xref: /freebsd/sys/security/audit/audit.c (revision f39bffc62c1395bde25d152c7f68fdf7cbaab414)
1 /*-
2  * SPDX-License-Identifier: BSD-3-Clause
3  *
4  * Copyright (c) 1999-2005 Apple Inc.
5  * Copyright (c) 2006-2007, 2016-2017 Robert N. M. Watson
6  * All rights reserved.
7  *
8  * Portions of this software were developed by BAE Systems, the University of
9  * Cambridge Computer Laboratory, and Memorial University under DARPA/AFRL
10  * contract FA8650-15-C-7558 ("CADETS"), as part of the DARPA Transparent
11  * Computing (TC) research program.
12  *
13  * Redistribution and use in source and binary forms, with or without
14  * modification, are permitted provided that the following conditions
15  * are met:
16  * 1.  Redistributions of source code must retain the above copyright
17  *     notice, this list of conditions and the following disclaimer.
18  * 2.  Redistributions in binary form must reproduce the above copyright
19  *     notice, this list of conditions and the following disclaimer in the
20  *     documentation and/or other materials provided with the distribution.
21  * 3.  Neither the name of Apple Inc. ("Apple") nor the names of
22  *     its contributors may be used to endorse or promote products derived
23  *     from this software without specific prior written permission.
24  *
25  * THIS SOFTWARE IS PROVIDED BY APPLE AND ITS CONTRIBUTORS "AS IS" AND
26  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
27  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
28  * ARE DISCLAIMED. IN NO EVENT SHALL APPLE OR ITS CONTRIBUTORS BE LIABLE FOR
29  * ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
30  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
31  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
32  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT,
33  * STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING
34  * IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
35  * POSSIBILITY OF SUCH DAMAGE.
36  */
37 
38 #include <sys/cdefs.h>
39 __FBSDID("$FreeBSD$");
40 
41 #include <sys/param.h>
42 #include <sys/condvar.h>
43 #include <sys/conf.h>
44 #include <sys/file.h>
45 #include <sys/filedesc.h>
46 #include <sys/fcntl.h>
47 #include <sys/ipc.h>
48 #include <sys/jail.h>
49 #include <sys/kernel.h>
50 #include <sys/kthread.h>
51 #include <sys/malloc.h>
52 #include <sys/mount.h>
53 #include <sys/namei.h>
54 #include <sys/priv.h>
55 #include <sys/proc.h>
56 #include <sys/queue.h>
57 #include <sys/socket.h>
58 #include <sys/socketvar.h>
59 #include <sys/protosw.h>
60 #include <sys/domain.h>
61 #include <sys/sysctl.h>
62 #include <sys/sysproto.h>
63 #include <sys/sysent.h>
64 #include <sys/systm.h>
65 #include <sys/ucred.h>
66 #include <sys/uio.h>
67 #include <sys/un.h>
68 #include <sys/unistd.h>
69 #include <sys/vnode.h>
70 
71 #include <bsm/audit.h>
72 #include <bsm/audit_internal.h>
73 #include <bsm/audit_kevents.h>
74 
75 #include <netinet/in.h>
76 #include <netinet/in_pcb.h>
77 
78 #include <security/audit/audit.h>
79 #include <security/audit/audit_private.h>
80 
81 #include <vm/uma.h>
82 
83 FEATURE(audit, "BSM audit support");
84 
85 static uma_zone_t	audit_record_zone;
86 static MALLOC_DEFINE(M_AUDITCRED, "audit_cred", "Audit cred storage");
87 MALLOC_DEFINE(M_AUDITDATA, "audit_data", "Audit data storage");
88 MALLOC_DEFINE(M_AUDITPATH, "audit_path", "Audit path storage");
89 MALLOC_DEFINE(M_AUDITTEXT, "audit_text", "Audit text storage");
90 MALLOC_DEFINE(M_AUDITGIDSET, "audit_gidset", "Audit GID set storage");
91 
92 static SYSCTL_NODE(_security, OID_AUTO, audit, CTLFLAG_RW, 0,
93     "TrustedBSD audit controls");
94 
95 /*
96  * Audit control settings that are set/read by system calls and are hence
97  * non-static.
