2 * Copyright (c) 2003, Jeffrey Roberson <jeff@freebsd.org>
5 * Redistribution and use in source and binary forms, with or without
6 * modification, are permitted provided that the following conditions
8 * 1. Redistributions of source code must retain the above copyright
9 * notice unmodified, this list of conditions, and the following
11 * 2. Redistributions in binary form must reproduce the above copyright
12 * notice, this list of conditions and the following disclaimer in the
13 * documentation and/or other materials provided with the distribution.
15 * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR
16 * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES
17 * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.
18 * IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT,
19 * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT
20 * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
21 * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
22 * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
23 * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF
24 * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
27 #include <sys/cdefs.h>
28 __FBSDID("$FreeBSD$");
30 #include "opt_compat.h"
31 #include "opt_posix.h"
32 #include <sys/param.h>
33 #include <sys/kernel.h>
35 #include <sys/mutex.h>
38 #include <sys/posix4.h>
39 #include <sys/racct.h>
40 #include <sys/resourcevar.h>
41 #include <sys/rwlock.h>
42 #include <sys/sched.h>
43 #include <sys/sysctl.h>
45 #include <sys/syscallsubr.h>
46 #include <sys/sysent.h>
47 #include <sys/systm.h>
48 #include <sys/sysproto.h>
49 #include <sys/signalvar.h>
50 #include <sys/sysctl.h>
51 #include <sys/ucontext.h>
53 #include <sys/rtprio.h>
55 #include <sys/limits.h>
57 #include <machine/frame.h>
59 #include <security/audit/audit.h>
61 static SYSCTL_NODE(_kern, OID_AUTO, threads, CTLFLAG_RW, 0,
64 static int max_threads_per_proc = 1500;
65 SYSCTL_INT(_kern_threads, OID_AUTO, max_threads_per_proc, CTLFLAG_RW,
66 &max_threads_per_proc, 0, "Limit on threads per proc");
68 static int max_threads_hits;
69 SYSCTL_INT(_kern_threads, OID_AUTO, max_threads_hits, CTLFLAG_RD,
70 &max_threads_hits, 0, "kern.threads.max_threads_per_proc hit count");
72 #ifdef COMPAT_FREEBSD32
75 suword_lwpid(void *addr, lwpid_t lwpid)
79 if (SV_CURPROC_FLAG(SV_LP64))
80 error = suword(addr, lwpid);
82 error = suword32(addr, lwpid);
87 #define suword_lwpid suword
90 static int create_thread(struct thread *td, mcontext_t *ctx,
91 void (*start_func)(void *), void *arg,
92 char *stack_base, size_t stack_size,
94 long *child_tid, long *parent_tid,
95 int flags, struct rtprio *rtp);
98 * System call interface.
101 sys_thr_create(struct thread *td, struct thr_create_args *uap)
102 /* ucontext_t *ctx, long *id, int flags */
107 if ((error = copyin(uap->ctx, &ctx, sizeof(ctx))))
110 error = create_thread(td, &ctx.uc_mcontext, NULL, NULL,
111 NULL, 0, NULL, uap->id, NULL, uap->flags, NULL);
116 sys_thr_new(struct thread *td, struct thr_new_args *uap)
117 /* struct thr_param * */
119 struct thr_param param;
122 if (uap->param_size < 0 || uap->param_size > sizeof(param))
124 bzero(¶m, sizeof(param));
125 if ((error = copyin(uap->param, ¶m, uap->param_size)))
127 return (kern_thr_new(td, ¶m));
131 kern_thr_new(struct thread *td, struct thr_param *param)
133 struct rtprio rtp, *rtpp;
137 if (param->rtp != 0) {
138 error = copyin(param->rtp, &rtp, sizeof(struct rtprio));
143 error = create_thread(td, NULL, param->start_func, param->arg,
144 param->stack_base, param->stack_size, param->tls_base,
145 param->child_tid, param->parent_tid, param->flags,
151 create_thread(struct thread *td, mcontext_t *ctx,
152 void (*start_func)(void *), void *arg,
153 char *stack_base, size_t stack_size,
155 long *child_tid, long *parent_tid,
156 int flags, struct rtprio *rtp)
159 struct thread *newtd;
165 /* Have race condition but it is cheap. */
166 if (p->p_numthreads >= max_threads_per_proc) {
173 case RTP_PRIO_REALTIME:
175 /* Only root can set scheduler policy */
176 if (priv_check(td, PRIV_SCHED_SETPOLICY) != 0)
178 if (rtp->prio > RTP_PRIO_MAX)
181 case RTP_PRIO_NORMAL:
190 PROC_LOCK(td->td_proc);
191 error = racct_add(p, RACCT_NTHR, 1);
192 PROC_UNLOCK(td->td_proc);
197 /* Initialize our td */
198 newtd = thread_alloc(0);
204 cpu_set_upcall(newtd, td);
207 * Try the copyout as soon as we allocate the td so we don't
208 * have to tear things down in a failure case below.
