1 /*-
2 * SPDX-License-Identifier: BSD-2-Clause
3 *
4 * Copyright (C) 2001 Julian Elischer <julian@freebsd.org>.
5 * All rights reserved.
6 *
7 * Redistribution and use in source and binary forms, with or without
8 * modification, are permitted provided that the following conditions
9 * are met:
10 * 1. Redistributions of source code must retain the above copyright
11 * notice(s), this list of conditions and the following disclaimer as
12 * the first lines of this file unmodified other than the possible
13 * addition of one or more copyright notices.
14 * 2. Redistributions in binary form must reproduce the above copyright
15 * notice(s), this list of conditions and the following disclaimer in the
16 * documentation and/or other materials provided with the distribution.
17 *
18 * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY
19 * EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
20 * WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
21 * DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY
22 * DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
23 * (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
24 * SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
25 * CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
26 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
27 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH
28 * DAMAGE.
29 */
30
31 #include "opt_witness.h"
32 #include "opt_hwpmc_hooks.h"
33
34 #include <sys/systm.h>
35 #include <sys/asan.h>
36 #include <sys/kernel.h>
37 #include <sys/lock.h>
38 #include <sys/msan.h>
39 #include <sys/mutex.h>
40 #include <sys/proc.h>
41 #include <sys/bitstring.h>
42 #include <sys/epoch.h>
43 #include <sys/rangelock.h>
44 #include <sys/resourcevar.h>
45 #include <sys/sdt.h>
46 #include <sys/smp.h>
47 #include <sys/sched.h>
48 #include <sys/sleepqueue.h>
49 #include <sys/selinfo.h>
50 #include <sys/syscallsubr.h>
51 #include <sys/dtrace_bsd.h>
52 #include <sys/sysent.h>
53 #include <sys/turnstile.h>
54 #include <sys/taskqueue.h>
55 #include <sys/ktr.h>
56 #include <sys/rwlock.h>
57 #include <sys/umtxvar.h>
58 #include <sys/vmmeter.h>
59 #include <sys/cpuset.h>
60 #ifdef HWPMC_HOOKS
61 #include <sys/pmckern.h>
62 #endif
63 #include <sys/priv.h>
64
65 #include <security/audit/audit.h>
66
67 #include <vm/pmap.h>
68 #include <vm/vm.h>
69 #include <vm/vm_extern.h>
70 #include <vm/uma.h>
71 #include <vm/vm_phys.h>
72 #include <sys/eventhandler.h>
73
74 /*
75 * Asserts below verify the stability of struct thread and struct proc
76 * layout, as exposed by KBI to modules. On head, the KBI is allowed
77 * to drift, change to the structures must be accompanied by the
78 * assert update.
79 *
80 * On the stable branches after KBI freeze, conditions must not be
81 * violated. Typically new fields are moved to the end of the
82 * structures.
83 */
84 #ifdef __amd64__
85 _Static_assert(offsetof(struct thread, td_flags) == 0x108,
86 "struct thread KBI td_flags");
87 _Static_assert(offsetof(struct thread, td_pflags) == 0x114,
88 "struct thread KBI td_pflags");
89 _Static_assert(offsetof(struct thread, td_frame) == 0x4b8,
90 "struct thread KBI td_frame");
91 _Static_assert(offsetof(struct thread, td_emuldata) == 0x6c0,
92 "struct thread KBI td_emuldata");
93 _Static_assert(offsetof(struct proc, p_flag) == 0xb8,
94 "struct proc KBI p_flag");
95 _Static_assert(offsetof(struct proc, p_pid) == 0xc4,
96 "struct proc KBI p_pid");
97 _Static_assert(offsetof(struct proc, p_filemon) == 0x3c8,
98 "struct proc KBI p_filemon");
99 _Static_assert(offsetof(struct proc, p_comm) == 0x3e0,
100 "struct proc KBI p_comm");
101 _Static_assert(offsetof(struct proc, p_emuldata) == 0x4d0,
102 "struct proc KBI p_emuldata");
103 #endif
104 #ifdef __i386__
105 _Static_assert(offsetof(struct thread, td_flags) == 0x9c,
106 "struct thread KBI td_flags");
107 _Static_assert(offsetof(struct thread, td_pflags) == 0xa8,
108 "struct thread KBI td_pflags");
109 _Static_assert(offsetof(struct thread, td_frame) == 0x318,
110 "struct thread KBI td_frame");
111 _Static_assert(offsetof(struct thread, td_emuldata) == 0x35c,
112 "struct thread KBI td_emuldata");
113 _Static_assert(offsetof(struct proc, p_flag) == 0x6c,
114 "struct proc KBI p_flag");
115 _Static_assert(offsetof(struct proc, p_pid) == 0x78,
116 "struct proc KBI p_pid");
117 _Static_assert(offsetof(struct proc, p_filemon) == 0x270,
118 "struct proc KBI p_filemon");
119 _Static_assert(offsetof(struct proc, p_comm) == 0x284,
120 "struct proc KBI p_comm");
121 _Static_assert(offsetof(struct proc, p_emuldata) == 0x318,
122 "struct proc KBI p_emuldata");
123 #endif
124
125 SDT_PROVIDER_DECLARE(proc);
126 SDT_PROBE_DEFINE(proc, , , lwp__exit);
127
128 /*
129 * thread related storage.
130 */
131 static uma_zone_t thread_zone;
132
133 struct thread_domain_data {
134 struct thread *tdd_zombies;
135 int tdd_reapticks;
136 } __aligned(CACHE_LINE_SIZE);
137
138 static struct thread_domain_data thread_domain_data[MAXMEMDOM];
139
140 static struct task thread_reap_task;
141 static struct callout thread_reap_callout;
142
143 static void thread_zombie(struct thread *);
144 static void thread_reap(void);
145 static void thread_reap_all(void);
146 static void thread_reap_task_cb(void *, int);
147 static void thread_reap_callout_cb(void *);
148 static void thread_unsuspend_one(struct thread *td, struct proc *p,
149 bool boundary);
150 static void thread_free_batched(struct thread *td);
151
152 static __exclusive_cache_line struct mtx tid_lock;
153 static bitstr_t *tid_bitmap;
154
155 static MALLOC_DEFINE(M_TIDHASH, "tidhash", "thread hash");
156
157 static int maxthread;
158 SYSCTL_INT(_kern, OID_AUTO, maxthread, CTLFLAG_RDTUN,
159 &maxthread, 0, "Maximum number of threads");
160
161 static __exclusive_cache_line int nthreads;
162
163 static LIST_HEAD(tidhashhead, thread) *tidhashtbl;
164 static u_long tidhash;
165 static u_long tidhashlock;
166 static struct rwlock *tidhashtbl_lock;
167 #define TIDHASH(tid) (&tidhashtbl[(tid) & tidhash])
168 #define TIDHASHLOCK(tid) (&tidhashtbl_lock[(tid) & tidhashlock])
169
170 EVENTHANDLER_LIST_DEFINE(thread_ctor);
171 EVENTHANDLER_LIST_DEFINE(thread_dtor);
172 EVENTHANDLER_LIST_DEFINE(thread_init);
173 EVENTHANDLER_LIST_DEFINE(thread_fini);
174
175 static bool
thread_count_inc_try(void)176 thread_count_inc_try(void)
177 {
178 int nthreads_new;
179
180 nthreads_new = atomic_fetchadd_int(&nthreads, 1) + 1;
181 if (nthreads_new >= maxthread - 100) {
182 if (priv_check_cred(curthread->td_ucred, PRIV_MAXPROC) != 0 ||
183 nthreads_new >= maxthread) {
184 atomic_subtract_int(&nthreads, 1);
185 return (false);
186 }
187 }
188 return (true);
189 }
190
191 static bool
thread_count_inc(void)192 thread_count_inc(void)
193 {
194 static struct timeval lastfail;
195 static int curfail;
196
197 thread_reap();
198 if (thread_count_inc_try()) {
199 return (true);
200 }
201
202 thread_reap_all();
203 if (thread_count_inc_try()) {
204 return (true);
205 }
206
207 if (ppsratecheck(&lastfail, &curfail, 1)) {
208 printf("maxthread limit exceeded by uid %u "
209 "(pid %d); consider increasing kern.maxthread\n",
210 curthread->td_ucred->cr_ruid, curproc->p_pid);
211 }
212 return (false);
213 }
214
215 static void
thread_count_sub(int n)216 thread_count_sub(int n)
217 {
218
219 atomic_subtract_int(&nthreads, n);
220 }
221
222 static void
thread_count_dec(void)223 thread_count_dec(void)
224 {
225
226 thread_count_sub(1);
227 }
228
229 static lwpid_t
tid_alloc(void)230 tid_alloc(void)
231 {
232 static lwpid_t trytid;
233 lwpid_t tid;
234
235 mtx_lock(&tid_lock);
236 /*
237 * It is an invariant that the bitmap is big enough to hold maxthread
238 * IDs. If we got to this point there has to be at least one free.
