2 * SPDX-License-Identifier: BSD-2-Clause-FreeBSD
4 * Copyright (C) 2001 Julian Elischer <julian@freebsd.org>.
7 * Redistribution and use in source and binary forms, with or without
8 * modification, are permitted provided that the following conditions
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.
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
31 #include "opt_witness.h"
32 #include "opt_hwpmc_hooks.h"
34 #include <sys/cdefs.h>
35 __FBSDID("$FreeBSD$");
37 #include <sys/param.h>
38 #include <sys/systm.h>
39 #include <sys/kernel.h>
41 #include <sys/mutex.h>
43 #include <sys/bitstring.h>
44 #include <sys/epoch.h>
45 #include <sys/rangelock.h>
46 #include <sys/resourcevar.h>
49 #include <sys/sched.h>
50 #include <sys/sleepqueue.h>
51 #include <sys/selinfo.h>
52 #include <sys/syscallsubr.h>
53 #include <sys/dtrace_bsd.h>
54 #include <sys/sysent.h>
55 #include <sys/turnstile.h>
56 #include <sys/taskqueue.h>
58 #include <sys/rwlock.h>
60 #include <sys/vmmeter.h>
61 #include <sys/cpuset.h>
63 #include <sys/pmckern.h>
67 #include <security/audit/audit.h>
71 #include <vm/vm_extern.h>
73 #include <vm/vm_phys.h>
74 #include <sys/eventhandler.h>
77 * Asserts below verify the stability of struct thread and struct proc
78 * layout, as exposed by KBI to modules. On head, the KBI is allowed
79 * to drift, change to the structures must be accompanied by the
82 * On the stable branches after KBI freeze, conditions must not be
83 * violated. Typically new fields are moved to the end of the
87 _Static_assert(offsetof(struct thread, td_flags) == 0xfc,
88 "struct thread KBI td_flags");
89 _Static_assert(offsetof(struct thread, td_pflags) == 0x104,
90 "struct thread KBI td_pflags");
91 _Static_assert(offsetof(struct thread, td_frame) == 0x4a0,
92 "struct thread KBI td_frame");
93 _Static_assert(offsetof(struct thread, td_emuldata) == 0x6b0,
94 "struct thread KBI td_emuldata");
95 _Static_assert(offsetof(struct proc, p_flag) == 0xb8,
96 "struct proc KBI p_flag");
97 _Static_assert(offsetof(struct proc, p_pid) == 0xc4,
98 "struct proc KBI p_pid");
99 _Static_assert(offsetof(struct proc, p_filemon) == 0x3c0,
100 "struct proc KBI p_filemon");
101 _Static_assert(offsetof(struct proc, p_comm) == 0x3d8,
102 "struct proc KBI p_comm");
103 _Static_assert(offsetof(struct proc, p_emuldata) == 0x4b8,
104 "struct proc KBI p_emuldata");
107 _Static_assert(offsetof(struct thread, td_flags) == 0x98,
108 "struct thread KBI td_flags");
109 _Static_assert(offsetof(struct thread, td_pflags) == 0xa0,
110 "struct thread KBI td_pflags");
111 _Static_assert(offsetof(struct thread, td_frame) == 0x300,
112 "struct thread KBI td_frame");
113 _Static_assert(offsetof(struct thread, td_emuldata) == 0x344,
114 "struct thread KBI td_emuldata");
115 _Static_assert(offsetof(struct proc, p_flag) == 0x6c,
116 "struct proc KBI p_flag");
117 _Static_assert(offsetof(struct proc, p_pid) == 0x78,
118 "struct proc KBI p_pid");
119 _Static_assert(offsetof(struct proc, p_filemon) == 0x26c,
120 "struct proc KBI p_filemon");
121 _Static_assert(offsetof(struct proc, p_comm) == 0x280,
122 "struct proc KBI p_comm");
123 _Static_assert(offsetof(struct proc, p_emuldata) == 0x30c,
124 "struct proc KBI p_emuldata");
127 SDT_PROVIDER_DECLARE(proc);
128 SDT_PROBE_DEFINE(proc, , , lwp__exit);
131 * thread related storage.
133 static uma_zone_t thread_zone;
135 struct thread_domain_data {
136 struct thread *tdd_zombies;
138 } __aligned(CACHE_LINE_SIZE);
140 static struct thread_domain_data thread_domain_data[MAXMEMDOM];
142 static struct task thread_reap_task;
143 static struct callout thread_reap_callout;
145 static void thread_zombie(struct thread *);
146 static void thread_reap(void);
147 static void thread_reap_all(void);
148 static void thread_reap_task_cb(void *, int);
149 static void thread_reap_callout_cb(void *);
150 static int thread_unsuspend_one(struct thread *td, struct proc *p,
152 static void thread_free_batched(struct thread *td);
154 static __exclusive_cache_line struct mtx tid_lock;
155 static bitstr_t *tid_bitmap;
157 static MALLOC_DEFINE(M_TIDHASH, "tidhash", "thread hash");
159 static int maxthread;
160 SYSCTL_INT(_kern, OID_AUTO, maxthread, CTLFLAG_RDTUN,
161 &maxthread, 0, "Maximum number of threads");
163 static __exclusive_cache_line int nthreads;
165 static LIST_HEAD(tidhashhead, thread) *tidhashtbl;
166 static u_long tidhash;
167 static u_long tidhashlock;
168 static struct rwlock *tidhashtbl_lock;
169 #define TIDHASH(tid) (&tidhashtbl[(tid) & tidhash])
170 #define TIDHASHLOCK(tid) (&tidhashtbl_lock[(tid) & tidhashlock])
172 EVENTHANDLER_LIST_DEFINE(thread_ctor);
173 EVENTHANDLER_LIST_DEFINE(thread_dtor);
174 EVENTHANDLER_LIST_DEFINE(thread_init);
175 EVENTHANDLER_LIST_DEFINE(thread_fini);
178 thread_count_inc_try(void)
182 nthreads_new = atomic_fetchadd_int(&nthreads, 1) + 1;
183 if (nthreads_new >= maxthread - 100) {
184 if (priv_check_cred(curthread->td_ucred, PRIV_MAXPROC) != 0 ||
185 nthreads_new >= maxthread) {
186 atomic_subtract_int(&nthreads, 1);
194 thread_count_inc(void)
196 static struct timeval lastfail;
200 if (thread_count_inc_try()) {
205 if (thread_count_inc_try()) {
209 if (ppsratecheck(&lastfail, &curfail, 1)) {
210 printf("maxthread limit exceeded by uid %u "
211 "(pid %d); consider increasing kern.maxthread\n",
212 curthread->td_ucred->cr_ruid, curproc->p_pid);
218 thread_count_sub(int n)
221 atomic_subtract_int(&nthreads, n);
225 thread_count_dec(void)
234 static lwpid_t trytid;
239 * It is an invariant that the bitmap is big enough to hold maxthread
240 * IDs. If we got to this point there has to be at least one free.