98  *
99  * Define the audit control flags.
100  */
101 int __read_frequently	audit_enabled;
102 int			audit_suspended;
103 
104 /*
105  * Flags controlling behavior in low storage situations.  Should we panic if
106  * a write fails?  Should we fail stop if we're out of disk space?
107  */
108 int			audit_panic_on_write_fail;
109 int			audit_fail_stop;
110 int			audit_argv;
111 int			audit_arge;
112 
113 /*
114  * Are we currently "failing stop" due to out of disk space?
115  */
116 int			audit_in_failure;
117 
118 /*
119  * Global audit statistics.
120  */
121 struct audit_fstat	audit_fstat;
122 
123 /*
124  * Preselection mask for non-attributable events.
125  */
126 struct au_mask		audit_nae_mask;
127 
128 /*
129  * Mutex to protect global variables shared between various threads and
130  * processes.
131  */
132 struct mtx		audit_mtx;
133 
134 /*
135  * Queue of audit records ready for delivery to disk.  We insert new records
136  * at the tail, and remove records from the head.  Also, a count of the
137  * number of records used for checking queue depth.  In addition, a counter
138  * of records that we have allocated but are not yet in the queue, which is
139  * needed to estimate the total size of the combined set of records
140  * outstanding in the system.
141  */
142 struct kaudit_queue	audit_q;
143 int			audit_q_len;
144 int			audit_pre_q_len;
145 
146 /*
147  * Audit queue control settings (minimum free, low/high water marks, etc.)
148  */
149 struct au_qctrl		audit_qctrl;
150 
151 /*
152  * Condition variable to signal to the worker that it has work to do: either
153  * new records are in the queue, or a log replacement is taking place.
154  */
155 struct cv		audit_worker_cv;
156 
157 /*
158  * Condition variable to flag when crossing the low watermark, meaning that
159  * threads blocked due to hitting the high watermark can wake up and continue
160  * to commit records.
161  */
162 struct cv		audit_watermark_cv;
163 
164 /*
165  * Condition variable for  auditing threads wait on when in fail-stop mode.
166  * Threads wait on this CV forever (and ever), never seeing the light of day
167  * again.
168  */
169 static struct cv	audit_fail_cv;
170 
171 /*
172  * Optional DTrace audit provider support: function pointers for preselection
173  * and commit events.
174  */
175 #ifdef KDTRACE_HOOKS
176 void	*(*dtaudit_hook_preselect)(au_id_t auid, au_event_t event,
177 	    au_class_t class);
178 int	(*dtaudit_hook_commit)(struct kaudit_record *kar, au_id_t auid,
179 	    au_event_t event, au_class_t class, int sorf);
180 void	(*dtaudit_hook_bsm)(struct kaudit_record *kar, au_id_t auid,
181 	    au_event_t event, au_class_t class, int sorf,
182 	    void *bsm_data, size_t bsm_lenlen);
183 #endif
184 
185 /*
186  * Kernel audit information.  This will store the current audit address
187  * or host information that the kernel will use when it's generating
188  * audit records.  This data is modified by the A_GET{SET}KAUDIT auditon(2)
189  * command.
