1/*-
2 * SPDX-License-Identifier: BSD-2-Clause-FreeBSD
3 *
4 * Copyright (c) 2004 Tim J. Robbins
5 * Copyright (c) 2002 Doug Rabson
6 * Copyright (c) 2000 Marcel Moolenaar
7 * All rights reserved.
8 *
9 * Redistribution and use in source and binary forms, with or without
10 * modification, are permitted provided that the following conditions
11 * are met:
12 * 1. Redistributions of source code must retain the above copyright
13 *    notice, this list of conditions and the following disclaimer
14 *    in this position and unchanged.
15 * 2. Redistributions in binary form must reproduce the above copyright
16 *    notice, this list of conditions and the following disclaimer in the
17 *    documentation and/or other materials provided with the distribution.
18 *
19 * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR
20 * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES
21 * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.
22 * IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT,
23 * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT
24 * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
25 * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
26 * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
27 * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF
28 * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
29 */
30
31#include <sys/cdefs.h>
32__FBSDID("$FreeBSD$");
33
34#include "opt_compat.h"
35
36#include <sys/param.h>
37#include <sys/systm.h>
38#include <sys/imgact.h>
39#include <sys/ktr.h>
40#include <sys/lock.h>
41#include <sys/mutex.h>
42#include <sys/proc.h>
43#include <sys/ptrace.h>
44#include <sys/racct.h>
45#include <sys/sched.h>
46#include <sys/syscallsubr.h>
47#include <sys/sx.h>
48#include <sys/umtx.h>
49#include <sys/unistd.h>
50#include <sys/wait.h>
51
52#include <vm/vm.h>
53#include <vm/pmap.h>
54#include <vm/vm_map.h>
55
56#ifdef COMPAT_LINUX32
57#include <machine/../linux32/linux.h>
58#include <machine/../linux32/linux32_proto.h>
59#else
60#include <machine/../linux/linux.h>
61#include <machine/../linux/linux_proto.h>
62#endif
63#include <compat/linux/linux_emul.h>
64#include <compat/linux/linux_futex.h>
65#include <compat/linux/linux_misc.h>
66#include <compat/linux/linux_util.h>
67
68#ifdef LINUX_LEGACY_SYSCALLS
69int
70linux_fork(struct thread *td, struct linux_fork_args *args)
71{
72	struct fork_req fr;
73	int error;
74	struct proc *p2;
75	struct thread *td2;
76
77	bzero(&fr, sizeof(fr));
78	fr.fr_flags = RFFDG | RFPROC | RFSTOPPED;
79	fr.fr_procp = &p2;
80	if ((error = fork1(td, &fr)) != 0)
81		return (error);
82
83	td2 = FIRST_THREAD_IN_PROC(p2);
84
85	linux_proc_init(td, td2, 0);
86
87	td->td_retval[0] = p2->p_pid;
88
89	/*
90	 * Make this runnable after we are finished with it.
91	 */
92	thread_lock(td2);
93	TD_SET_CAN_RUN(td2);
94	sched_add(td2, SRQ_BORING);
95	thread_unlock(td2);
96
97	return (0);
98}
99
100int
101linux_vfork(struct thread *td, struct linux_vfork_args *args)
102{
103	struct fork_req fr;
104	int error;
105	struct proc *p2;
106	struct thread *td2;
107
108	bzero(&fr, sizeof(fr));
109	fr.fr_flags = RFFDG | RFPROC | RFMEM | RFPPWAIT | RFSTOPPED;
110	fr.fr_procp = &p2;
111	if ((error = fork1(td, &fr)) != 0)
112		return (error);
113
114	td2 = FIRST_THREAD_IN_PROC(p2);
115
116	linux_proc_init(td, td2, 0);
117
118	td->td_retval[0] = p2->p_pid;
119
120	/*
121	 * Make this runnable after we are finished with it.
122	 */
123	thread_lock(td2);
124	TD_SET_CAN_RUN(td2);
125	sched_add(td2, SRQ_BORING);
126	thread_unlock(td2);
127
128	return (0);
129}
130#endif
131
132static int
133linux_clone_proc(struct thread *td, struct linux_clone_args *args)
134{
135	struct fork_req fr;
136	int error, ff = RFPROC | RFSTOPPED;
137	struct proc *p2;
138	struct thread *td2;
139	int exit_signal;
140	struct linux_emuldata *em;
141
142	exit_signal = args->flags & 0x000000ff;
143	if (LINUX_SIG_VALID(exit_signal)) {
144		exit_signal = linux_to_bsd_signal(exit_signal);
145	} else if (exit_signal != 0)
146		return (EINVAL);
147
148	if (args->flags & LINUX_CLONE_VM)
149		ff |= RFMEM;
150	if (args->flags & LINUX_CLONE_SIGHAND)
151		ff |= RFSIGSHARE;
152	/*
153	 * XXX: In Linux, sharing of fs info (chroot/cwd/umask)
154	 * and open files is independent.  In FreeBSD, its in one
155	 * structure but in reality it does not cause any problems
156	 * because both of these flags are usually set together.