239 */
240 if (trytid >= maxthread)
241 trytid = 0;
242 bit_ffc_at(tid_bitmap, trytid, maxthread, &tid);
243 if (tid == -1) {
244 KASSERT(trytid != 0, ("unexpectedly ran out of IDs"));
245 trytid = 0;
246 bit_ffc_at(tid_bitmap, trytid, maxthread, &tid);
247 KASSERT(tid != -1, ("unexpectedly ran out of IDs"));
248 }
249 bit_set(tid_bitmap, tid);
250 trytid = tid + 1;
251 mtx_unlock(&tid_lock);
252 return (tid + NO_PID);
253 }
254
255 static void
tid_free_locked(lwpid_t rtid)256 tid_free_locked(lwpid_t rtid)
257 {
258 lwpid_t tid;
259
260 mtx_assert(&tid_lock, MA_OWNED);
261 KASSERT(rtid >= NO_PID,
262 ("%s: invalid tid %d\n", __func__, rtid));
263 tid = rtid - NO_PID;
264 KASSERT(bit_test(tid_bitmap, tid) != 0,
265 ("thread ID %d not allocated\n", rtid));
266 bit_clear(tid_bitmap, tid);
267 }
268
269 static void
tid_free(lwpid_t rtid)270 tid_free(lwpid_t rtid)
271 {
272
273 mtx_lock(&tid_lock);
274 tid_free_locked(rtid);
275 mtx_unlock(&tid_lock);
276 }
277
278 static void
tid_free_batch(lwpid_t * batch,int n)279 tid_free_batch(lwpid_t *batch, int n)
280 {
281 int i;
282
283 mtx_lock(&tid_lock);
284 for (i = 0; i < n; i++) {
285 tid_free_locked(batch[i]);
286 }
287 mtx_unlock(&tid_lock);
288 }
289
290 /*
291 * Batching for thread reapping.
292 */
293 struct tidbatch {
294 lwpid_t tab[16];
295 int n;
296 };
297
298 static void
tidbatch_prep(struct tidbatch * tb)299 tidbatch_prep(struct tidbatch *tb)
300 {
301
302 tb->n = 0;
303 }
304
305 static void
tidbatch_add(struct tidbatch * tb,struct thread * td)306 tidbatch_add(struct tidbatch *tb, struct thread *td)
307 {
308
309 KASSERT(tb->n < nitems(tb->tab),
310 ("%s: count too high %d", __func__, tb->n));
311 tb->tab[tb->n] = td->td_tid;
312 tb->n++;
313 }
314
315 static void
tidbatch_process(struct tidbatch * tb)316 tidbatch_process(struct tidbatch *tb)
317 {
318
319 KASSERT(tb->n <= nitems(tb->tab),
320 ("%s: count too high %d", __func__, tb->n));
321 if (tb->n == nitems(tb->tab)) {
322 tid_free_batch(tb->tab, tb->n);
323 tb->n = 0;
324 }
325 }
326
327 static void
tidbatch_final(struct tidbatch * tb)328 tidbatch_final(struct tidbatch *tb)
329 {
330
331 KASSERT(tb->n <= nitems(tb->tab),
332 ("%s: count too high %d", __func__, tb->n));
333 if (tb->n != 0) {
334 tid_free_batch(tb->tab, tb->n);
335 }
336 }
337
338 /*
339 * Batching thread count free, for consistency
340 */
341 struct tdcountbatch {
342 int n;
343 };
344
345 static void
tdcountbatch_prep(struct tdcountbatch * tb)346 tdcountbatch_prep(struct tdcountbatch *tb)
347 {
348
349 tb->n = 0;
350 }
351
352 static void
tdcountbatch_add(struct tdcountbatch * tb,struct thread * td __unused)353 tdcountbatch_add(struct tdcountbatch *tb, struct thread *td __unused)
354 {
355
356 tb->n++;
357 }
358
359 static void
tdcountbatch_process(struct tdcountbatch * tb)360 tdcountbatch_process(struct tdcountbatch *tb)
361 {
362
363 if (tb->n == 32) {
364 thread_count_sub(tb->n);
365 tb->n = 0;
366 }
367 }
368
369 static void
tdcountbatch_final(struct tdcountbatch * tb)370 tdcountbatch_final(struct tdcountbatch *tb)
371 {
372
373 if (tb->n != 0) {
374 thread_count_sub(tb->n);
375 }
376 }
377
378 /*
379 * Prepare a thread for use.
380 */
381 static int
thread_ctor(void * mem,int size,void * arg,int flags)382 thread_ctor(void *mem, int size, void *arg, int flags)
383 {
384 struct thread *td;
385
386 td = (struct thread *)mem;
387 TD_SET_STATE(td, TDS_INACTIVE);
388 td->td_lastcpu = td->td_oncpu = NOCPU;
389
390 /*
391 * Note that td_critnest begins life as 1 because the thread is not
392 * running and is thereby implicitly waiting to be on the receiving
393 * end of a context switch.
394 */
395 td->td_critnest = 1;
396 td->td_lend_user_pri = PRI_MAX;
397 #ifdef AUDIT
398 audit_thread_alloc(td);
399 #endif
400 #ifdef KDTRACE_HOOKS
401 kdtrace_thread_ctor(td);
402 #endif
403 umtx_thread_alloc(td);
404 MPASS(td->td_sel == NULL);
405 return (0);
406 }
407
408 /*
409 * Reclaim a thread after use.
410 */
411 static void
thread_dtor(void * mem,int size,void * arg)412 thread_dtor(void *mem, int size, void *arg)
413 {
414 struct thread *td;
415
416 td = (struct thread *)mem;
417
418 #ifdef INVARIANTS
419 /* Verify that this thread is in a safe state to free. */
420 switch (TD_GET_STATE(td)) {
421 case TDS_INHIBITED:
422 case TDS_RUNNING:
423 case TDS_CAN_RUN:
424 case TDS_RUNQ:
425 /*
426 * We must never unlink a thread that is in one of
427 * these states, because it is currently active.
428 */
429 panic("bad state for thread unlinking");
430 /* NOTREACHED */
431 case TDS_INACTIVE:
432 break;
433 default:
434 panic("bad thread state");
435 /* NOTREACHED */
436 }
437 #endif
438 #ifdef AUDIT
439 audit_thread_free(td);
440 #endif
441 #ifdef KDTRACE_HOOKS
442 kdtrace_thread_dtor(td);
443 #endif
444 /* Free all OSD associated to this thread. */
445 osd_thread_exit(td);
446 ast_kclear(td);
447 seltdfini(td);
448 }
449
450 /*
451 * Initialize type-stable parts of a thread (when newly created).
452 */
453 static int
thread_init(void * mem,int size,int flags)454 thread_init(void *mem, int size, int flags)
455 {
456 struct thread *td;
457
458 td = (struct thread *)mem;
459
460 td->td_allocdomain = vm_phys_domain(vtophys(td));
461 td->td_sleepqueue = sleepq_alloc();
462 td->td_turnstile = turnstile_alloc();
463 EVENTHANDLER_DIRECT_INVOKE(thread_init, td);
464 umtx_thread_init(td);
465 td->td_kstack = 0;
466 td->td_sel = NULL;
467 return (0);
468 }
469
470 /*
471 * Tear down type-stable parts of a thread (just before being discarded).
472 */
473 static void
thread_fini(void * mem,int size)474 thread_fini(void *mem, int size)
475 {
476 struct thread *td;
477
478 td = (struct thread *)mem;
479 EVENTHANDLER_DIRECT_INVOKE(thread_fini, td);
480 turnstile_free(td->td_turnstile);
481 sleepq_free(td->td_sleepqueue);
482 umtx_thread_fini(td);
483 MPASS(td->td_sel == NULL);
484 }
485
486 /*
487 * For a newly created process,
488 * link up all the structures and its initial threads etc.