242 if (trytid >= maxthread)
244 bit_ffc_at(tid_bitmap, trytid, maxthread, &tid);
246 KASSERT(trytid != 0, ("unexpectedly ran out of IDs"));
248 bit_ffc_at(tid_bitmap, trytid, maxthread, &tid);
249 KASSERT(tid != -1, ("unexpectedly ran out of IDs"));
251 bit_set(tid_bitmap, tid);
253 mtx_unlock(&tid_lock);
254 return (tid + NO_PID);
258 tid_free_locked(lwpid_t rtid)
262 mtx_assert(&tid_lock, MA_OWNED);
263 KASSERT(rtid >= NO_PID,
264 ("%s: invalid tid %d\n", __func__, rtid));
266 KASSERT(bit_test(tid_bitmap, tid) != 0,
267 ("thread ID %d not allocated\n", rtid));
268 bit_clear(tid_bitmap, tid);
272 tid_free(lwpid_t rtid)
276 tid_free_locked(rtid);
277 mtx_unlock(&tid_lock);
281 tid_free_batch(lwpid_t *batch, int n)
286 for (i = 0; i < n; i++) {
287 tid_free_locked(batch[i]);
289 mtx_unlock(&tid_lock);
293 * Batching for thread reapping.
301 tidbatch_prep(struct tidbatch *tb)
308 tidbatch_add(struct tidbatch *tb, struct thread *td)
311 KASSERT(tb->n < nitems(tb->tab),
312 ("%s: count too high %d", __func__, tb->n));
313 tb->tab[tb->n] = td->td_tid;
318 tidbatch_process(struct tidbatch *tb)
321 KASSERT(tb->n <= nitems(tb->tab),
322 ("%s: count too high %d", __func__, tb->n));
323 if (tb->n == nitems(tb->tab)) {
324 tid_free_batch(tb->tab, tb->n);
330 tidbatch_final(struct tidbatch *tb)
333 KASSERT(tb->n <= nitems(tb->tab),
334 ("%s: count too high %d", __func__, tb->n));
336 tid_free_batch(tb->tab, tb->n);
341 * Prepare a thread for use.
344 thread_ctor(void *mem, int size, void *arg, int flags)
348 td = (struct thread *)mem;
349 td->td_state = TDS_INACTIVE;
350 td->td_lastcpu = td->td_oncpu = NOCPU;
353 * Note that td_critnest begins life as 1 because the thread is not
354 * running and is thereby implicitly waiting to be on the receiving
355 * end of a context switch.
358 td->td_lend_user_pri = PRI_MAX;
360 audit_thread_alloc(td);
363 kdtrace_thread_ctor(td);
365 umtx_thread_alloc(td);
366 MPASS(td->td_sel == NULL);
371 * Reclaim a thread after use.
374 thread_dtor(void *mem, int size, void *arg)
378 td = (struct thread *)mem;
381 /* Verify that this thread is in a safe state to free. */
382 switch (td->td_state) {
388 * We must never unlink a thread that is in one of
389 * these states, because it is currently active.
391 panic("bad state for thread unlinking");
396 panic("bad thread state");
401 audit_thread_free(td);
404 kdtrace_thread_dtor(td);
406 /* Free all OSD associated to this thread. */
408 td_softdep_cleanup(td);
409 MPASS(td->td_su == NULL);
414 * Initialize type-stable parts of a thread (when newly created).
417 thread_init(void *mem, int size, int flags)
421 td = (struct thread *)mem;
423 td->td_allocdomain = vm_phys_domain(vtophys(td));
424 td->td_sleepqueue = sleepq_alloc();
425 td->td_turnstile = turnstile_alloc();
427 EVENTHANDLER_DIRECT_INVOKE(thread_init, td);
428 umtx_thread_init(td);
435 * Tear down type-stable parts of a thread (just before being discarded).
438 thread_fini(void *mem, int size)
442 td = (struct thread *)mem;
443 EVENTHANDLER_DIRECT_INVOKE(thread_fini, td);
444 rlqentry_free(td->td_rlqe);
445 turnstile_free(td->td_turnstile);
446 sleepq_free(td->td_sleepqueue);
447 umtx_thread_fini(td);
448 MPASS(td->td_sel == NULL);
452 * For a newly created process,
453 * link up all the structures and its initial threads etc.
455 * {arch}/{arch}/machdep.c {arch}_init(), init386() etc.