209 * Here we copy out tid to two places, one for child and one
210 * for parent, because pthread can create a detached thread,
211 * if parent wants to safely access child tid, it has to provide
212 * its storage, because child thread may exit quickly and
213 * memory is freed before parent thread can access it.
215 if ((child_tid != NULL &&
216 suword_lwpid(child_tid, newtd->td_tid)) ||
217 (parent_tid != NULL &&
218 suword_lwpid(parent_tid, newtd->td_tid))) {
224 bzero(&newtd->td_startzero,
225 __rangeof(struct thread, td_startzero, td_endzero));
227 bcopy(&td->td_startcopy, &newtd->td_startcopy,
228 __rangeof(struct thread, td_startcopy, td_endcopy));
229 newtd->td_proc = td->td_proc;
230 newtd->td_ucred = crhold(td->td_ucred);
232 if (ctx != NULL) { /* old way to set user context */
233 error = set_mcontext(newtd, ctx);
236 crfree(td->td_ucred);
240 /* Set up our machine context. */
241 stack.ss_sp = stack_base;
242 stack.ss_size = stack_size;
243 /* Set upcall address to user thread entry function. */
244 cpu_set_upcall_kse(newtd, start_func, arg, &stack);
245 /* Setup user TLS address and TLS pointer register. */
246 error = cpu_set_user_tls(newtd, tls_base);
249 crfree(td->td_ucred);
254 PROC_LOCK(td->td_proc);
255 td->td_proc->p_flag |= P_HADTHREADS;
256 thread_link(newtd, p);
257 bcopy(p->p_comm, newtd->td_name, sizeof(newtd->td_name));
259 /* let the scheduler know about these things. */
260 sched_fork_thread(td, newtd);
263 newtd->td_flags |= TDF_ASTPENDING | TDF_NEEDSUSPCHK;
270 if (!(td->td_pri_class == PRI_TIMESHARE &&
271 rtp->type == RTP_PRIO_NORMAL)) {
272 rtp_to_pri(rtp, newtd);
273 sched_prio(newtd, newtd->td_user_pri);
274 } /* ignore timesharing class */
276 TD_SET_CAN_RUN(newtd);
277 sched_add(newtd, SRQ_BORING);
278 thread_unlock(newtd);
286 racct_sub(p, RACCT_NTHR, 1);
294 sys_thr_self(struct thread *td, struct thr_self_args *uap)
299 error = suword_lwpid(uap->id, (unsigned)td->td_tid);
306 sys_thr_exit(struct thread *td, struct thr_exit_args *uap)
313 /* Signal userland that it can free the stack. */
314 if ((void *)uap->state != NULL) {
315 suword_lwpid(uap->state, 1);
316 kern_umtx_wake(td, uap->state, INT_MAX, 0);
319 rw_wlock(&tidhash_lock);
323 if (p->p_numthreads != 1) {
324 racct_sub(p, RACCT_NTHR, 1);
325 LIST_REMOVE(td, td_hash);
326 rw_wunlock(&tidhash_lock);
328 umtx_thread_exit(td);
336 * Ignore attempts to shut down last thread in the proc. This
337 * will actually call _exit(2) in the usermode trampoline when