489 * called from:
490 * {arch}/{arch}/machdep.c {arch}_init(), init386() etc.
491 * proc_dtor() (should go away)
492 * proc_init()
493 */
494 void
proc_linkup0(struct proc * p,struct thread * td)495 proc_linkup0(struct proc *p, struct thread *td)
496 {
497 TAILQ_INIT(&p->p_threads); /* all threads in proc */
498 proc_linkup(p, td);
499 }
500
501 void
proc_linkup(struct proc * p,struct thread * td)502 proc_linkup(struct proc *p, struct thread *td)
503 {
504
505 sigqueue_init(&p->p_sigqueue, p);
506 p->p_ksi = ksiginfo_alloc(M_WAITOK);
507 if (p->p_ksi != NULL) {
508 /* XXX p_ksi may be null if ksiginfo zone is not ready */
509 p->p_ksi->ksi_flags = KSI_EXT | KSI_INS;
510 }
511 LIST_INIT(&p->p_mqnotifier);
512 p->p_numthreads = 0;
513 thread_link(td, p);
514 }
515
516 static void
ast_suspend(struct thread * td,int tda __unused)517 ast_suspend(struct thread *td, int tda __unused)
518 {
519 struct proc *p;
520
521 p = td->td_proc;
522 /*
523 * We need to check to see if we have to exit or wait due to a
524 * single threading requirement or some other STOP condition.
525 */
526 PROC_LOCK(p);
527 thread_suspend_check(0);
528 PROC_UNLOCK(p);
529 }
530
531 extern int max_threads_per_proc;
532
533 /*
534 * Initialize global thread allocation resources.
535 */
536 void
threadinit(void)537 threadinit(void)
538 {
539 u_long i;
540 lwpid_t tid0;
541
542 /*
543 * Place an upper limit on threads which can be allocated.
544 *
545 * Note that other factors may make the de facto limit much lower.
546 *
547 * Platform limits are somewhat arbitrary but deemed "more than good
548 * enough" for the foreseable future.
549 */
550 if (maxthread == 0) {
551 #ifdef _LP64
552 maxthread = MIN(maxproc * max_threads_per_proc, 1000000);
553 #else
554 maxthread = MIN(maxproc * max_threads_per_proc, 100000);
555 #endif
556 }
557
558 mtx_init(&tid_lock, "TID lock", NULL, MTX_DEF);
559 tid_bitmap = bit_alloc(maxthread, M_TIDHASH, M_WAITOK);
560 /*
561 * Handle thread0.
562 */
563 thread_count_inc();
564 tid0 = tid_alloc();
565 if (tid0 != THREAD0_TID)
566 panic("tid0 %d != %d\n", tid0, THREAD0_TID);
567
568 /*
569 * Thread structures are specially aligned so that (at least) the
570 * 5 lower bits of a pointer to 'struct thead' must be 0. These bits
571 * are used by synchronization primitives to store flags in pointers to
572 * such structures.
573 */
574 thread_zone = uma_zcreate("THREAD", sched_sizeof_thread(),
575 thread_ctor, thread_dtor, thread_init, thread_fini,
576 UMA_ALIGN_CACHE_AND_MASK(32 - 1), UMA_ZONE_NOFREE);
577 tidhashtbl = hashinit(maxproc / 2, M_TIDHASH, &tidhash);
578 tidhashlock = (tidhash + 1) / 64;
579 if (tidhashlock > 0)
580 tidhashlock--;
581 tidhashtbl_lock = malloc(sizeof(*tidhashtbl_lock) * (tidhashlock + 1),
582 M_TIDHASH, M_WAITOK | M_ZERO);
583 for (i = 0; i < tidhashlock + 1; i++)
584 rw_init(&tidhashtbl_lock[i], "tidhash");
585
586 TASK_INIT(&thread_reap_task, 0, thread_reap_task_cb, NULL);
587 callout_init(&thread_reap_callout, 1);
588 callout_reset(&thread_reap_callout, 5 * hz,
589 thread_reap_callout_cb, NULL);
590 ast_register(TDA_SUSPEND, ASTR_ASTF_REQUIRED, 0, ast_suspend);
591 }
592
593 /*
594 * Place an unused thread on the zombie list.
595 */
596 void
thread_zombie(struct thread * td)597 thread_zombie(struct thread *td)
598 {
599 struct thread_domain_data *tdd;
600 struct thread *ztd;
601
602 tdd = &thread_domain_data[td->td_allocdomain];
603 ztd = atomic_load_ptr(&tdd->tdd_zombies);
604 for (;;) {
605 td->td_zombie = ztd;
606 if (atomic_fcmpset_rel_ptr((uintptr_t *)&tdd->tdd_zombies,
607 (uintptr_t *)&ztd, (uintptr_t)td))
608 break;
609 continue;
610 }
611 }
612
613 /*
614 * Release a thread that has exited after cpu_throw().
615 */
616 void
thread_stash(struct thread * td)617 thread_stash(struct thread *td)
618 {
619 atomic_subtract_rel_int(&td->td_proc->p_exitthreads, 1);
620 thread_zombie(td);
621 }
622
623 /*
624 * Reap zombies from passed domain.
625 */
626 static void
thread_reap_domain(struct thread_domain_data * tdd)627 thread_reap_domain(struct thread_domain_data *tdd)
628 {
629 struct thread *itd, *ntd;
630 struct tidbatch tidbatch;
631 struct credbatch credbatch;
632 struct limbatch limbatch;
633 struct tdcountbatch tdcountbatch;
634
635 /*
636 * Reading upfront is pessimal if followed by concurrent atomic_swap,
637 * but most of the time the list is empty.
638 */
639 if (tdd->tdd_zombies == NULL)
640 return;
641
642 itd = (struct thread *)atomic_swap_ptr((uintptr_t *)&tdd->tdd_zombies,
643 (uintptr_t)NULL);
644 if (itd == NULL)
645 return;
646
647 /*
648 * Multiple CPUs can get here, the race is fine as ticks is only
649 * advisory.
650 */
651 tdd->tdd_reapticks = ticks;
652
653 tidbatch_prep(&tidbatch);
654 credbatch_prep(&credbatch);
655 limbatch_prep(&limbatch);
656 tdcountbatch_prep(&tdcountbatch);
657
658 while (itd != NULL) {
659 ntd = itd->td_zombie;
660 EVENTHANDLER_DIRECT_INVOKE(thread_dtor, itd);
661
662 tidbatch_add(&tidbatch, itd);
663 credbatch_add(&credbatch, itd);
664 limbatch_add(&limbatch, itd);
665 tdcountbatch_add(&tdcountbatch, itd);
666
667 thread_free_batched(itd);
668
669 tidbatch_process(&tidbatch);
670 credbatch_process(&credbatch);
671 limbatch_process(&limbatch);
672 tdcountbatch_process(&tdcountbatch);
673
674 itd = ntd;
675 }
676
677 tidbatch_final(&tidbatch);
678 credbatch_final(&credbatch);
679 limbatch_final(&limbatch);
680 tdcountbatch_final(&tdcountbatch);
681 }
682
683 /*
684 * Reap zombies from all domains.
685 */
686 static void
thread_reap_all(void)687 thread_reap_all(void)
688 {
689 struct thread_domain_data *tdd;
690 int i, domain;
691
692 domain = PCPU_GET(domain);
693 for (i = 0; i < vm_ndomains; i++) {
694 tdd = &thread_domain_data[(i + domain) % vm_ndomains];
695 thread_reap_domain(tdd);
696 }
697 }
698
699 /*
700 * Reap zombies from local domain.
701 */
702 static void
thread_reap(void)703 thread_reap(void)
704 {
705 struct thread_domain_data *tdd;
706 int domain;
707
708 domain = PCPU_GET(domain);
709 tdd = &thread_domain_data[domain];
710
711 thread_reap_domain(tdd);
712 }
713
714 static void
thread_reap_task_cb(void * arg __unused,int pending __unused)715 thread_reap_task_cb(void *arg __unused, int pending __unused)
716 {
717
718 thread_reap_all();
719 }
720
721 static void
thread_reap_callout_cb(void * arg __unused)722 thread_reap_callout_cb(void *arg __unused)
723 {
724 struct thread_domain_data *tdd;
725 int i, cticks, lticks;
726 bool wantreap;
727
728 wantreap = false;
729 cticks = atomic_load_int(&ticks);
730 for (i = 0; i < vm_ndomains; i++) {
731 tdd = &thread_domain_data[i];
732 lticks = tdd->tdd_reapticks;
733 if (tdd->tdd_zombies != NULL &&
734 (u_int)(cticks - lticks) > 5 * hz) {
735 wantreap = true;
736 break;
737 }
738 }
739
740 if (wantreap)
741 taskqueue_enqueue(taskqueue_thread, &thread_reap_task);
742 callout_reset(&thread_reap_callout, 5 * hz,
743 thread_reap_callout_cb, NULL);
744 }
745
746 /*
747 * Calling this function guarantees that any thread that exited before
748 * the call is reaped when the function returns. By 'exited' we mean
749 * a thread removed from the process linkage with thread_unlink().