190  */
191 static struct auditinfo_addr	audit_kinfo;
192 static struct rwlock		audit_kinfo_lock;
193 
194 #define	KINFO_LOCK_INIT()	rw_init(&audit_kinfo_lock, \
195 				    "audit_kinfo_lock")
196 #define	KINFO_RLOCK()		rw_rlock(&audit_kinfo_lock)
197 #define	KINFO_WLOCK()		rw_wlock(&audit_kinfo_lock)
198 #define	KINFO_RUNLOCK()		rw_runlock(&audit_kinfo_lock)
199 #define	KINFO_WUNLOCK()		rw_wunlock(&audit_kinfo_lock)
200 
201 void
202 audit_set_kinfo(struct auditinfo_addr *ak)
203 {
204 
205 	KASSERT(ak->ai_termid.at_type == AU_IPv4 ||
206 	    ak->ai_termid.at_type == AU_IPv6,
207 	    ("audit_set_kinfo: invalid address type"));
208 
209 	KINFO_WLOCK();
210 	audit_kinfo = *ak;
211 	KINFO_WUNLOCK();
212 }
213 
214 void
215 audit_get_kinfo(struct auditinfo_addr *ak)
216 {
217 
218 	KASSERT(audit_kinfo.ai_termid.at_type == AU_IPv4 ||
219 	    audit_kinfo.ai_termid.at_type == AU_IPv6,
220 	    ("audit_set_kinfo: invalid address type"));
221 
222 	KINFO_RLOCK();
223 	*ak = audit_kinfo;
224 	KINFO_RUNLOCK();
225 }
226 
227 /*
228  * Construct an audit record for the passed thread.
229  */
230 static int
231 audit_record_ctor(void *mem, int size, void *arg, int flags)
232 {
233 	struct kaudit_record *ar;
234 	struct thread *td;
235 	struct ucred *cred;
236 	struct prison *pr;
237 
238 	KASSERT(sizeof(*ar) == size, ("audit_record_ctor: wrong size"));
239 
240 	td = arg;
241 	ar = mem;
242 	bzero(ar, sizeof(*ar));
243 	ar->k_ar.ar_magic = AUDIT_RECORD_MAGIC;
244 	nanotime(&ar->k_ar.ar_starttime);
245 
246 	/*
247 	 * Export the subject credential.
248 	 */
249 	cred = td->td_ucred;
250 	cru2x(cred, &ar->k_ar.ar_subj_cred);
251 	ar->k_ar.ar_subj_ruid = cred->cr_ruid;
252 	ar->k_ar.ar_subj_rgid = cred->cr_rgid;
253 	ar->k_ar.ar_subj_egid = cred->cr_groups[0];
254 	ar->k_ar.ar_subj_auid = cred->cr_audit.ai_auid;
255 	ar->k_ar.ar_subj_asid = cred->cr_audit.ai_asid;
256 	ar->k_ar.ar_subj_pid = td->td_proc->p_pid;
257 	ar->k_ar.ar_subj_amask = cred->cr_audit.ai_mask;
258 	ar->k_ar.ar_subj_term_addr = cred->cr_audit.ai_termid;
259 	/*
260 	 * If this process is jailed, make sure we capture the name of the
261 	 * jail so we can use it to generate a zonename token when we covert
262 	 * this record to BSM.
263 	 */
264 	if (jailed(cred)) {
265 		pr = cred->cr_prison;
266 		(void) strlcpy(ar->k_ar.ar_jailname, pr->pr_name,
267 		    sizeof(ar->k_ar.ar_jailname));
268 	} else
269 		ar->k_ar.ar_jailname[0] = '\0';
270 	return (0);
271 }
272 
273 static void
274 audit_record_dtor(void *mem, int size, void *arg)
275 {
276 	struct kaudit_record *ar;
277 
278 	KASSERT(sizeof(*ar) == size, ("audit_record_dtor: wrong size"));
279 
280 	ar = mem;
281 	if (ar->k_ar.ar_arg_upath1 != NULL)
282 		free(ar->k_ar.ar_arg_upath1, M_AUDITPATH);
283 	if (ar->k_ar.ar_arg_upath2 != NULL)
284 		free(ar->k_ar.ar_arg_upath2, M_AUDITPATH);
285 	if (ar->k_ar.ar_arg_text != NULL)
286 		free(ar->k_ar.ar_arg_text, M_AUDITTEXT);
287 	if (ar->k_udata != NULL)
288 		free(ar->k_udata, M_AUDITDATA);
289 	if (ar->k_ar.ar_arg_argv != NULL)
290 		free(ar->k_ar.ar_arg_argv, M_AUDITTEXT);
291 	if (ar->k_ar.ar_arg_envv != NULL)
292 		free(ar->k_ar.ar_arg_envv, M_AUDITTEXT);
293 	if (ar->k_ar.ar_arg_groups.gidset != NULL)
294 		free(ar->k_ar.ar_arg_groups.gidset, M_AUDITGIDSET);
295 }
296 
297 /*
298  * Initialize the Audit subsystem: configuration state, work queue,
299  * synchronization primitives, worker thread, and trigger device node.  Also
300  * call into the BSM assembly code to initialize it.