157	 */
158	if (!(args->flags & (LINUX_CLONE_FILES | LINUX_CLONE_FS)))
159		ff |= RFFDG;
160
161	if (args->flags & LINUX_CLONE_PARENT_SETTID)
162		if (args->parent_tidptr == NULL)
163			return (EINVAL);
164
165	if (args->flags & LINUX_CLONE_VFORK)
166		ff |= RFPPWAIT;
167
168	bzero(&fr, sizeof(fr));
169	fr.fr_flags = ff;
170	fr.fr_procp = &p2;
171	error = fork1(td, &fr);
172	if (error)
173		return (error);
174
175	td2 = FIRST_THREAD_IN_PROC(p2);
176
177	/* create the emuldata */
178	linux_proc_init(td, td2, args->flags);
179
180	em = em_find(td2);
181	KASSERT(em != NULL, ("clone_proc: emuldata not found.\n"));
182
183	if (args->flags & LINUX_CLONE_CHILD_SETTID)
184		em->child_set_tid = args->child_tidptr;
185	else
186		em->child_set_tid = NULL;
187
188	if (args->flags & LINUX_CLONE_CHILD_CLEARTID)
189		em->child_clear_tid = args->child_tidptr;
190	else
191		em->child_clear_tid = NULL;
192
193	if (args->flags & LINUX_CLONE_PARENT_SETTID) {
194		error = copyout(&p2->p_pid, args->parent_tidptr,
195		    sizeof(p2->p_pid));
196		if (error)
197			linux_msg(td, "copyout p_pid failed!");
198	}
199
200	PROC_LOCK(p2);
201	p2->p_sigparent = exit_signal;
202	PROC_UNLOCK(p2);
203	/*
204	 * In a case of stack = NULL, we are supposed to COW calling process
205	 * stack. This is what normal fork() does, so we just keep tf_rsp arg
206	 * intact.
207	 */
208	linux_set_upcall_kse(td2, PTROUT(args->stack));
209
210	if (args->flags & LINUX_CLONE_SETTLS)
211		linux_set_cloned_tls(td2, args->tls);
212
213	/*
214	 * If CLONE_PARENT is set, then the parent of the new process will be
215	 * the same as that of the calling process.
216	 */
217	if (args->flags & LINUX_CLONE_PARENT) {
218		sx_xlock(&proctree_lock);
219		PROC_LOCK(p2);
220		proc_reparent(p2, td->td_proc->p_pptr, true);
221		PROC_UNLOCK(p2);
222		sx_xunlock(&proctree_lock);
223	}
224
225	/*
226	 * Make this runnable after we are finished with it.
227	 */
228	thread_lock(td2);
229	TD_SET_CAN_RUN(td2);
230	sched_add(td2, SRQ_BORING);
231	thread_unlock(td2);
232
233	td->td_retval[0] = p2->p_pid;
234
235	return (0);
236}
237
238static int
239linux_clone_thread(struct thread *td, struct linux_clone_args *args)
240{
241	struct linux_emuldata *em;
242	struct thread *newtd;
243	struct proc *p;
244	int error;
245
246	LINUX_CTR4(clone_thread, "thread(%d) flags %x ptid %p ctid %p",
247	    td->td_tid, (unsigned)args->flags,
248	    args->parent_tidptr, args->child_tidptr);
249
250	if ((args->flags & LINUX_CLONE_PARENT) != 0)
251		return (EINVAL);
252	if (args->flags & LINUX_CLONE_PARENT_SETTID)
253		if (args->parent_tidptr == NULL)
254			return (EINVAL);
255
256	/* Threads should be created with own stack */
257	if (args->stack == NULL)
258		return (EINVAL);
259
260	p = td->td_proc;
261
262#ifdef RACCT
263	if (racct_enable) {
264		PROC_LOCK(p);
265		error = racct_add(p, RACCT_NTHR, 1);
266		PROC_UNLOCK(p);
267		if (error != 0)
268			return (EPROCLIM);
269	}
270#endif
271
272	/* Initialize our td */
273	error = kern_thr_alloc(p, 0, &newtd);
274	if (error)
275		goto fail;
276
277	cpu_copy_thread(newtd, td);
278
279	bzero(&newtd->td_startzero,
280	    __rangeof(struct thread, td_startzero, td_endzero));
281	newtd->td_pflags2 = 0;
282	newtd->td_errno = 0;
283	bcopy(&td->td_startcopy, &newtd->td_startcopy,
284	    __rangeof(struct thread, td_startcopy, td_endcopy));
285
286	newtd->td_proc = p;
287	thread_cow_get(newtd, td);
288
289	/* create the emuldata */
290	linux_proc_init(td, newtd, args->flags);
291
292	em = em_find(newtd);
293	KASSERT(em != NULL, ("clone_thread: emuldata not found.\n"));
294
295	if (args->flags & LINUX_CLONE_SETTLS)
296		linux_set_cloned_tls(newtd, args->tls);
297
298	if (args->flags & LINUX_CLONE_CHILD_SETTID)
299		em->child_set_tid = args->child_tidptr;