750 * Practically this means that caller must lock/unlock corresponding
751 * process lock before the call, to synchronize with thread_exit().
752 */
753 void
thread_reap_barrier(void)754 thread_reap_barrier(void)
755 {
756 struct task *t;
757
758 /*
759 * First do context switches to each CPU to ensure that all
760 * PCPU pc_deadthreads are moved to zombie list.
761 */
762 quiesce_all_cpus("", PDROP);
763
764 /*
765 * Second, fire the task in the same thread as normal
766 * thread_reap() is done, to serialize reaping.
767 */
768 t = malloc(sizeof(*t), M_TEMP, M_WAITOK);
769 TASK_INIT(t, 0, thread_reap_task_cb, t);
770 taskqueue_enqueue(taskqueue_thread, t);
771 taskqueue_drain(taskqueue_thread, t);
772 free(t, M_TEMP);
773 }
774
775 /*
776 * Allocate a thread.
777 */
778 struct thread *
thread_alloc(int pages)779 thread_alloc(int pages)
780 {
781 struct thread *td;
782 lwpid_t tid;
783
784 if (!thread_count_inc()) {
785 return (NULL);
786 }
787
788 tid = tid_alloc();
789 td = uma_zalloc(thread_zone, M_WAITOK);
790 KASSERT(td->td_kstack == 0, ("thread_alloc got thread with kstack"));
791 if (!vm_thread_new(td, pages)) {
792 uma_zfree(thread_zone, td);
793 tid_free(tid);
794 thread_count_dec();
795 return (NULL);
796 }
797 td->td_tid = tid;
798 bzero(&td->td_sa.args, sizeof(td->td_sa.args));
799 kasan_thread_alloc(td);
800 kmsan_thread_alloc(td);
801 cpu_thread_alloc(td);
802 EVENTHANDLER_DIRECT_INVOKE(thread_ctor, td);
803 return (td);
804 }
805
806 int
thread_recycle(struct thread * td,int pages)807 thread_recycle(struct thread *td, int pages)
808 {
809 if (td->td_kstack == 0 || td->td_kstack_pages != pages) {
810 if (td->td_kstack != 0)
811 vm_thread_dispose(td);
812 if (!vm_thread_new(td, pages))
813 return (ENOMEM);
814 cpu_thread_alloc(td);
815 }
816 kasan_thread_alloc(td);
817 kmsan_thread_alloc(td);
818 return (0);
819 }
820
821 /*
822 * Deallocate a thread.
823 */
824 static void
thread_free_batched(struct thread * td)825 thread_free_batched(struct thread *td)
826 {
827
828 lock_profile_thread_exit(td);
829 if (td->td_cpuset)
830 cpuset_rel(td->td_cpuset);
831 td->td_cpuset = NULL;
832 cpu_thread_free(td);
833 if (td->td_kstack != 0)
834 vm_thread_dispose(td);
835 callout_drain(&td->td_slpcallout);
836 /*
837 * Freeing handled by the caller.
838 */
839 td->td_tid = -1;
840 kmsan_thread_free(td);
841 uma_zfree(thread_zone, td);
842 }
843
844 void
thread_free(struct thread * td)845 thread_free(struct thread *td)
846 {
847 lwpid_t tid;
848
849 EVENTHANDLER_DIRECT_INVOKE(thread_dtor, td);
850 tid = td->td_tid;
851 thread_free_batched(td);
852 tid_free(tid);
853 thread_count_dec();
854 }
855
856 void
thread_cow_get_proc(struct thread * newtd,struct proc * p)857 thread_cow_get_proc(struct thread *newtd, struct proc *p)
858 {
859
860 PROC_LOCK_ASSERT(p, MA_OWNED);
861 newtd->td_realucred = crcowget(p->p_ucred);
862 newtd->td_ucred = newtd->td_realucred;
863 newtd->td_limit = lim_hold(p->p_limit);
864 newtd->td_cowgen = p->p_cowgen;
865 }
866
867 void
thread_cow_get(struct thread * newtd,struct thread * td)868 thread_cow_get(struct thread *newtd, struct thread *td)
869 {
870
871 MPASS(td->td_realucred == td->td_ucred);
872 newtd->td_realucred = crcowget(td->td_realucred);
873 newtd->td_ucred = newtd->td_realucred;
874 newtd->td_limit = lim_hold(td->td_limit);
875 newtd->td_cowgen = td->td_cowgen;
876 }
877
878 void
thread_cow_free(struct thread * td)879 thread_cow_free(struct thread *td)
880 {
881
882 if (td->td_realucred != NULL)
883 crcowfree(td);
884 if (td->td_limit != NULL)
885 lim_free(td->td_limit);
886 }
887
888 void
thread_cow_update(struct thread * td)889 thread_cow_update(struct thread *td)
890 {
891 struct proc *p;
892 struct ucred *oldcred;
893 struct plimit *oldlimit;
894
895 p = td->td_proc;
896 PROC_LOCK(p);
897 oldcred = crcowsync();
898 oldlimit = lim_cowsync();
899 td->td_cowgen = p->p_cowgen;
900 PROC_UNLOCK(p);
901 if (oldcred != NULL)
902 crfree(oldcred);
903 if (oldlimit != NULL)
904 lim_free(oldlimit);
905 }
906
907 void
thread_cow_synced(struct thread * td)908 thread_cow_synced(struct thread *td)
909 {
910 struct proc *p;
911
912 p = td->td_proc;
913 PROC_LOCK_ASSERT(p, MA_OWNED);
914 MPASS(td->td_cowgen != p->p_cowgen);
915 MPASS(td->td_ucred == p->p_ucred);
916 MPASS(td->td_limit == p->p_limit);
917 td->td_cowgen = p->p_cowgen;
918 }
919
920 /*
921 * Discard the current thread and exit from its context.
922 * Always called with scheduler locked.
923 *
924 * Because we can't free a thread while we're operating under its context,
925 * push the current thread into our CPU's deadthread holder. This means
926 * we needn't worry about someone else grabbing our context before we
927 * do a cpu_throw().
928 */
929 void
thread_exit(void)930 thread_exit(void)
931 {
932 uint64_t runtime, new_switchtime;
933 struct thread *td;
934 struct thread *td2;
935 struct proc *p;
936
937 td = curthread;
938 p = td->td_proc;
939
940 PROC_SLOCK_ASSERT(p, MA_OWNED);
941 mtx_assert(&Giant, MA_NOTOWNED);
942
943 PROC_LOCK_ASSERT(p, MA_OWNED);
944 KASSERT(p != NULL, ("thread exiting without a process"));
945 CTR3(KTR_PROC, "thread_exit: thread %p (pid %ld, %s)", td,
946 (long)p->p_pid, td->td_name);
947 SDT_PROBE0(proc, , , lwp__exit);
948 KASSERT(TAILQ_EMPTY(&td->td_sigqueue.sq_list), ("signal pending"));
949 MPASS(td->td_realucred == td->td_ucred);
950
951 /*
952 * drop FPU & debug register state storage, or any other
953 * architecture specific resources that
954 * would not be on a new untouched process.
955 */
956 cpu_thread_exit(td);
957
958 /*
959 * The last thread is left attached to the process
960 * So that the whole bundle gets recycled. Skip
961 * all this stuff if we never had threads.
962 * EXIT clears all sign of other threads when
963 * it goes to single threading, so the last thread always
964 * takes the short path.
965 */
966 if (p->p_flag & P_HADTHREADS) {
967 if (p->p_numthreads > 1) {
968 atomic_add_int(&td->td_proc->p_exitthreads, 1);
969 thread_unlink(td);
970 td2 = FIRST_THREAD_IN_PROC(p);
971 sched_exit_thread(td2, td);
972
973 /*
974 * The test below is NOT true if we are the
975 * sole exiting thread. P_STOPPED_SINGLE is unset
976 * in exit1() after it is the only survivor.