456 * proc_dtor() (should go away)
460 proc_linkup0(struct proc *p, struct thread *td)
462 TAILQ_INIT(&p->p_threads); /* all threads in proc */
467 proc_linkup(struct proc *p, struct thread *td)
470 sigqueue_init(&p->p_sigqueue, p);
471 p->p_ksi = ksiginfo_alloc(1);
472 if (p->p_ksi != NULL) {
473 /* XXX p_ksi may be null if ksiginfo zone is not ready */
474 p->p_ksi->ksi_flags = KSI_EXT | KSI_INS;
476 LIST_INIT(&p->p_mqnotifier);
481 extern int max_threads_per_proc;
484 * Initialize global thread allocation resources.
494 * Place an upper limit on threads which can be allocated.
496 * Note that other factors may make the de facto limit much lower.
498 * Platform limits are somewhat arbitrary but deemed "more than good
499 * enough" for the foreseable future.
501 if (maxthread == 0) {
503 maxthread = MIN(maxproc * max_threads_per_proc, 1000000);
505 maxthread = MIN(maxproc * max_threads_per_proc, 100000);
509 mtx_init(&tid_lock, "TID lock", NULL, MTX_DEF);
510 tid_bitmap = bit_alloc(maxthread, M_TIDHASH, M_WAITOK);
516 if (tid0 != THREAD0_TID)
517 panic("tid0 %d != %d\n", tid0, THREAD0_TID);
519 flags = UMA_ZONE_NOFREE;
522 * Force thread structures to be allocated from the direct map.
523 * Otherwise, superpage promotions and demotions may temporarily
524 * invalidate thread structure mappings. For most dynamically allocated
525 * structures this is not a problem, but translation faults cannot be
526 * handled without accessing curthread.
528 flags |= UMA_ZONE_CONTIG;
530 thread_zone = uma_zcreate("THREAD", sched_sizeof_thread(),
531 thread_ctor, thread_dtor, thread_init, thread_fini,
533 tidhashtbl = hashinit(maxproc / 2, M_TIDHASH, &tidhash);
534 tidhashlock = (tidhash + 1) / 64;
537 tidhashtbl_lock = malloc(sizeof(*tidhashtbl_lock) * (tidhashlock + 1),
538 M_TIDHASH, M_WAITOK | M_ZERO);
539 for (i = 0; i < tidhashlock + 1; i++)
540 rw_init(&tidhashtbl_lock[i], "tidhash");
542 TASK_INIT(&thread_reap_task, 0, thread_reap_task_cb, NULL);
543 callout_init(&thread_reap_callout, 1);
544 callout_reset(&thread_reap_callout, 5 * hz, thread_reap_callout_cb, NULL);
548 * Place an unused thread on the zombie list.
551 thread_zombie(struct thread *td)
553 struct thread_domain_data *tdd;
556 tdd = &thread_domain_data[td->td_allocdomain];
557 ztd = atomic_load_ptr(&tdd->tdd_zombies);
560 if (atomic_fcmpset_rel_ptr((uintptr_t *)&tdd->tdd_zombies,
561 (uintptr_t *)&ztd, (uintptr_t)td))
568 * Release a thread that has exited after cpu_throw().
571 thread_stash(struct thread *td)
573 atomic_subtract_rel_int(&td->td_proc->p_exitthreads, 1);
578 * Reap zombies from passed domain.
581 thread_reap_domain(struct thread_domain_data *tdd)
583 struct thread *itd, *ntd;
584 struct tidbatch tidbatch;
585 struct credbatch credbatch;
591 * Reading upfront is pessimal if followed by concurrent atomic_swap,
592 * but most of the time the list is empty.
594 if (tdd->tdd_zombies == NULL)
597 itd = (struct thread *)atomic_swap_ptr((uintptr_t *)&tdd->tdd_zombies,
603 * Multiple CPUs can get here, the race is fine as ticks is only
606 tdd->tdd_reapticks = ticks;
608 tidbatch_prep(&tidbatch);
609 credbatch_prep(&credbatch);
614 while (itd != NULL) {
615 ntd = itd->td_zombie;
616 EVENTHANDLER_DIRECT_INVOKE(thread_dtor, itd);
617 tidbatch_add(&tidbatch, itd);
618 credbatch_add(&credbatch, itd);
619 MPASS(itd->td_limit != NULL);
620 if (lim != itd->td_limit) {
622 lim_freen(lim, limcount);
628 thread_free_batched(itd);
629 tidbatch_process(&tidbatch);
630 credbatch_process(&credbatch);
633 thread_count_sub(tdcount);
639 tidbatch_final(&tidbatch);
640 credbatch_final(&credbatch);
642 thread_count_sub(tdcount);
644 MPASS(limcount != 0);
645 lim_freen(lim, limcount);
649 * Reap zombies from all domains.
652 thread_reap_all(void)
654 struct thread_domain_data *tdd;
657 domain = PCPU_GET(domain);
658 for (i = 0; i < vm_ndomains; i++) {
659 tdd = &thread_domain_data[(i + domain) % vm_ndomains];
660 thread_reap_domain(tdd);
665 * Reap zombies from local domain.
670 struct thread_domain_data *tdd;
673 domain = PCPU_GET(domain);
674 tdd = &thread_domain_data[domain];
676 thread_reap_domain(tdd);
680 thread_reap_task_cb(void *arg __unused, int pending __unused)
687 thread_reap_callout_cb(void *arg __unused)
689 struct thread_domain_data *tdd;
690 int i, cticks, lticks;
694 cticks = atomic_load_int(&ticks);
695 for (i = 0; i < vm_ndomains; i++) {
696 tdd = &thread_domain_data[i];
697 lticks = tdd->tdd_reapticks;
698 if (tdd->tdd_zombies != NULL &&
699 (u_int)(cticks - lticks) > 5 * hz) {
706 taskqueue_enqueue(taskqueue_thread, &thread_reap_task);
707 callout_reset(&thread_reap_callout, 5 * hz, thread_reap_callout_cb, NULL);
714 thread_alloc(int pages)
719 if (!thread_count_inc()) {
724 td = uma_zalloc(thread_zone, M_WAITOK);
725 KASSERT(td->td_kstack == 0, ("thread_alloc got thread with kstack"));
726 if (!vm_thread_new(td, pages)) {
727 uma_zfree(thread_zone, td);
733 cpu_thread_alloc(td);
734 EVENTHANDLER_DIRECT_INVOKE(thread_ctor, td);
739 thread_alloc_stack(struct thread *td, int pages)
742 KASSERT(td->td_kstack == 0,
743 ("thread_alloc_stack called on a thread with kstack"));
744 if (!vm_thread_new(td, pages))
746 cpu_thread_alloc(td);
751 * Deallocate a thread.
