched_donate_wait_recv(current_thread->tid,
wakeup_cond, &msg, timeout);
CPUDEBUG(FIXEDPRI,
"fixedpri_schedloop(%d): Donated: %d %s\n",
psched->tid, current_thread->tid,
msg_sched_rcodes[rc & ~SCHEDULE_MSGRECV]);
#ifdef DEBUG
current_thread->lastcode = rc;
#endif
switch (rc & ~SCHEDULE_MSGRECV) {
case SCHEDULE_NOTREADY:
/*
* Thread was not ready to recv donation.
* Forget about this thread.
*/
break;
case SCHEDULE_BLOCKED:
/* Thread blocked, so forget about it. */
break;
case SCHEDULE_YIELDED:
/* user specified yield */
assert(! runq_onrunq(psched, current_thread));
runq_insert_tail(psched, current_thread);
break;
case SCHEDULE_PREEMPTED:
/* system directed yield */
assert(! runq_onrunq(psched, current_thread));
runq_insert_head(psched, current_thread);
break;
case SCHEDULE_TIMEDOUT:
/* Time based preemption */
pstate = current_thread;
if (pstate->policy == SCHED_RR) {
if (--pstate->ticks == 0) {
runq_insert_tail(psched, pstate);
pstate->ticks = SCHED_RR_INTERVAL;
}
else
runq_insert_head(psched, pstate);
}
else if (pstate->policy == SCHED_FIFO) {
runq_insert_head(psched, pstate);
}
break;
default:
if (rc == SCHEDULE_MSGRECV)
runq_insert_head(psched,
current_thread);
else
panic("fixedpri_schedloop: "
"Bad return code:%d", rc);
break;
}
/* Back to the top to try again. */
if (! (rc & SCHEDULE_MSGRECV))
continue;
}
else {
/*
* No threads to run so block waiting for a message.
*/
CPUDEBUG(FIXEDPRI,
"fixedpri_schedloop(%d): Recv\n",psched->tid);
rc = pthread_sched_message_recv(&msg, -1);
if (rc == OSKIT_ECANCELED)
pthread_exit((void *) PTHREAD_CANCELED);
}
/*
* Process messages.
*/
consume:
/* runq_check(psched); */
#ifdef MEASURE
before = get_tsc();
#endif
/*
* Map tid in message to thread state structure. Avoid
* lookup if possible.
*/
assert(msg.tid);
if (msg.type != MSG_SCHED_NEWTHREAD) {
if (current_thread && msg.tid == current_thread->tid)
pstate = current_thread;
else
pstate = fixedpri_getstate(psched, msg.tid);
}
CPUDEBUG(FIXEDPRI,
"fixedpri_schedloop(%d): MSG: %s %d 0x%x(%d)\n",
psched->tid,
msg_sched_typenames[msg.type], msg.opaque,
(int) pstate, (pstate ? pstate->tid : 0));
switch (msg.type) {
case MSG_SCHED_NEWTHREAD:
/*
* New thread has joined us. Create a state structure
* and add it to the runq.
*/
pstate = fixedpri_thread_init(psched, msg.tid,
msg.opaque, msg.opaque2);
/* and add it to the runq */
runq_insert_tail(psched, pstate);
break;
case MSG_SCHED_UNBLOCK:
/*
* A thread wants to be rescheduled. Place at the
* tail of the queue as defined by POSIX.
*/
if (! runq_onrunq(psched, pstate))
runq_insert_tail(psched, pstate);
break;
case MSG_SCHED_SETSTATE:
/*
* Thread priority has been changed.
*/
pstate->base_priority = (int) msg.opaque;
if (runq_onrunq(psched, pstate)) {
runq_remove(psched, pstate);
pstate->priority = (int) msg.opaque;
runq_insert_tail(psched, pstate);
}
else
pstate->priority = (int) msg.opaque;
break;
case MSG_SCHED_EXITED:
/*
* The thread has exited.
*/
CPUDEBUG(FIXEDPRI,
"fixedpri_schedloop(%d): "
"Exit: Thread 0x%x(%d)\n",
pthread_self(), (int) pstate, pstate->tid);
tid = pstate->tid;
fixedpri_remstate(psched, tid);
free(pstate);
/*
* Special case is the main thread in the root
* scheduler. If that exits, the scheduler exits.
*/
if (psched->maintid && tid == psched->maintid)
pthread_exit(0);
break;
default:
panic("fixedpri_schedloop: Bad message: %d 0x%x\n",
msg.type, pstate);
break;
}
#ifdef MEASURE
after = get_tsc();
if (after > before) {
stats.msgs++;
stats.msg_cycles += (after - before);
}
#endif
/*runq_check(psched);*/
}
/*
* Never reached.
*/
pthread_cleanup_pop(1);
}
/*
* Handle async cancel of the scheduler. Cleanup resources before the thread
* disappears completely.
*
* XXX NOT CLEANING UP THREADS! What does it mean anyway to cancel a
* scheduler before all its threads have been canceled? Also, there is an
* obvious race condition that needs to be addressed; what happens if a
* scheduler is canceled before its sees a particular thread for the first
* time (and thus does not even know it exists as its child).
*/
void
fixedpri_canceled(void *arg)
{
fixedpri_sched_t *psched = (fixedpri_sched_t *) arg;
CPUDEBUG(FIXEDPRI,
"fixedpri_terminate: Scheduler exiting:%d\n",
pthread_self());
free(psched);
}
#ifdef MEASURE
void
dump_fixedpri_stats()
{
printf("FP msgs: %d\n", stats.msgs);
printf("FP msg Cycles: %u\n", stats.msg_cycles);
}
#endif
#endif CPU_INHERIT
|