977 */
978 if (P_SHOULDSTOP(p) == P_STOPPED_SINGLE) {
979 if (p->p_numthreads == p->p_suspcount) {
980 thread_lock(p->p_singlethread);
981 thread_unsuspend_one(p->p_singlethread,
982 p, false);
983 }
984 }
985
986 PCPU_SET(deadthread, td);
987 } else {
988 /*
989 * The last thread is exiting.. but not through exit()
990 */
991 panic ("thread_exit: Last thread exiting on its own");
992 }
993 }
994 #ifdef HWPMC_HOOKS
995 /*
996 * If this thread is part of a process that is being tracked by hwpmc(4),
997 * inform the module of the thread's impending exit.
998 */
999 if (PMC_PROC_IS_USING_PMCS(td->td_proc)) {
1000 PMC_SWITCH_CONTEXT(td, PMC_FN_CSW_OUT);
1001 PMC_CALL_HOOK_UNLOCKED(td, PMC_FN_THR_EXIT, NULL);
1002 } else if (PMC_SYSTEM_SAMPLING_ACTIVE())
1003 PMC_CALL_HOOK_UNLOCKED(td, PMC_FN_THR_EXIT_LOG, NULL);
1004 #endif
1005 PROC_UNLOCK(p);
1006 PROC_STATLOCK(p);
1007 thread_lock(td);
1008 PROC_SUNLOCK(p);
1009
1010 /* Do the same timestamp bookkeeping that mi_switch() would do. */
1011 new_switchtime = cpu_ticks();
1012 runtime = new_switchtime - PCPU_GET(switchtime);
1013 td->td_runtime += runtime;
1014 td->td_incruntime += runtime;
1015 PCPU_SET(switchtime, new_switchtime);
1016 PCPU_SET(switchticks, ticks);
1017 VM_CNT_INC(v_swtch);
1018
1019 /* Save our resource usage in our process. */
1020 td->td_ru.ru_nvcsw++;
1021 ruxagg_locked(p, td);
1022 rucollect(&p->p_ru, &td->td_ru);
1023 PROC_STATUNLOCK(p);
1024
1025 TD_SET_STATE(td, TDS_INACTIVE);
1026 #ifdef WITNESS
1027 witness_thread_exit(td);
1028 #endif
1029 CTR1(KTR_PROC, "thread_exit: cpu_throw() thread %p", td);
1030 sched_throw(td);
1031 panic("I'm a teapot!");
1032 /* NOTREACHED */
1033 }
1034
1035 /*
1036 * Do any thread specific cleanups that may be needed in wait()
1037 * called with Giant, proc and schedlock not held.
1038 */
1039 void
thread_wait(struct proc * p)1040 thread_wait(struct proc *p)
1041 {
1042 struct thread *td;
1043
1044 mtx_assert(&Giant, MA_NOTOWNED);
1045 KASSERT(p->p_numthreads == 1, ("multiple threads in thread_wait()"));
1046 KASSERT(p->p_exitthreads == 0, ("p_exitthreads leaking"));
1047 td = FIRST_THREAD_IN_PROC(p);
1048 /* Lock the last thread so we spin until it exits cpu_throw(). */
1049 thread_lock(td);
1050 thread_unlock(td);
1051 lock_profile_thread_exit(td);
1052 cpuset_rel(td->td_cpuset);
1053 td->td_cpuset = NULL;
1054 cpu_thread_clean(td);
1055 thread_cow_free(td);
1056 callout_drain(&td->td_slpcallout);
1057 thread_reap(); /* check for zombie threads etc. */
1058 }
1059
1060 /*
1061 * Link a thread to a process.
1062 * set up anything that needs to be initialized for it to
1063 * be used by the process.
1064 */
1065 void
thread_link(struct thread * td,struct proc * p)1066 thread_link(struct thread *td, struct proc *p)
1067 {
1068
1069 /*
1070 * XXX This can't be enabled because it's called for proc0 before
1071 * its lock has been created.
1072 * PROC_LOCK_ASSERT(p, MA_OWNED);
1073 */
1074 TD_SET_STATE(td, TDS_INACTIVE);
1075 td->td_proc = p;
1076 td->td_flags = TDF_INMEM;
1077
1078 LIST_INIT(&td->td_contested);
1079 LIST_INIT(&td->td_lprof[0]);
1080 LIST_INIT(&td->td_lprof[1]);
1081 #ifdef EPOCH_TRACE
1082 SLIST_INIT(&td->td_epochs);
1083 #endif
1084 sigqueue_init(&td->td_sigqueue, p);
1085 callout_init(&td->td_slpcallout, 1);
1086 TAILQ_INSERT_TAIL(&p->p_threads, td, td_plist);
1087 p->p_numthreads++;
1088 }
1089
1090 /*
1091 * Called from:
1092 * thread_exit()
1093 */
1094 void
thread_unlink(struct thread * td)1095 thread_unlink(struct thread *td)
1096 {
1097 struct proc *p = td->td_proc;
1098
1099 PROC_LOCK_ASSERT(p, MA_OWNED);
1100 #ifdef EPOCH_TRACE
1101 MPASS(SLIST_EMPTY(&td->td_epochs));
1102 #endif
1103
1104 TAILQ_REMOVE(&p->p_threads, td, td_plist);
1105 p->p_numthreads--;
1106 /* could clear a few other things here */
1107 /* Must NOT clear links to proc! */
1108 }
1109
1110 static int
calc_remaining(struct proc * p,int mode)1111 calc_remaining(struct proc *p, int mode)
1112 {
1113 int remaining;
1114
1115 PROC_LOCK_ASSERT(p, MA_OWNED);
1116 PROC_SLOCK_ASSERT(p, MA_OWNED);
1117 if (mode == SINGLE_EXIT)
1118 remaining = p->p_numthreads;
1119 else if (mode == SINGLE_BOUNDARY)
1120 remaining = p->p_numthreads - p->p_boundary_count;
1121 else if (mode == SINGLE_NO_EXIT || mode == SINGLE_ALLPROC)
1122 remaining = p->p_numthreads - p->p_suspcount;
1123 else
1124 panic("calc_remaining: wrong mode %d", mode);
1125 return (remaining);
1126 }
1127
1128 static int
remain_for_mode(int mode)1129 remain_for_mode(int mode)
1130 {
1131
1132 return (mode == SINGLE_ALLPROC ? 0 : 1);
1133 }
1134
1135 static void
weed_inhib(int mode,struct thread * td2,struct proc * p)1136 weed_inhib(int mode, struct thread *td2, struct proc *p)
1137 {
1138 PROC_LOCK_ASSERT(p, MA_OWNED);
1139 PROC_SLOCK_ASSERT(p, MA_OWNED);
1140 THREAD_LOCK_ASSERT(td2, MA_OWNED);
1141
1142 /*
1143 * Since the thread lock is dropped by the scheduler we have
1144 * to retry to check for races.
1145 */
1146 restart:
1147 switch (mode) {
1148 case SINGLE_EXIT:
1149 if (TD_IS_SUSPENDED(td2)) {
1150 thread_unsuspend_one(td2, p, true);
1151 thread_lock(td2);
1152 goto restart;
1153 }
1154 if (TD_CAN_ABORT(td2)) {
1155 sleepq_abort(td2, EINTR);
1156 return;
1157 }
1158 break;
1159 case SINGLE_BOUNDARY:
1160 case SINGLE_NO_EXIT:
1161 if (TD_IS_SUSPENDED(td2) &&
1162 (td2->td_flags & TDF_BOUNDARY) == 0) {
1163 thread_unsuspend_one(td2, p, false);
1164 thread_lock(td2);
1165 goto restart;
1166 }
1167 if (TD_CAN_ABORT(td2)) {
1168 sleepq_abort(td2, ERESTART);
1169 return;
1170 }
1171 break;
1172 case SINGLE_ALLPROC:
1173 /*
1174 * ALLPROC suspend tries to avoid spurious EINTR for
1175 * threads sleeping interruptable, by suspending the
1176 * thread directly, similarly to sig_suspend_threads().
1177 * Since such sleep is not neccessary performed at the user
1178 * boundary, TDF_ALLPROCSUSP is used to avoid immediate
1179 * un-suspend.