754 thread_free_batched(struct thread *td)
757 lock_profile_thread_exit(td);
759 cpuset_rel(td->td_cpuset);
760 td->td_cpuset = NULL;
762 if (td->td_kstack != 0)
763 vm_thread_dispose(td);
764 callout_drain(&td->td_slpcallout);
766 * Freeing handled by the caller.
769 uma_zfree(thread_zone, td);
773 thread_free(struct thread *td)
777 EVENTHANDLER_DIRECT_INVOKE(thread_dtor, td);
779 thread_free_batched(td);
785 thread_cow_get_proc(struct thread *newtd, struct proc *p)
788 PROC_LOCK_ASSERT(p, MA_OWNED);
789 newtd->td_realucred = crcowget(p->p_ucred);
790 newtd->td_ucred = newtd->td_realucred;
791 newtd->td_limit = lim_hold(p->p_limit);
792 newtd->td_cowgen = p->p_cowgen;
796 thread_cow_get(struct thread *newtd, struct thread *td)
799 MPASS(td->td_realucred == td->td_ucred);
800 newtd->td_realucred = crcowget(td->td_realucred);
801 newtd->td_ucred = newtd->td_realucred;
802 newtd->td_limit = lim_hold(td->td_limit);
803 newtd->td_cowgen = td->td_cowgen;
807 thread_cow_free(struct thread *td)
810 if (td->td_realucred != NULL)
812 if (td->td_limit != NULL)
813 lim_free(td->td_limit);
817 thread_cow_update(struct thread *td)
820 struct ucred *oldcred;
821 struct plimit *oldlimit;
826 oldcred = crcowsync();
827 if (td->td_limit != p->p_limit) {
828 oldlimit = td->td_limit;
829 td->td_limit = lim_hold(p->p_limit);
831 td->td_cowgen = p->p_cowgen;
835 if (oldlimit != NULL)
840 * Discard the current thread and exit from its context.
841 * Always called with scheduler locked.
843 * Because we can't free a thread while we're operating under its context,
844 * push the current thread into our CPU's deadthread holder. This means
845 * we needn't worry about someone else grabbing our context before we
851 uint64_t runtime, new_switchtime;
860 PROC_SLOCK_ASSERT(p, MA_OWNED);
861 mtx_assert(&Giant, MA_NOTOWNED);
863 PROC_LOCK_ASSERT(p, MA_OWNED);
864 KASSERT(p != NULL, ("thread exiting without a process"));
865 CTR3(KTR_PROC, "thread_exit: thread %p (pid %ld, %s)", td,
866 (long)p->p_pid, td->td_name);
867 SDT_PROBE0(proc, , , lwp__exit);
868 KASSERT(TAILQ_EMPTY(&td->td_sigqueue.sq_list), ("signal pending"));
869 MPASS(td->td_realucred == td->td_ucred);
872 * drop FPU & debug register state storage, or any other
873 * architecture specific resources that
874 * would not be on a new untouched process.
879 * The last thread is left attached to the process
880 * So that the whole bundle gets recycled. Skip
881 * all this stuff if we never had threads.
882 * EXIT clears all sign of other threads when
883 * it goes to single threading, so the last thread always
884 * takes the short path.
886 if (p->p_flag & P_HADTHREADS) {
887 if (p->p_numthreads > 1) {
888 atomic_add_int(&td->td_proc->p_exitthreads, 1);
890 td2 = FIRST_THREAD_IN_PROC(p);
891 sched_exit_thread(td2, td);
894 * The test below is NOT true if we are the
895 * sole exiting thread. P_STOPPED_SINGLE is unset
896 * in exit1() after it is the only survivor.
898 if (P_SHOULDSTOP(p) == P_STOPPED_SINGLE) {
899 if (p->p_numthreads == p->p_suspcount) {
900 thread_lock(p->p_singlethread);
901 wakeup_swapper = thread_unsuspend_one(
902 p->p_singlethread, p, false);
908 PCPU_SET(deadthread, td);
911 * The last thread is exiting.. but not through exit()
913 panic ("thread_exit: Last thread exiting on its own");
918 * If this thread is part of a process that is being tracked by hwpmc(4),
919 * inform the module of the thread's impending exit.
921 if (PMC_PROC_IS_USING_PMCS(td->td_proc)) {
922 PMC_SWITCH_CONTEXT(td, PMC_FN_CSW_OUT);
923 PMC_CALL_HOOK_UNLOCKED(td, PMC_FN_THR_EXIT, NULL);
924 } else if (PMC_SYSTEM_SAMPLING_ACTIVE())
925 PMC_CALL_HOOK_UNLOCKED(td, PMC_FN_THR_EXIT_LOG, NULL);
932 /* Do the same timestamp bookkeeping that mi_switch() would do. */
933 new_switchtime = cpu_ticks();
934 runtime = new_switchtime - PCPU_GET(switchtime);
935 td->td_runtime += runtime;
936 td->td_incruntime += runtime;
937 PCPU_SET(switchtime, new_switchtime);
938 PCPU_SET(switchticks, ticks);
941 /* Save our resource usage in our process. */
942 td->td_ru.ru_nvcsw++;
943 ruxagg_locked(p, td);
944 rucollect(&p->p_ru, &td->td_ru);
947 td->td_state = TDS_INACTIVE;
949 witness_thread_exit(td);
951 CTR1(KTR_PROC, "thread_exit: cpu_throw() thread %p", td);
953 panic("I'm a teapot!");
958 * Do any thread specific cleanups that may be needed in wait()
959 * called with Giant, proc and schedlock not held.