300	else
301		em->child_set_tid = NULL;
302
303	if (args->flags & LINUX_CLONE_CHILD_CLEARTID)
304		em->child_clear_tid = args->child_tidptr;
305	else
306		em->child_clear_tid = NULL;
307
308	cpu_thread_clean(newtd);
309
310	linux_set_upcall_kse(newtd, PTROUT(args->stack));
311
312	PROC_LOCK(p);
313	p->p_flag |= P_HADTHREADS;
314	thread_link(newtd, p);
315	bcopy(p->p_comm, newtd->td_name, sizeof(newtd->td_name));
316
317	thread_lock(td);
318	/* let the scheduler know about these things. */
319	sched_fork_thread(td, newtd);
320	thread_unlock(td);
321	if (P_SHOULDSTOP(p))
322		newtd->td_flags |= TDF_ASTPENDING | TDF_NEEDSUSPCHK;
323
324	if (p->p_ptevents & PTRACE_LWP)
325		newtd->td_dbgflags |= TDB_BORN;
326	PROC_UNLOCK(p);
327
328	tidhash_add(newtd);
329
330	LINUX_CTR2(clone_thread, "thread(%d) successful clone to %d",
331	    td->td_tid, newtd->td_tid);
332
333	if (args->flags & LINUX_CLONE_PARENT_SETTID) {
334		error = copyout(&newtd->td_tid, args->parent_tidptr,
335		    sizeof(newtd->td_tid));
336		if (error)
337			linux_msg(td, "clone_thread: copyout td_tid failed!");
338	}
339
340	/*
341	 * Make this runnable after we are finished with it.
342	 */
343	thread_lock(newtd);
344	TD_SET_CAN_RUN(newtd);
345	sched_add(newtd, SRQ_BORING);
346	thread_unlock(newtd);
347
348	td->td_retval[0] = newtd->td_tid;
349
350	return (0);
351
352fail:
353#ifdef RACCT
354	if (racct_enable) {
355		PROC_LOCK(p);
356		racct_sub(p, RACCT_NTHR, 1);
357		PROC_UNLOCK(p);
358	}
359#endif
360	return (error);
361}
362
363int
364linux_clone(struct thread *td, struct linux_clone_args *args)
365{
366
367	if (args->flags & LINUX_CLONE_THREAD)
368		return (linux_clone_thread(td, args));
369	else
370		return (linux_clone_proc(td, args));
371}
372
373int
374linux_exit(struct thread *td, struct linux_exit_args *args)
375{
376	struct linux_emuldata *em;
377
378	em = em_find(td);
379	KASSERT(em != NULL, ("exit: emuldata not found.\n"));
380
381	LINUX_CTR2(exit, "thread(%d) (%d)", em->em_tid, args->rval);
382
383	umtx_thread_exit(td);
384
385	linux_thread_detach(td);
386
387	/*
388	 * XXX. When the last two threads of a process
389	 * exit via pthread_exit() try thr_exit() first.
390	 */
391	kern_thr_exit(td);
392	exit1(td, args->rval, 0);
393		/* NOTREACHED */
394}
395
396int
397linux_set_tid_address(struct thread *td, struct linux_set_tid_address_args *args)
398{
399	struct linux_emuldata *em;
400
401	em = em_find(td);
402	KASSERT(em != NULL, ("set_tid_address: emuldata not found.\n"));
403
404	em->child_clear_tid = args->tidptr;
405
406	td->td_retval[0] = em->em_tid;
407
408	LINUX_CTR3(set_tid_address, "tidptr(%d) %p, returns %d",
409	    em->em_tid, args->tidptr, td->td_retval[0]);
410
411	return (0);
412}
413
414void
415linux_thread_detach(struct thread *td)
416{
417	struct linux_sys_futex_args cup;
418	struct linux_emuldata *em;
419	int *child_clear_tid;
420	int error;
421
422	em = em_find(td);
423	KASSERT(em != NULL, ("thread_detach: emuldata not found.\n"));
424
425	LINUX_CTR1(thread_detach, "thread(%d)", em->em_tid);
426
427	release_futexes(td, em);
428
429	child_clear_tid = em->child_clear_tid;
430
431	if (child_clear_tid != NULL) {
432
433		LINUX_CTR2(thread_detach, "thread(%d) %p",
434		    em->em_tid, child_clear_tid);
435
436		error = suword32(child_clear_tid, 0);
437		if (error != 0)
438			return;
439
440		cup.uaddr = child_clear_tid;
441		cup.op = LINUX_FUTEX_WAKE;
442		cup.val = 1;		/* wake one */
443		cup.timeout = NULL;
444		cup.uaddr2 = NULL;
445		cup.val3 = 0;
446		error = linux_sys_futex(td, &cup);
447		/*
448		 * this cannot happen at the moment and if this happens it
449		 * probably means there is a user space bug
450		 */
451		if (error != 0)
452			linux_msg(td, "futex stuff in thread_detach failed.");
453	}
454}
455