1180 */
1181 if (TD_IS_SUSPENDED(td2) &&
1182 (td2->td_flags & TDF_ALLPROCSUSP) == 0) {
1183 thread_unsuspend_one(td2, p, false);
1184 thread_lock(td2);
1185 goto restart;
1186 }
1187 if (TD_CAN_ABORT(td2)) {
1188 td2->td_flags |= TDF_ALLPROCSUSP;
1189 sleepq_abort(td2, ERESTART);
1190 return;
1191 }
1192 break;
1193 default:
1194 break;
1195 }
1196 thread_unlock(td2);
1197 }
1198
1199 /*
1200 * Enforce single-threading.
1201 *
1202 * Returns 1 if the caller must abort (another thread is waiting to
1203 * exit the process or similar). Process is locked!
1204 * Returns 0 when you are successfully the only thread running.
1205 * A process has successfully single threaded in the suspend mode when
1206 * There are no threads in user mode. Threads in the kernel must be
1207 * allowed to continue until they get to the user boundary. They may even
1208 * copy out their return values and data before suspending. They may however be
1209 * accelerated in reaching the user boundary as we will wake up
1210 * any sleeping threads that are interruptable. (PCATCH).
1211 */
1212 int
thread_single(struct proc * p,int mode)1213 thread_single(struct proc *p, int mode)
1214 {
1215 struct thread *td;
1216 struct thread *td2;
1217 int remaining;
1218
1219 td = curthread;
1220 KASSERT(mode == SINGLE_EXIT || mode == SINGLE_BOUNDARY ||
1221 mode == SINGLE_ALLPROC || mode == SINGLE_NO_EXIT,
1222 ("invalid mode %d", mode));
1223 /*
1224 * If allowing non-ALLPROC singlethreading for non-curproc
1225 * callers, calc_remaining() and remain_for_mode() should be
1226 * adjusted to also account for td->td_proc != p. For now
1227 * this is not implemented because it is not used.
1228 */
1229 KASSERT((mode == SINGLE_ALLPROC && td->td_proc != p) ||
1230 (mode != SINGLE_ALLPROC && td->td_proc == p),
1231 ("mode %d proc %p curproc %p", mode, p, td->td_proc));
1232 mtx_assert(&Giant, MA_NOTOWNED);
1233 PROC_LOCK_ASSERT(p, MA_OWNED);
1234
1235 /*
1236 * Is someone already single threading?
1237 * Or may be singlethreading is not needed at all.
1238 */
1239 if (mode == SINGLE_ALLPROC) {
1240 while ((p->p_flag & P_STOPPED_SINGLE) != 0) {
1241 if ((p->p_flag2 & P2_WEXIT) != 0)
1242 return (1);
1243 msleep(&p->p_flag, &p->p_mtx, PCATCH, "thrsgl", 0);
1244 }
1245 if ((p->p_flag & (P_STOPPED_SIG | P_TRACED)) != 0 ||
1246 (p->p_flag2 & P2_WEXIT) != 0)
1247 return (1);
1248 } else if ((p->p_flag & P_HADTHREADS) == 0)
1249 return (0);
1250 if (p->p_singlethread != NULL && p->p_singlethread != td)
1251 return (1);
1252
1253 if (mode == SINGLE_EXIT) {
1254 p->p_flag |= P_SINGLE_EXIT;
1255 p->p_flag &= ~P_SINGLE_BOUNDARY;
1256 } else {
1257 p->p_flag &= ~P_SINGLE_EXIT;
1258 if (mode == SINGLE_BOUNDARY)
1259 p->p_flag |= P_SINGLE_BOUNDARY;
1260 else
1261 p->p_flag &= ~P_SINGLE_BOUNDARY;
1262 }
1263 if (mode == SINGLE_ALLPROC)
1264 p->p_flag |= P_TOTAL_STOP;
1265 p->p_flag |= P_STOPPED_SINGLE;
1266 PROC_SLOCK(p);
1267 p->p_singlethread = td;
1268 remaining = calc_remaining(p, mode);
1269 while (remaining != remain_for_mode(mode)) {
1270 if (P_SHOULDSTOP(p) != P_STOPPED_SINGLE)
1271 goto stopme;
1272 FOREACH_THREAD_IN_PROC(p, td2) {
1273 if (td2 == td)
1274 continue;
1275 thread_lock(td2);
1276 ast_sched_locked(td2, TDA_SUSPEND);
1277 if (TD_IS_INHIBITED(td2)) {
1278 weed_inhib(mode, td2, p);
1279 #ifdef SMP
1280 } else if (TD_IS_RUNNING(td2)) {
1281 forward_signal(td2);
1282 thread_unlock(td2);
1283 #endif
1284 } else
1285 thread_unlock(td2);
1286 }
1287 remaining = calc_remaining(p, mode);
1288
1289 /*
1290 * Maybe we suspended some threads.. was it enough?
1291 */
1292 if (remaining == remain_for_mode(mode))
1293 break;
1294
1295 stopme:
1296 /*
1297 * Wake us up when everyone else has suspended.
1298 * In the mean time we suspend as well.
1299 */
1300 thread_suspend_switch(td, p);
1301 remaining = calc_remaining(p, mode);
1302 }
1303 if (mode == SINGLE_EXIT) {
1304 /*
1305 * Convert the process to an unthreaded process. The
1306 * SINGLE_EXIT is called by exit1() or execve(), in
1307 * both cases other threads must be retired.
1308 */
1309 KASSERT(p->p_numthreads == 1, ("Unthreading with >1 threads"));
1310 p->p_singlethread = NULL;
1311 p->p_flag &= ~(P_STOPPED_SINGLE | P_SINGLE_EXIT | P_HADTHREADS);
1312
1313 /*
1314 * Wait for any remaining threads to exit cpu_throw().
1315 */
1316 while (p->p_exitthreads != 0) {
1317 PROC_SUNLOCK(p);
1318 PROC_UNLOCK(p);
1319 sched_relinquish(td);
1320 PROC_LOCK(p);
1321 PROC_SLOCK(p);
1322 }
1323 } else if (mode == SINGLE_BOUNDARY) {
1324 /*
1325 * Wait until all suspended threads are removed from
1326 * the processors. The thread_suspend_check()
1327 * increments p_boundary_count while it is still
1328 * running, which makes it possible for the execve()
1329 * to destroy vmspace while our other threads are
1330 * still using the address space.
1331 *
1332 * We lock the thread, which is only allowed to
1333 * succeed after context switch code finished using
1334 * the address space.
1335 */
1336 FOREACH_THREAD_IN_PROC(p, td2) {
1337 if (td2 == td)
1338 continue;
1339 thread_lock(td2);
1340 KASSERT((td2->td_flags & TDF_BOUNDARY) != 0,
1341 ("td %p not on boundary", td2));
1342 KASSERT(TD_IS_SUSPENDED(td2),
1343 ("td %p is not suspended", td2));
1344 thread_unlock(td2);
1345 }
1346 }
1347 PROC_SUNLOCK(p);
1348 return (0);
1349 }
1350
1351 bool
thread_suspend_check_needed(void)1352 thread_suspend_check_needed(void)
1353 {
1354 struct proc *p;
1355 struct thread *td;
1356
1357 td = curthread;
1358 p = td->td_proc;
1359 PROC_LOCK_ASSERT(p, MA_OWNED);
1360 return (P_SHOULDSTOP(p) || ((p->p_flag & P_TRACED) != 0 &&
1361 (td->td_dbgflags & TDB_SUSPEND) != 0));
1362 }
1363
1364 /*
1365 * Called in from locations that can safely check to see
1366 * whether we have to suspend or at least throttle for a
1367 * single-thread event (e.g. fork).
1368 *
1369 * Such locations include userret().
1370 * If the "return_instead" argument is non zero, the thread must be able to
1371 * accept 0 (caller may continue), or 1 (caller must abort) as a result.
1372 *
1373 * The 'return_instead' argument tells the function if it may do a
1374 * thread_exit() or suspend, or whether the caller must abort and back
1375 * out instead.
1376 *
1377 * If the thread that set the single_threading request has set the
1378 * P_SINGLE_EXIT bit in the process flags then this call will never return
1379 * if 'return_instead' is false, but will exit.