962 thread_wait(struct proc *p)
966 mtx_assert(&Giant, MA_NOTOWNED);
967 KASSERT(p->p_numthreads == 1, ("multiple threads in thread_wait()"));
968 KASSERT(p->p_exitthreads == 0, ("p_exitthreads leaking"));
969 td = FIRST_THREAD_IN_PROC(p);
970 /* Lock the last thread so we spin until it exits cpu_throw(). */
973 lock_profile_thread_exit(td);
974 cpuset_rel(td->td_cpuset);
975 td->td_cpuset = NULL;
976 cpu_thread_clean(td);
978 callout_drain(&td->td_slpcallout);
979 thread_reap(); /* check for zombie threads etc. */
983 * Link a thread to a process.
984 * set up anything that needs to be initialized for it to
985 * be used by the process.
988 thread_link(struct thread *td, struct proc *p)
992 * XXX This can't be enabled because it's called for proc0 before
993 * its lock has been created.
994 * PROC_LOCK_ASSERT(p, MA_OWNED);
996 td->td_state = TDS_INACTIVE;
998 td->td_flags = TDF_INMEM;
1000 LIST_INIT(&td->td_contested);
1001 LIST_INIT(&td->td_lprof[0]);
1002 LIST_INIT(&td->td_lprof[1]);
1004 SLIST_INIT(&td->td_epochs);
1006 sigqueue_init(&td->td_sigqueue, p);
1007 callout_init(&td->td_slpcallout, 1);
1008 TAILQ_INSERT_TAIL(&p->p_threads, td, td_plist);
1017 thread_unlink(struct thread *td)
1019 struct proc *p = td->td_proc;
1021 PROC_LOCK_ASSERT(p, MA_OWNED);
1023 MPASS(SLIST_EMPTY(&td->td_epochs));
1026 TAILQ_REMOVE(&p->p_threads, td, td_plist);
1028 /* could clear a few other things here */
1029 /* Must NOT clear links to proc! */
1033 calc_remaining(struct proc *p, int mode)
1037 PROC_LOCK_ASSERT(p, MA_OWNED);
1038 PROC_SLOCK_ASSERT(p, MA_OWNED);
1039 if (mode == SINGLE_EXIT)
1040 remaining = p->p_numthreads;
1041 else if (mode == SINGLE_BOUNDARY)
1042 remaining = p->p_numthreads - p->p_boundary_count;
1043 else if (mode == SINGLE_NO_EXIT || mode == SINGLE_ALLPROC)
1044 remaining = p->p_numthreads - p->p_suspcount;
1046 panic("calc_remaining: wrong mode %d", mode);
1051 remain_for_mode(int mode)
1054 return (mode == SINGLE_ALLPROC ? 0 : 1);
1058 weed_inhib(int mode, struct thread *td2, struct proc *p)
1062 PROC_LOCK_ASSERT(p, MA_OWNED);
1063 PROC_SLOCK_ASSERT(p, MA_OWNED);
1064 THREAD_LOCK_ASSERT(td2, MA_OWNED);
1069 * Since the thread lock is dropped by the scheduler we have
1070 * to retry to check for races.
1075 if (TD_IS_SUSPENDED(td2)) {
1076 wakeup_swapper |= thread_unsuspend_one(td2, p, true);
1080 if (TD_CAN_ABORT(td2)) {
1081 wakeup_swapper |= sleepq_abort(td2, EINTR);
1082 return (wakeup_swapper);
1085 case SINGLE_BOUNDARY:
1086 case SINGLE_NO_EXIT:
1087 if (TD_IS_SUSPENDED(td2) &&
1088 (td2->td_flags & TDF_BOUNDARY) == 0) {
1089 wakeup_swapper |= thread_unsuspend_one(td2, p, false);
1093 if (TD_CAN_ABORT(td2)) {
1094 wakeup_swapper |= sleepq_abort(td2, ERESTART);
1095 return (wakeup_swapper);
1098 case SINGLE_ALLPROC:
1100 * ALLPROC suspend tries to avoid spurious EINTR for
1101 * threads sleeping interruptable, by suspending the
1102 * thread directly, similarly to sig_suspend_threads().
1103 * Since such sleep is not performed at the user
1104 * boundary, TDF_BOUNDARY flag is not set, and TDF_ALLPROCSUSP
1105 * is used to avoid immediate un-suspend.
1107 if (TD_IS_SUSPENDED(td2) && (td2->td_flags & (TDF_BOUNDARY |
1108 TDF_ALLPROCSUSP)) == 0) {
1109 wakeup_swapper |= thread_unsuspend_one(td2, p, false);
1113 if (TD_CAN_ABORT(td2)) {
1114 if ((td2->td_flags & TDF_SBDRY) == 0) {
1115 thread_suspend_one(td2);
1116 td2->td_flags |= TDF_ALLPROCSUSP;
1118 wakeup_swapper |= sleepq_abort(td2, ERESTART);
1119 return (wakeup_swapper);
1127 return (wakeup_swapper);
1131 * Enforce single-threading.
1133 * Returns 1 if the caller must abort (another thread is waiting to
1134 * exit the process or similar). Process is locked!
1135 * Returns 0 when you are successfully the only thread running.