341 rw_wunlock(&tidhash_lock);
346 sys_thr_kill(struct thread *td, struct thr_kill_args *uap)
347 /* long id, int sig */
356 ksi.ksi_signo = uap->sig;
357 ksi.ksi_code = SI_LWP;
358 ksi.ksi_pid = p->p_pid;
359 ksi.ksi_uid = td->td_ucred->cr_ruid;
361 if (uap->sig != 0 && !_SIG_VALID(uap->sig)) {
366 FOREACH_THREAD_IN_PROC(p, ttd) {
371 tdksignal(ttd, uap->sig, &ksi);
378 ttd = tdfind((lwpid_t)uap->id, p->p_pid);
383 else if (!_SIG_VALID(uap->sig))
386 tdksignal(ttd, uap->sig, &ksi);
387 PROC_UNLOCK(ttd->td_proc);
393 sys_thr_kill2(struct thread *td, struct thr_kill2_args *uap)
394 /* pid_t pid, long id, int sig */
401 AUDIT_ARG_SIGNUM(uap->sig);
404 ksi.ksi_signo = uap->sig;
405 ksi.ksi_code = SI_LWP;
406 ksi.ksi_pid = td->td_proc->p_pid;
407 ksi.ksi_uid = td->td_ucred->cr_ruid;
409 if ((p = pfind(uap->pid)) == NULL)
411 AUDIT_ARG_PROCESS(p);
412 error = p_cansignal(td, p, uap->sig);
417 if (uap->sig != 0 && !_SIG_VALID(uap->sig)) {
421 FOREACH_THREAD_IN_PROC(p, ttd) {
426 tdksignal(ttd, uap->sig, &ksi);
432 ttd = tdfind((lwpid_t)uap->id, uap->pid);
436 AUDIT_ARG_PROCESS(p);
437 error = p_cansignal(td, p, uap->sig);
440 else if (!_SIG_VALID(uap->sig))
443 tdksignal(ttd, uap->sig, &ksi);
450 sys_thr_suspend(struct thread *td, struct thr_suspend_args *uap)
451 /* const struct timespec *timeout */
453 struct timespec ts, *tsp;
457 if (uap->timeout != NULL) {
458 error = umtx_copyin_timeout(uap->timeout, &ts);
464 return (kern_thr_suspend(td, tsp));
468 kern_thr_suspend(struct thread *td, struct timespec *tsp)
470 struct proc *p = td->td_proc;
475 if (td->td_pflags & TDP_WAKEUP) {
476 td->td_pflags &= ~TDP_WAKEUP;
481 if (tsp->tv_sec == 0 && tsp->tv_nsec == 0)
484 TIMESPEC_TO_TIMEVAL(&tv, tsp);
490 if (error == 0 && (td->td_flags & TDF_THRWAKEUP) == 0)
491 error = msleep((void *)td, &p->p_mtx,
492 PCATCH, "lthr", timo);
494 if (td->td_flags & TDF_THRWAKEUP) {
496 td->td_flags &= ~TDF_THRWAKEUP;
502 if (error == EWOULDBLOCK)
504 else if (error == ERESTART) {
512 sys_thr_wake(struct thread *td, struct thr_wake_args *uap)
518 if (uap->id == td->td_tid) {
519 td->td_pflags |= TDP_WAKEUP;
524 ttd = tdfind((lwpid_t)uap->id, p->p_pid);
528 ttd->td_flags |= TDF_THRWAKEUP;
536 sys_thr_set_name(struct thread *td, struct thr_set_name_args *uap)
539 char name[MAXCOMLEN + 1];
545 if (uap->name != NULL) {
546 error = copyinstr(uap->name, name, sizeof(name),
552 ttd = tdfind((lwpid_t)uap->id, p->p_pid);
555 strcpy(ttd->td_name, name);
557 sched_clear_tdname(ttd);