1380 *
1381 * P_SINGLE_EXIT | return_instead == 0| return_instead != 0
1382 *---------------+--------------------+---------------------
1383 * 0 | returns 0 | returns 0 or 1
1384 * | when ST ends | immediately
1385 *---------------+--------------------+---------------------
1386 * 1 | thread exits | returns 1
1387 * | | immediately
1388 * 0 = thread_exit() or suspension ok,
1389 * other = return error instead of stopping the thread.
1390 *
1391 * While a full suspension is under effect, even a single threading
1392 * thread would be suspended if it made this call (but it shouldn't).
1393 * This call should only be made from places where
1394 * thread_exit() would be safe as that may be the outcome unless
1395 * return_instead is set.
1396 */
1397 int
thread_suspend_check(int return_instead)1398 thread_suspend_check(int return_instead)
1399 {
1400 struct thread *td;
1401 struct proc *p;
1402
1403 td = curthread;
1404 p = td->td_proc;
1405 mtx_assert(&Giant, MA_NOTOWNED);
1406 PROC_LOCK_ASSERT(p, MA_OWNED);
1407 while (thread_suspend_check_needed()) {
1408 if (P_SHOULDSTOP(p) == P_STOPPED_SINGLE) {
1409 KASSERT(p->p_singlethread != NULL,
1410 ("singlethread not set"));
1411 /*
1412 * The only suspension in action is a
1413 * single-threading. Single threader need not stop.
1414 * It is safe to access p->p_singlethread unlocked
1415 * because it can only be set to our address by us.
1416 */
1417 if (p->p_singlethread == td)
1418 return (0); /* Exempt from stopping. */
1419 }
1420 if ((p->p_flag & P_SINGLE_EXIT) && return_instead)
1421 return (EINTR);
1422
1423 /* Should we goto user boundary if we didn't come from there? */
1424 if (P_SHOULDSTOP(p) == P_STOPPED_SINGLE &&
1425 (p->p_flag & P_SINGLE_BOUNDARY) && return_instead)
1426 return (ERESTART);
1427
1428 /*
1429 * Ignore suspend requests if they are deferred.
1430 */
1431 if ((td->td_flags & TDF_SBDRY) != 0) {
1432 KASSERT(return_instead,
1433 ("TDF_SBDRY set for unsafe thread_suspend_check"));
1434 KASSERT((td->td_flags & (TDF_SEINTR | TDF_SERESTART)) !=
1435 (TDF_SEINTR | TDF_SERESTART),
1436 ("both TDF_SEINTR and TDF_SERESTART"));
1437 return (TD_SBDRY_INTR(td) ? TD_SBDRY_ERRNO(td) : 0);
1438 }
1439
1440 /*
1441 * If the process is waiting for us to exit,
1442 * this thread should just suicide.
1443 * Assumes that P_SINGLE_EXIT implies P_STOPPED_SINGLE.
1444 */
1445 if ((p->p_flag & P_SINGLE_EXIT) && (p->p_singlethread != td)) {
1446 PROC_UNLOCK(p);
1447
1448 /*
1449 * Allow Linux emulation layer to do some work
1450 * before thread suicide.
1451 */
1452 if (__predict_false(p->p_sysent->sv_thread_detach != NULL))
1453 (p->p_sysent->sv_thread_detach)(td);
1454 umtx_thread_exit(td);
1455 kern_thr_exit(td);
1456 panic("stopped thread did not exit");
1457 }
1458
1459 PROC_SLOCK(p);
1460 thread_stopped(p);
1461 if (P_SHOULDSTOP(p) == P_STOPPED_SINGLE) {
1462 if (p->p_numthreads == p->p_suspcount + 1) {
1463 thread_lock(p->p_singlethread);
1464 thread_unsuspend_one(p->p_singlethread, p,
1465 false);
1466 }
1467 }
1468 PROC_UNLOCK(p);
1469 thread_lock(td);
1470 /*
1471 * When a thread suspends, it just
1472 * gets taken off all queues.
1473 */
1474 thread_suspend_one(td);
1475 if (return_instead == 0) {
1476 p->p_boundary_count++;
1477 td->td_flags |= TDF_BOUNDARY;
1478 }
1479 PROC_SUNLOCK(p);
1480 mi_switch(SW_INVOL | SWT_SUSPEND);
1481 PROC_LOCK(p);
1482 }
1483 return (0);
1484 }
1485
1486 /*
1487 * Check for possible stops and suspensions while executing a
1488 * casueword or similar transiently failing operation.
1489 *
1490 * The sleep argument controls whether the function can handle a stop
1491 * request itself or it should return ERESTART and the request is
1492 * proceed at the kernel/user boundary in ast.
1493 *
1494 * Typically, when retrying due to casueword(9) failure (rv == 1), we
1495 * should handle the stop requests there, with exception of cases when
1496 * the thread owns a kernel resource, for instance busied the umtx
1497 * key, or when functions return immediately if thread_check_susp()
1498 * returned non-zero. On the other hand, retrying the whole lock
1499 * operation, we better not stop there but delegate the handling to
1500 * ast.
1501 *
1502 * If the request is for thread termination P_SINGLE_EXIT, we cannot
1503 * handle it at all, and simply return EINTR.
1504 */
1505 int
thread_check_susp(struct thread * td,bool sleep)1506 thread_check_susp(struct thread *td, bool sleep)
1507 {
1508 struct proc *p;
1509 int error;
1510
1511 /*
1512 * The check for TDA_SUSPEND is racy, but it is enough to
1513 * eventually break the lockstep loop.
1514 */
1515 if (!td_ast_pending(td, TDA_SUSPEND))
1516 return (0);
1517 error = 0;
1518 p = td->td_proc;
1519 PROC_LOCK(p);
1520 if (p->p_flag & P_SINGLE_EXIT)
1521 error = EINTR;
1522 else if (P_SHOULDSTOP(p) ||
1523 ((p->p_flag & P_TRACED) && (td->td_dbgflags & TDB_SUSPEND)))
1524 error = sleep ? thread_suspend_check(0) : ERESTART;
1525 PROC_UNLOCK(p);
1526 return (error);
1527 }
1528
1529 void
thread_suspend_switch(struct thread * td,struct proc * p)1530 thread_suspend_switch(struct thread *td, struct proc *p)
1531 {
1532
1533 KASSERT(!TD_IS_SUSPENDED(td), ("already suspended"));
1534 PROC_LOCK_ASSERT(p, MA_OWNED);
1535 PROC_SLOCK_ASSERT(p, MA_OWNED);
1536 /*
1537 * We implement thread_suspend_one in stages here to avoid
1538 * dropping the proc lock while the thread lock is owned.
1539 */
1540 if (p == td->td_proc) {
1541 thread_stopped(p);
1542 p->p_suspcount++;
1543 }
1544 PROC_UNLOCK(p);
1545 thread_lock(td);
1546 ast_unsched_locked(td, TDA_SUSPEND);
1547 TD_SET_SUSPENDED(td);
1548 sched_sleep(td, 0);
1549 PROC_SUNLOCK(p);
1550 DROP_GIANT();
1551 mi_switch(SW_VOL | SWT_SUSPEND);
1552 PICKUP_GIANT();
1553 PROC_LOCK(p);
1554 PROC_SLOCK(p);
1555 }
1556
1557 void
thread_suspend_one(struct thread * td)1558 thread_suspend_one(struct thread *td)
1559 {
1560 struct proc *p;
1561
1562 p = td->td_proc;
1563 PROC_SLOCK_ASSERT(p, MA_OWNED);
1564 THREAD_LOCK_ASSERT(td, MA_OWNED);
1565 KASSERT(!TD_IS_SUSPENDED(td), ("already suspended"));
1566 p->p_suspcount++;
1567 ast_unsched_locked(td, TDA_SUSPEND);
1568 TD_SET_SUSPENDED(td);
1569 sched_sleep(td, 0);
1570 }
1571
1572 static void
thread_unsuspend_one(struct thread * td,struct proc * p,bool boundary)1573 thread_unsuspend_one(struct thread *td, struct proc *p, bool boundary)
1574 {
1575
1576 THREAD_LOCK_ASSERT(td, MA_OWNED);
1577 KASSERT(TD_IS_SUSPENDED(td), ("Thread not suspended"));
1578 TD_CLR_SUSPENDED(td);
1579 td->td_flags &= ~TDF_ALLPROCSUSP;
1580 if (td->td_proc == p) {
1581 PROC_SLOCK_ASSERT(p, MA_OWNED);
1582 p->p_suspcount--;
1583 if (boundary && (td->td_flags & TDF_BOUNDARY) != 0) {
1584 td->td_flags &= ~TDF_BOUNDARY;
1585 p->p_boundary_count--;
1586 }
1587 }
1588 setrunnable(td, 0);
1589 }
1590
1591 void
thread_run_flash(struct thread * td)1592 thread_run_flash(struct thread *td)
1593 {
1594 struct proc *p;
1595
1596 p = td->td_proc;
1597 PROC_LOCK_ASSERT(p, MA_OWNED);
1598
1599 if (TD_ON_SLEEPQ(td))
1600 sleepq_remove_nested(td);
1601 else
1602 thread_lock(td);
1603
1604 THREAD_LOCK_ASSERT(td, MA_OWNED);
1605 KASSERT(TD_IS_SUSPENDED(td), ("Thread not suspended"));
1606
1607 TD_CLR_SUSPENDED(td);
1608 PROC_SLOCK(p);
1609 MPASS(p->p_suspcount > 0);
1610 p->p_suspcount--;
1611 PROC_SUNLOCK(p);
1612 setrunnable(td, 0);
1613 }
1614
1615 /*
1616 * Allow all threads blocked by single threading to continue running.