301  */
302 static void
303 audit_init(void)
304 {
305 
306 	audit_enabled = 0;
307 	audit_suspended = 0;
308 	audit_panic_on_write_fail = 0;
309 	audit_fail_stop = 0;
310 	audit_in_failure = 0;
311 	audit_argv = 0;
312 	audit_arge = 0;
313 
314 	audit_fstat.af_filesz = 0;	/* '0' means unset, unbounded. */
315 	audit_fstat.af_currsz = 0;
316 	audit_nae_mask.am_success = 0;
317 	audit_nae_mask.am_failure = 0;
318 
319 	TAILQ_INIT(&audit_q);
320 	audit_q_len = 0;
321 	audit_pre_q_len = 0;
322 	audit_qctrl.aq_hiwater = AQ_HIWATER;
323 	audit_qctrl.aq_lowater = AQ_LOWATER;
324 	audit_qctrl.aq_bufsz = AQ_BUFSZ;
325 	audit_qctrl.aq_minfree = AU_FS_MINFREE;
326 
327 	audit_kinfo.ai_termid.at_type = AU_IPv4;
328 	audit_kinfo.ai_termid.at_addr[0] = INADDR_ANY;
329 
330 	mtx_init(&audit_mtx, "audit_mtx", NULL, MTX_DEF);
331 	KINFO_LOCK_INIT();
332 	cv_init(&audit_worker_cv, "audit_worker_cv");
333 	cv_init(&audit_watermark_cv, "audit_watermark_cv");
334 	cv_init(&audit_fail_cv, "audit_fail_cv");
335 
336 	audit_record_zone = uma_zcreate("audit_record",
337 	    sizeof(struct kaudit_record), audit_record_ctor,
338 	    audit_record_dtor, NULL, NULL, UMA_ALIGN_PTR, 0);
339 
340 	/* Initialize the BSM audit subsystem. */
341 	kau_init();
342 
343 	audit_trigger_init();
344 
345 	/* Register shutdown handler. */
346 	EVENTHANDLER_REGISTER(shutdown_pre_sync, audit_shutdown, NULL,
347 	    SHUTDOWN_PRI_FIRST);
348 
349 	/* Start audit worker thread. */
350 	audit_worker_init();
351 }
352 
353 SYSINIT(audit_init, SI_SUB_AUDIT, SI_ORDER_FIRST, audit_init, NULL);
354 
355 /*
356  * Drain the audit queue and close the log at shutdown.  Note that this can
357  * be called both from the system shutdown path and also from audit
358  * configuration syscalls, so 'arg' and 'howto' are ignored.
359  *
360  * XXXRW: In FreeBSD 7.x and 8.x, this fails to wait for the record queue to
361  * drain before returning, which could lead to lost records on shutdown.
362  */
363 void
364 audit_shutdown(void *arg, int howto)
365 {
366 
367 	audit_rotate_vnode(NULL, NULL);
368 }
369 
370 /*
371  * Return the current thread's audit record, if any.
372  */
373 struct kaudit_record *
374 currecord(void)
375 {
376 
377 	return (curthread->td_ar);
378 }
379 
380 /*
381  * XXXAUDIT: There are a number of races present in the code below due to
382  * release and re-grab of the mutex.  The code should be revised to become
383  * slightly less racy.
384  *
385  * XXXAUDIT: Shouldn't there be logic here to sleep waiting on available
386  * pre_q space, suspending the system call until there is room?