1136 * A process has successfully single threaded in the suspend mode when
1137 * There are no threads in user mode. Threads in the kernel must be
1138 * allowed to continue until they get to the user boundary. They may even
1139 * copy out their return values and data before suspending. They may however be
1140 * accelerated in reaching the user boundary as we will wake up
1141 * any sleeping threads that are interruptable. (PCATCH).
1144 thread_single(struct proc *p, int mode)
1148 int remaining, wakeup_swapper;
1151 KASSERT(mode == SINGLE_EXIT || mode == SINGLE_BOUNDARY ||
1152 mode == SINGLE_ALLPROC || mode == SINGLE_NO_EXIT,
1153 ("invalid mode %d", mode));
1155 * If allowing non-ALLPROC singlethreading for non-curproc
1156 * callers, calc_remaining() and remain_for_mode() should be
1157 * adjusted to also account for td->td_proc != p. For now
1158 * this is not implemented because it is not used.
1160 KASSERT((mode == SINGLE_ALLPROC && td->td_proc != p) ||
1161 (mode != SINGLE_ALLPROC && td->td_proc == p),
1162 ("mode %d proc %p curproc %p", mode, p, td->td_proc));
1163 mtx_assert(&Giant, MA_NOTOWNED);
1164 PROC_LOCK_ASSERT(p, MA_OWNED);
1166 if ((p->p_flag & P_HADTHREADS) == 0 && mode != SINGLE_ALLPROC)
1169 /* Is someone already single threading? */
1170 if (p->p_singlethread != NULL && p->p_singlethread != td)
1173 if (mode == SINGLE_EXIT) {
1174 p->p_flag |= P_SINGLE_EXIT;
1175 p->p_flag &= ~P_SINGLE_BOUNDARY;
1177 p->p_flag &= ~P_SINGLE_EXIT;
1178 if (mode == SINGLE_BOUNDARY)
1179 p->p_flag |= P_SINGLE_BOUNDARY;
1181 p->p_flag &= ~P_SINGLE_BOUNDARY;
1183 if (mode == SINGLE_ALLPROC)
1184 p->p_flag |= P_TOTAL_STOP;
1185 p->p_flag |= P_STOPPED_SINGLE;
1187 p->p_singlethread = td;
1188 remaining = calc_remaining(p, mode);
1189 while (remaining != remain_for_mode(mode)) {
1190 if (P_SHOULDSTOP(p) != P_STOPPED_SINGLE)
1193 FOREACH_THREAD_IN_PROC(p, td2) {
1197 td2->td_flags |= TDF_ASTPENDING | TDF_NEEDSUSPCHK;
1198 if (TD_IS_INHIBITED(td2)) {
1199 wakeup_swapper |= weed_inhib(mode, td2, p);
1201 } else if (TD_IS_RUNNING(td2) && td != td2) {
1202 forward_signal(td2);
1210 remaining = calc_remaining(p, mode);
1213 * Maybe we suspended some threads.. was it enough?
1215 if (remaining == remain_for_mode(mode))
1220 * Wake us up when everyone else has suspended.
1221 * In the mean time we suspend as well.
1223 thread_suspend_switch(td, p);
1224 remaining = calc_remaining(p, mode);
1226 if (mode == SINGLE_EXIT) {
1228 * Convert the process to an unthreaded process. The
1229 * SINGLE_EXIT is called by exit1() or execve(), in
1230 * both cases other threads must be retired.
1232 KASSERT(p->p_numthreads == 1, ("Unthreading with >1 threads"));
1233 p->p_singlethread = NULL;
1234 p->p_flag &= ~(P_STOPPED_SINGLE | P_SINGLE_EXIT | P_HADTHREADS);
1237 * Wait for any remaining threads to exit cpu_throw().
1239 while (p->p_exitthreads != 0) {
1242 sched_relinquish(td);
1246 } else if (mode == SINGLE_BOUNDARY) {
1248 * Wait until all suspended threads are removed from
1249 * the processors. The thread_suspend_check()
1250 * increments p_boundary_count while it is still
1251 * running, which makes it possible for the execve()
1252 * to destroy vmspace while our other threads are
1253 * still using the address space.
1255 * We lock the thread, which is only allowed to
1256 * succeed after context switch code finished using
1257 * the address space.
1259 FOREACH_THREAD_IN_PROC(p, td2) {
1263 KASSERT((td2->td_flags & TDF_BOUNDARY) != 0,
1264 ("td %p not on boundary", td2));
1265 KASSERT(TD_IS_SUSPENDED(td2),
1266 ("td %p is not suspended", td2));
1275 thread_suspend_check_needed(void)
1282 PROC_LOCK_ASSERT(p, MA_OWNED);
1283 return (P_SHOULDSTOP(p) || ((p->p_flag & P_TRACED) != 0 &&
1284 (td->td_dbgflags & TDB_SUSPEND) != 0));
1288 * Called in from locations that can safely check to see
1289 * whether we have to suspend or at least throttle for a
1290 * single-thread event (e.g. fork).
1292 * Such locations include userret().
1293 * If the "return_instead" argument is non zero, the thread must be able to
1294 * accept 0 (caller may continue), or 1 (caller must abort) as a result.
1296 * The 'return_instead' argument tells the function if it may do a
1297 * thread_exit() or suspend, or whether the caller must abort and back
1300 * If the thread that set the single_threading request has set the
1301 * P_SINGLE_EXIT bit in the process flags then this call will never return
1302 * if 'return_instead' is false, but will exit.
1304 * P_SINGLE_EXIT | return_instead == 0| return_instead != 0
1305 *---------------+--------------------+---------------------
1306 * 0 | returns 0 | returns 0 or 1
1307 * | when ST ends | immediately
1308 *---------------+--------------------+---------------------
1309 * 1 | thread exits | returns 1
1311 * 0 = thread_exit() or suspension ok,
1312 * other = return error instead of stopping the thread.