1617 */
1618 void
thread_unsuspend(struct proc * p)1619 thread_unsuspend(struct proc *p)
1620 {
1621 struct thread *td;
1622
1623 PROC_LOCK_ASSERT(p, MA_OWNED);
1624 PROC_SLOCK_ASSERT(p, MA_OWNED);
1625 if (!P_SHOULDSTOP(p)) {
1626 FOREACH_THREAD_IN_PROC(p, td) {
1627 thread_lock(td);
1628 if (TD_IS_SUSPENDED(td))
1629 thread_unsuspend_one(td, p, true);
1630 else
1631 thread_unlock(td);
1632 }
1633 } else if (P_SHOULDSTOP(p) == P_STOPPED_SINGLE &&
1634 p->p_numthreads == p->p_suspcount) {
1635 /*
1636 * Stopping everything also did the job for the single
1637 * threading request. Now we've downgraded to single-threaded,
1638 * let it continue.
1639 */
1640 if (p->p_singlethread->td_proc == p) {
1641 thread_lock(p->p_singlethread);
1642 thread_unsuspend_one(p->p_singlethread, p, false);
1643 }
1644 }
1645 }
1646
1647 /*
1648 * End the single threading mode..
1649 */
1650 void
thread_single_end(struct proc * p,int mode)1651 thread_single_end(struct proc *p, int mode)
1652 {
1653 struct thread *td;
1654
1655 KASSERT(mode == SINGLE_EXIT || mode == SINGLE_BOUNDARY ||
1656 mode == SINGLE_ALLPROC || mode == SINGLE_NO_EXIT,
1657 ("invalid mode %d", mode));
1658 PROC_LOCK_ASSERT(p, MA_OWNED);
1659 KASSERT((mode == SINGLE_ALLPROC && (p->p_flag & P_TOTAL_STOP) != 0) ||
1660 (mode != SINGLE_ALLPROC && (p->p_flag & P_TOTAL_STOP) == 0),
1661 ("mode %d does not match P_TOTAL_STOP", mode));
1662 KASSERT(mode == SINGLE_ALLPROC || p->p_singlethread == curthread,
1663 ("thread_single_end from other thread %p %p",
1664 curthread, p->p_singlethread));
1665 KASSERT(mode != SINGLE_BOUNDARY ||
1666 (p->p_flag & P_SINGLE_BOUNDARY) != 0,
1667 ("mis-matched SINGLE_BOUNDARY flags %x", p->p_flag));
1668 p->p_flag &= ~(P_STOPPED_SINGLE | P_SINGLE_EXIT | P_SINGLE_BOUNDARY |
1669 P_TOTAL_STOP);
1670 PROC_SLOCK(p);
1671 p->p_singlethread = NULL;
1672
1673 /*
1674 * If there are other threads they may now run,
1675 * unless of course there is a blanket 'stop order'
1676 * on the process. The single threader must be allowed
1677 * to continue however as this is a bad place to stop.
1678 */
1679 if (p->p_numthreads != remain_for_mode(mode) && !P_SHOULDSTOP(p)) {
1680 FOREACH_THREAD_IN_PROC(p, td) {
1681 thread_lock(td);
1682 if (TD_IS_SUSPENDED(td))
1683 thread_unsuspend_one(td, p, true);
1684 else
1685 thread_unlock(td);
1686 }
1687 }
1688 KASSERT(mode != SINGLE_BOUNDARY || p->p_boundary_count == 0,
1689 ("inconsistent boundary count %d", p->p_boundary_count));
1690 PROC_SUNLOCK(p);
1691 wakeup(&p->p_flag);
1692 }
1693
1694 /*
1695 * Locate a thread by number and return with proc lock held.
1696 *
1697 * thread exit establishes proc -> tidhash lock ordering, but lookup
1698 * takes tidhash first and needs to return locked proc.
1699 *
1700 * The problem is worked around by relying on type-safety of both
1701 * structures and doing the work in 2 steps:
1702 * - tidhash-locked lookup which saves both thread and proc pointers
1703 * - proc-locked verification that the found thread still matches
1704 */
1705 static bool
tdfind_hash(lwpid_t tid,pid_t pid,struct proc ** pp,struct thread ** tdp)1706 tdfind_hash(lwpid_t tid, pid_t pid, struct proc **pp, struct thread **tdp)
1707 {
1708 #define RUN_THRESH 16
1709 struct proc *p;
1710 struct thread *td;
1711 int run;
1712 bool locked;
1713
1714 run = 0;
1715 rw_rlock(TIDHASHLOCK(tid));
1716 locked = true;
1717 LIST_FOREACH(td, TIDHASH(tid), td_hash) {
1718 if (td->td_tid != tid) {
1719 run++;
1720 continue;
1721 }
1722 p = td->td_proc;
1723 if (pid != -1 && p->p_pid != pid) {
1724 td = NULL;
1725 break;
1726 }
1727 if (run > RUN_THRESH) {
1728 if (rw_try_upgrade(TIDHASHLOCK(tid))) {
1729 LIST_REMOVE(td, td_hash);
1730 LIST_INSERT_HEAD(TIDHASH(td->td_tid),
1731 td, td_hash);
1732 rw_wunlock(TIDHASHLOCK(tid));
1733 locked = false;
1734 break;
1735 }
1736 }
1737 break;
1738 }
1739 if (locked)
1740 rw_runlock(TIDHASHLOCK(tid));
1741 if (td == NULL)
1742 return (false);
1743 *pp = p;
1744 *tdp = td;
1745 return (true);
1746 }
1747
1748 struct thread *
tdfind(lwpid_t tid,pid_t pid)1749 tdfind(lwpid_t tid, pid_t pid)
1750 {
1751 struct proc *p;
1752 struct thread *td;
1753
1754 td = curthread;
1755 if (td->td_tid == tid) {
1756 if (pid != -1 && td->td_proc->p_pid != pid)
1757 return (NULL);
1758 PROC_LOCK(td->td_proc);
1759 return (td);
1760 }
1761
1762 for (;;) {
1763 if (!tdfind_hash(tid, pid, &p, &td))
1764 return (NULL);
1765 PROC_LOCK(p);
1766 if (td->td_tid != tid) {
1767 PROC_UNLOCK(p);
1768 continue;
1769 }
1770 if (td->td_proc != p) {
1771 PROC_UNLOCK(p);
1772 continue;
1773 }
1774 if (p->p_state == PRS_NEW) {
1775 PROC_UNLOCK(p);
1776 return (NULL);
1777 }
1778 return (td);
1779 }
1780 }
1781
1782 void
tidhash_add(struct thread * td)1783 tidhash_add(struct thread *td)
1784 {
1785 rw_wlock(TIDHASHLOCK(td->td_tid));
1786 LIST_INSERT_HEAD(TIDHASH(td->td_tid), td, td_hash);
1787 rw_wunlock(TIDHASHLOCK(td->td_tid));
1788 }
1789
1790 void
tidhash_remove(struct thread * td)1791 tidhash_remove(struct thread *td)
1792 {
1793
1794 rw_wlock(TIDHASHLOCK(td->td_tid));
1795 LIST_REMOVE(td, td_hash);
1796 rw_wunlock(TIDHASHLOCK(td->td_tid));
1797 }
1798