387  */
388 struct kaudit_record *
389 audit_new(int event, struct thread *td)
390 {
391 	struct kaudit_record *ar;
392 	int no_record;
393 
394 	mtx_lock(&audit_mtx);
395 	no_record = (audit_suspended || !audit_enabled);
396 	mtx_unlock(&audit_mtx);
397 	if (no_record)
398 		return (NULL);
399 
400 	/*
401 	 * Note: the number of outstanding uncommitted audit records is
402 	 * limited to the number of concurrent threads servicing system calls
403 	 * in the kernel.
404 	 */
405 	ar = uma_zalloc_arg(audit_record_zone, td, M_WAITOK);
406 	ar->k_ar.ar_event = event;
407 
408 	mtx_lock(&audit_mtx);
409 	audit_pre_q_len++;
410 	mtx_unlock(&audit_mtx);
411 
412 	return (ar);
413 }
414 
415 void
416 audit_free(struct kaudit_record *ar)
417 {
418 
419 	uma_zfree(audit_record_zone, ar);
420 }
421 
422 void
423 audit_commit(struct kaudit_record *ar, int error, int retval)
424 {
425 	au_event_t event;
426 	au_class_t class;
427 	au_id_t auid;
428 	int sorf;
429 	struct au_mask *aumask;
430 
431 	if (ar == NULL)
432 		return;
433 
434 	ar->k_ar.ar_errno = error;
435 	ar->k_ar.ar_retval = retval;
436 	nanotime(&ar->k_ar.ar_endtime);
437 
438 	/*
439 	 * Decide whether to commit the audit record by checking the error
440 	 * value from the system call and using the appropriate audit mask.
441 	 */
442 	if (ar->k_ar.ar_subj_auid == AU_DEFAUDITID)
443 		aumask = &audit_nae_mask;
444 	else
445 		aumask = &ar->k_ar.ar_subj_amask;
446 
447 	if (error)
448 		sorf = AU_PRS_FAILURE;
449 	else
450 		sorf = AU_PRS_SUCCESS;
451 
452 	/*
453 	 * syscalls.master sometimes contains a prototype event number, which
454 	 * we will transform into a more specific event number now that we
455 	 * have more complete information gathered during the system call.
456 	 */
457 	switch(ar->k_ar.ar_event) {
458 	case AUE_OPEN_RWTC:
459 		ar->k_ar.ar_event = audit_flags_and_error_to_openevent(
460 		    ar->k_ar.ar_arg_fflags, error);
461 		break;
462 
463 	case AUE_OPENAT_RWTC:
464 		ar->k_ar.ar_event = audit_flags_and_error_to_openatevent(
465 		    ar->k_ar.ar_arg_fflags, error);
466 		break;
467 
468 	case AUE_SYSCTL:
469 		ar->k_ar.ar_event = audit_ctlname_to_sysctlevent(
470 		    ar->k_ar.ar_arg_ctlname, ar->k_ar.ar_valid_arg);
471 		break;
472 
473 	case AUE_AUDITON:
474 		/* Convert the auditon() command to an event. */
475 		ar->k_ar.ar_event = auditon_command_event(ar->k_ar.ar_arg_cmd);
476 		break;
477 
478 	case AUE_MSGSYS:
479 		if (ARG_IS_VALID(ar, ARG_SVIPC_WHICH))
480 			ar->k_ar.ar_event =
481 			    audit_msgsys_to_event(ar->k_ar.ar_arg_svipc_which);
482 		break;
483 
484 	case AUE_SEMSYS:
485 		if (ARG_IS_VALID(ar, ARG_SVIPC_WHICH))
486 			ar->k_ar.ar_event =
487 			    audit_semsys_to_event(ar->k_ar.ar_arg_svipc_which);
488 		break;
489 
490 	case AUE_SHMSYS:
491 		if (ARG_IS_VALID(ar, ARG_SVIPC_WHICH))
492 			ar->k_ar.ar_event =
493 			    audit_shmsys_to_event(ar->k_ar.ar_arg_svipc_which);
494 		break;
495 	}
496 
497 	auid = ar->k_ar.ar_subj_auid;
498 	event = ar->k_ar.ar_event;
499 	class = au_event_class(event);
500 
501 	ar->k_ar_commit |= AR_COMMIT_KERNEL;
502 	if (au_preselect(event, class, aumask, sorf) != 0)
503 		ar->k_ar_commit |= AR_PRESELECT_TRAIL;
504 	if (audit_pipe_preselect(auid, event, class, sorf,
505 	    ar->k_ar_commit & AR_PRESELECT_TRAIL) != 0)
506 		ar->k_ar_commit |= AR_PRESELECT_PIPE;
507 #ifdef KDTRACE_HOOKS
508 	/*
509 	 * Expose the audit record to DTrace, both to allow the "commit" probe
510 	 * to fire if it's desirable, and also to allow a decision to be made
511 	 * about later firing with BSM in the audit worker.