1314 * While a full suspension is under effect, even a single threading
1315 * thread would be suspended if it made this call (but it shouldn't).
1316 * This call should only be made from places where
1317 * thread_exit() would be safe as that may be the outcome unless
1318 * return_instead is set.
1321 thread_suspend_check(int return_instead)
1329 mtx_assert(&Giant, MA_NOTOWNED);
1330 PROC_LOCK_ASSERT(p, MA_OWNED);
1331 while (thread_suspend_check_needed()) {
1332 if (P_SHOULDSTOP(p) == P_STOPPED_SINGLE) {
1333 KASSERT(p->p_singlethread != NULL,
1334 ("singlethread not set"));
1336 * The only suspension in action is a
1337 * single-threading. Single threader need not stop.
1338 * It is safe to access p->p_singlethread unlocked
1339 * because it can only be set to our address by us.
1341 if (p->p_singlethread == td)
1342 return (0); /* Exempt from stopping. */
1344 if ((p->p_flag & P_SINGLE_EXIT) && return_instead)
1347 /* Should we goto user boundary if we didn't come from there? */
1348 if (P_SHOULDSTOP(p) == P_STOPPED_SINGLE &&
1349 (p->p_flag & P_SINGLE_BOUNDARY) && return_instead)
1353 * Ignore suspend requests if they are deferred.
1355 if ((td->td_flags & TDF_SBDRY) != 0) {
1356 KASSERT(return_instead,
1357 ("TDF_SBDRY set for unsafe thread_suspend_check"));
1358 KASSERT((td->td_flags & (TDF_SEINTR | TDF_SERESTART)) !=
1359 (TDF_SEINTR | TDF_SERESTART),
1360 ("both TDF_SEINTR and TDF_SERESTART"));
1361 return (TD_SBDRY_INTR(td) ? TD_SBDRY_ERRNO(td) : 0);
1365 * If the process is waiting for us to exit,
1366 * this thread should just suicide.
1367 * Assumes that P_SINGLE_EXIT implies P_STOPPED_SINGLE.
1369 if ((p->p_flag & P_SINGLE_EXIT) && (p->p_singlethread != td)) {
1373 * Allow Linux emulation layer to do some work
1374 * before thread suicide.
1376 if (__predict_false(p->p_sysent->sv_thread_detach != NULL))
1377 (p->p_sysent->sv_thread_detach)(td);
1378 umtx_thread_exit(td);
1380 panic("stopped thread did not exit");
1385 if (P_SHOULDSTOP(p) == P_STOPPED_SINGLE) {
1386 if (p->p_numthreads == p->p_suspcount + 1) {
1387 thread_lock(p->p_singlethread);
1388 wakeup_swapper = thread_unsuspend_one(
1389 p->p_singlethread, p, false);
1397 * When a thread suspends, it just
1398 * gets taken off all queues.
1400 thread_suspend_one(td);
1401 if (return_instead == 0) {
1402 p->p_boundary_count++;
1403 td->td_flags |= TDF_BOUNDARY;
1406 mi_switch(SW_INVOL | SWT_SUSPEND);
1413 * Check for possible stops and suspensions while executing a
1414 * casueword or similar transiently failing operation.
1416 * The sleep argument controls whether the function can handle a stop
1417 * request itself or it should return ERESTART and the request is
1418 * proceed at the kernel/user boundary in ast.
1420 * Typically, when retrying due to casueword(9) failure (rv == 1), we
1421 * should handle the stop requests there, with exception of cases when
1422 * the thread owns a kernel resource, for instance busied the umtx
1423 * key, or when functions return immediately if thread_check_susp()
1424 * returned non-zero. On the other hand, retrying the whole lock
1425 * operation, we better not stop there but delegate the handling to
1428 * If the request is for thread termination P_SINGLE_EXIT, we cannot
1429 * handle it at all, and simply return EINTR.
1432 thread_check_susp(struct thread *td, bool sleep)
1438 * The check for TDF_NEEDSUSPCHK is racy, but it is enough to
1439 * eventually break the lockstep loop.
1441 if ((td->td_flags & TDF_NEEDSUSPCHK) == 0)
1446 if (p->p_flag & P_SINGLE_EXIT)
1448 else if (P_SHOULDSTOP(p) ||
1449 ((p->p_flag & P_TRACED) && (td->td_dbgflags & TDB_SUSPEND)))
1450 error = sleep ? thread_suspend_check(0) : ERESTART;
1456 thread_suspend_switch(struct thread *td, struct proc *p)
1459 KASSERT(!TD_IS_SUSPENDED(td), ("already suspended"));
1460 PROC_LOCK_ASSERT(p, MA_OWNED);
1461 PROC_SLOCK_ASSERT(p, MA_OWNED);
1463 * We implement thread_suspend_one in stages here to avoid
1464 * dropping the proc lock while the thread lock is owned.
1466 if (p == td->td_proc) {
1472 td->td_flags &= ~TDF_NEEDSUSPCHK;
1473 TD_SET_SUSPENDED(td);
1477 mi_switch(SW_VOL | SWT_SUSPEND);
1484 thread_suspend_one(struct thread *td)
1489 PROC_SLOCK_ASSERT(p, MA_OWNED);
1490 THREAD_LOCK_ASSERT(td, MA_OWNED);
1491 KASSERT(!TD_IS_SUSPENDED(td), ("already suspended"));
1493 td->td_flags &= ~TDF_NEEDSUSPCHK;
1494 TD_SET_SUSPENDED(td);
1499 thread_unsuspend_one(struct thread *td, struct proc *p, bool boundary)
1502 THREAD_LOCK_ASSERT(td, MA_OWNED);
1503 KASSERT(TD_IS_SUSPENDED(td), ("Thread not suspended"));
1504 TD_CLR_SUSPENDED(td);
1505 td->td_flags &= ~TDF_ALLPROCSUSP;
1506 if (td->td_proc == p) {
1507 PROC_SLOCK_ASSERT(p, MA_OWNED);
1509 if (boundary && (td->td_flags & TDF_BOUNDARY) != 0) {
1510 td->td_flags &= ~TDF_BOUNDARY;
1511 p->p_boundary_count--;
1514 return (setrunnable(td, 0));
1518 * Allow all threads blocked by single threading to continue running.
1521 thread_unsuspend(struct proc *p)
1526 PROC_LOCK_ASSERT(p, MA_OWNED);
1527 PROC_SLOCK_ASSERT(p, MA_OWNED);
1529 if (!P_SHOULDSTOP(p)) {
1530 FOREACH_THREAD_IN_PROC(p, td) {
1532 if (TD_IS_SUSPENDED(td)) {
1533 wakeup_swapper |= thread_unsuspend_one(td, p,
1538 } else if (P_SHOULDSTOP(p) == P_STOPPED_SINGLE &&
1539 p->p_numthreads == p->p_suspcount) {
1541 * Stopping everything also did the job for the single
1542 * threading request. Now we've downgraded to single-threaded,
1545 if (p->p_singlethread->td_proc == p) {
1546 thread_lock(p->p_singlethread);
1547 wakeup_swapper = thread_unsuspend_one(
1548 p->p_singlethread, p, false);
1556 * End the single threading mode..
1559 thread_single_end(struct proc *p, int mode)
1564 KASSERT(mode == SINGLE_EXIT || mode == SINGLE_BOUNDARY ||
1565 mode == SINGLE_ALLPROC || mode == SINGLE_NO_EXIT,
1566 ("invalid mode %d", mode));
1567 PROC_LOCK_ASSERT(p, MA_OWNED);
1568 KASSERT((mode == SINGLE_ALLPROC && (p->p_flag & P_TOTAL_STOP) != 0) ||
1569 (mode != SINGLE_ALLPROC && (p->p_flag & P_TOTAL_STOP) == 0),
1570 ("mode %d does not match P_TOTAL_STOP", mode));
1571 KASSERT(mode == SINGLE_ALLPROC || p->p_singlethread == curthread,
1572 ("thread_single_end from other thread %p %p",
1573 curthread, p->p_singlethread));
1574 KASSERT(mode != SINGLE_BOUNDARY ||
1575 (p->p_flag & P_SINGLE_BOUNDARY) != 0,
1576 ("mis-matched SINGLE_BOUNDARY flags %x", p->p_flag));
1577 p->p_flag &= ~(P_STOPPED_SINGLE | P_SINGLE_EXIT | P_SINGLE_BOUNDARY |
1580 p->p_singlethread = NULL;
1583 * If there are other threads they may now run,
1584 * unless of course there is a blanket 'stop order'
1585 * on the process. The single threader must be allowed
1586 * to continue however as this is a bad place to stop.
1588 if (p->p_numthreads != remain_for_mode(mode) && !P_SHOULDSTOP(p)) {
1589 FOREACH_THREAD_IN_PROC(p, td) {
1591 if (TD_IS_SUSPENDED(td)) {
1592 wakeup_swapper |= thread_unsuspend_one(td, p,
1593 mode == SINGLE_BOUNDARY);
1598 KASSERT(mode != SINGLE_BOUNDARY || p->p_boundary_count == 0,
1599 ("inconsistent boundary count %d", p->p_boundary_count));
1606 * Locate a thread by number and return with proc lock held.
1608 * thread exit establishes proc -> tidhash lock ordering, but lookup
1609 * takes tidhash first and needs to return locked proc.
1611 * The problem is worked around by relying on type-safety of both
1612 * structures and doing the work in 2 steps:
1613 * - tidhash-locked lookup which saves both thread and proc pointers
1614 * - proc-locked verification that the found thread still matches
1617 tdfind_hash(lwpid_t tid, pid_t pid, struct proc **pp, struct thread **tdp)
1619 #define RUN_THRESH 16
1626 rw_rlock(TIDHASHLOCK(tid));
1628 LIST_FOREACH(td, TIDHASH(tid), td_hash) {
1629 if (td->td_tid != tid) {
1634 if (pid != -1 && p->p_pid != pid) {
1638 if (run > RUN_THRESH) {
1639 if (rw_try_upgrade(TIDHASHLOCK(tid))) {
1640 LIST_REMOVE(td, td_hash);
1641 LIST_INSERT_HEAD(TIDHASH(td->td_tid),
1643 rw_wunlock(TIDHASHLOCK(tid));
1651 rw_runlock(TIDHASHLOCK(tid));
1660 tdfind(lwpid_t tid, pid_t pid)
1666 if (td->td_tid == tid) {
1667 if (pid != -1 && td->td_proc->p_pid != pid)
1669 PROC_LOCK(td->td_proc);
1674 if (!tdfind_hash(tid, pid, &p, &td))
1677 if (td->td_tid != tid) {
1681 if (td->td_proc != p) {
1685 if (p->p_state == PRS_NEW) {
1694 tidhash_add(struct thread *td)
1696 rw_wlock(TIDHASHLOCK(td->td_tid));
1697 LIST_INSERT_HEAD(TIDHASH(td->td_tid), td, td_hash);
1698 rw_wunlock(TIDHASHLOCK(td->td_tid));
1702 tidhash_remove(struct thread *td)
1705 rw_wlock(TIDHASHLOCK(td->td_tid));
1706 LIST_REMOVE(td, td_hash);
1707 rw_wunlock(TIDHASHLOCK(td->td_tid));