512 	 */
513 	if (dtaudit_hook_commit != NULL) {
514 		if (dtaudit_hook_commit(ar, auid, event, class, sorf) != 0)
515 			ar->k_ar_commit |= AR_PRESELECT_DTRACE;
516 	}
517 #endif
518 
519 	if ((ar->k_ar_commit & (AR_PRESELECT_TRAIL | AR_PRESELECT_PIPE |
520 	    AR_PRESELECT_USER_TRAIL | AR_PRESELECT_USER_PIPE |
521 	    AR_PRESELECT_DTRACE)) == 0) {
522 		mtx_lock(&audit_mtx);
523 		audit_pre_q_len--;
524 		mtx_unlock(&audit_mtx);
525 		audit_free(ar);
526 		return;
527 	}
528 
529 	/*
530 	 * Note: it could be that some records initiated while audit was
531 	 * enabled should still be committed?
532 	 */
533 	mtx_lock(&audit_mtx);
534 	if (audit_suspended || !audit_enabled) {
535 		audit_pre_q_len--;
536 		mtx_unlock(&audit_mtx);
537 		audit_free(ar);
538 		return;
539 	}
540 
541 	/*
542 	 * Constrain the number of committed audit records based on the
543 	 * configurable parameter.
544 	 */
545 	while (audit_q_len >= audit_qctrl.aq_hiwater)
546 		cv_wait(&audit_watermark_cv, &audit_mtx);
547 
548 	TAILQ_INSERT_TAIL(&audit_q, ar, k_q);
549 	audit_q_len++;
550 	audit_pre_q_len--;
551 	cv_signal(&audit_worker_cv);
552 	mtx_unlock(&audit_mtx);
553 }
554 
555 /*
556  * audit_syscall_enter() is called on entry to each system call.  It is
557  * responsible for deciding whether or not to audit the call (preselection),
558  * and if so, allocating a per-thread audit record.  audit_new() will fill in
559  * basic thread/credential properties.
560  */
561 void
562 audit_syscall_enter(unsigned short code, struct thread *td)
563 {
564 	struct au_mask *aumask;
565 #ifdef KDTRACE_HOOKS
566 	void *dtaudit_state;
567 #endif
568 	au_class_t class;
569 	au_event_t event;
570 	au_id_t auid;
571 	int record_needed;
572 
573 	KASSERT(td->td_ar == NULL, ("audit_syscall_enter: td->td_ar != NULL"));
574 	KASSERT((td->td_pflags & TDP_AUDITREC) == 0,
575 	    ("audit_syscall_enter: TDP_AUDITREC set"));
576 
577 	/*
578 	 * In FreeBSD, each ABI has its own system call table, and hence
579 	 * mapping of system call codes to audit events.  Convert the code to
580 	 * an audit event identifier using the process system call table
581 	 * reference.  In Darwin, there's only one, so we use the global
582 	 * symbol for the system call table.  No audit record is generated
583 	 * for bad system calls, as no operation has been performed.
584 	 */
585 	if (code >= td->td_proc->p_sysent->sv_size)
586 		return;
587 
588 	event = td->td_proc->p_sysent->sv_table[code].sy_auevent;
589 	if (event == AUE_NULL)
590 		return;
591 
592 	/*
593 	 * Check which audit mask to use; either the kernel non-attributable
594 	 * event mask or the process audit mask.
595 	 */
596 	auid = td->td_ucred->cr_audit.ai_auid;
597 	if (auid == AU_DEFAUDITID)
598 		aumask = &audit_nae_mask;
599 	else
600 		aumask = &td->td_ucred->cr_audit.ai_mask;
601 
602 	/*
603 	 * Determine whether trail or pipe preselection would like an audit
604 	 * record allocated for this system call.
605 	 */
606 	class = au_event_class(event);
607 	if (au_preselect(event, class, aumask, AU_PRS_BOTH)) {
608 		/*
609 		 * If we're out of space and need to suspend unprivileged
610 		 * processes, do that here rather than trying to allocate
611 		 * another audit record.
612 		 *
613 		 * Note: we might wish to be able to continue here in the
614 		 * future, if the system recovers.  That should be possible
615 		 * by means of checking the condition in a loop around
616 		 * cv_wait().  It might be desirable to reevaluate whether an
617 		 * audit record is still required for this event by
618 		 * re-calling au_preselect().
619 		 */
620 		if (audit_in_failure &&
621 		    priv_check(td, PRIV_AUDIT_FAILSTOP) != 0) {
622 			cv_wait(&audit_fail_cv, &audit_mtx);
623 			panic("audit_failing_stop: thread continued");
624 		}
625 		record_needed = 1;
626 	} else if (audit_pipe_preselect(auid, event, class, AU_PRS_BOTH, 0)) {
627 		record_needed = 1;
628 	} else {
629 		record_needed = 0;
630 	}
631 
632 	/*
633 	 * After audit trails and pipes have made their policy choices, DTrace
634 	 * may request that records be generated as well.  This is a slightly
635 	 * complex affair, as the DTrace audit provider needs the audit
636 	 * framework to maintain some state on the audit record, which has not
637 	 * been allocated at the point where the decision has to be made.
638 	 * This hook must run even if we are not changing the decision, as
639 	 * DTrace may want to stick event state onto a record we were going to
640 	 * produce due to the trail or pipes.  The event state returned by the
641 	 * DTrace provider must be safe without locks held between here and
642 	 * below -- i.e., dtaudit_state must must refer to stable memory.
643 	 */
644 #ifdef KDTRACE_HOOKS
645 	dtaudit_state = NULL;
646         if (dtaudit_hook_preselect != NULL) {
647 		dtaudit_state = dtaudit_hook_preselect(auid, event, class);
648 		if (dtaudit_state != NULL)
649 			record_needed = 1;
650 	}
651 #endif
652 
653 	/*
654 	 * If a record is required, allocate it and attach it to the thread
655 	 * for use throughout the system call.  Also attach DTrace state if
656 	 * required.
657 	 *
658 	 * XXXRW: If we decide to reference count the evname_elem underlying
659 	 * dtaudit_state, we will need to free here if no record is allocated
660 	 * or allocatable.
661 	 */
662 	if (record_needed) {
663 		td->td_ar = audit_new(event, td);
664 		if (td->td_ar != NULL) {
665 			td->td_pflags |= TDP_AUDITREC;
666 #ifdef KDTRACE_HOOKS
667 			td->td_ar->k_dtaudit_state = dtaudit_state;
668 #endif
669 		}
670 	} else
671 		td->td_ar = NULL;
672 }
673 
674 /*
675  * audit_syscall_exit() is called from the return of every system call, or in
676  * the event of exit1(), during the execution of exit1().  It is responsible
677  * for committing the audit record, if any, along with return condition.
678  */
679 void
680 audit_syscall_exit(int error, struct thread *td)
681 {
682 	int retval;
683 
684 	/*
685 	 * Commit the audit record as desired; once we pass the record into
686 	 * audit_commit(), the memory is owned by the audit subsystem.  The
687 	 * return value from the system call is stored on the user thread.
688 	 * If there was an error, the return value is set to -1, imitating
689 	 * the behavior of the cerror routine.
690 	 */
691 	if (error)
692 		retval = -1;
693 	else
694 		retval = td->td_retval[0];
695 
696 	audit_commit(td->td_ar, error, retval);
697 	td->td_ar = NULL;
698 	td->td_pflags &= ~TDP_AUDITREC;
699 }
700 
701 void
702 audit_cred_copy(struct ucred *src, struct ucred *dest)
703 {
704 
705 	bcopy(&src->cr_audit, &dest->cr_audit, sizeof(dest->cr_audit));
706 }
707 
708 void
709 audit_cred_destroy(struct ucred *cred)
710 {
711 
712 }
713 
714 void
715 audit_cred_init(struct ucred *cred)
716 {
717 
718 	bzero(&cred->cr_audit, sizeof(cred->cr_audit));
719 }
720 
721 /*
722  * Initialize audit information for the first kernel process (proc 0) and for
723  * the first user process (init).
724  */
725 void
726 audit_cred_kproc0(struct ucred *cred)
727 {
728 
729 	cred->cr_audit.ai_auid = AU_DEFAUDITID;
730 	cred->cr_audit.ai_termid.at_type = AU_IPv4;
731 }
732 
733 void
734 audit_cred_proc1(struct ucred *cred)
735 {
736 
737 	cred->cr_audit.ai_auid = AU_DEFAUDITID;
738 	cred->cr_audit.ai_termid.at_type = AU_IPv4;
739 }
740 
741 void
742 audit_thread_alloc(struct thread *td)
743 {
744 
745 	td->td_ar = NULL;
746 }
747 
748 void
749 audit_thread_free(struct thread *td)
750 {
751 
752 	KASSERT(td->td_ar == NULL, ("audit_thread_free: td_ar != NULL"));
753 	KASSERT((td->td_pflags & TDP_AUDITREC) == 0,
754 	    ("audit_thread_free: TDP_AUDITREC set"));
755 }
756 
757 void
758 audit_proc_coredump(struct thread *td, char *path, int errcode)
759 {
760 	struct kaudit_record *ar;
761 	struct au_mask *aumask;
762 	struct ucred *cred;
763 	au_class_t class;
764 	int ret, sorf;
765 	char **pathp;
766 	au_id_t auid;
767 
768 	ret = 0;
769 
770 	/*
771 	 * Make sure we are using the correct preselection mask.
772 	 */
773 	cred = td->td_ucred;
774 	auid = cred->cr_audit.ai_auid;
775 	if (auid == AU_DEFAUDITID)
776 		aumask = &audit_nae_mask;
777 	else
778 		aumask = &cred->cr_audit.ai_mask;
779 	/*
780 	 * It's possible for coredump(9) generation to fail.  Make sure that
781 	 * we handle this case correctly for preselection.
782 	 */
783 	if (errcode != 0)
784 		sorf = AU_PRS_FAILURE;
785 	else
786 		sorf = AU_PRS_SUCCESS;
787 	class = au_event_class(AUE_CORE);
788 	if (au_preselect(AUE_CORE, class, aumask, sorf) == 0 &&
789 	    audit_pipe_preselect(auid, AUE_CORE, class, sorf, 0) == 0)
790 		return;
791 
792 	/*
793 	 * If we are interested in seeing this audit record, allocate it.
794 	 * Where possible coredump records should contain a pathname and arg32
795 	 * (signal) tokens.
796 	 */
797 	ar = audit_new(AUE_CORE, td);
798 	if (ar == NULL)
799 		return;
800 	if (path != NULL) {
801 		pathp = &ar->k_ar.ar_arg_upath1;
802 		*pathp = malloc(MAXPATHLEN, M_AUDITPATH, M_WAITOK);
803 		audit_canon_path(td, AT_FDCWD, path, *pathp);
804 		ARG_SET_VALID(ar, ARG_UPATH1);
805 	}
806 	ar->k_ar.ar_arg_signum = td->td_proc->p_sig;
807 	ARG_SET_VALID(ar, ARG_SIGNUM);
808 	if (errcode != 0)
809 		ret = 1;
810 	audit_commit(ar, errcode, ret);
811 }
812