1 /*-
2 * Copyright (c) 1999-2008 Apple Inc.
3 * Copyright (c) 2006-2008 Robert N. M. Watson
4 * All rights reserved.
5 *
6 * Redistribution and use in source and binary forms, with or without
7 * modification, are permitted provided that the following conditions
8 * are met:
9 * 1. Redistributions of source code must retain the above copyright
10 * notice, this list of conditions and the following disclaimer.
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.
14 * 3. Neither the name of Apple Inc. ("Apple") nor the names of
15 * its contributors may be used to endorse or promote products derived
16 * from this software without specific prior written permission.
17 *
18 * THIS SOFTWARE IS PROVIDED BY APPLE AND ITS CONTRIBUTORS "AS IS" AND
19 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
20 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
21 * ARE DISCLAIMED. IN NO EVENT SHALL APPLE OR ITS CONTRIBUTORS BE LIABLE FOR
22 * ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
23 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
24 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
25 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT,
26 * STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING
27 * IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
28 * POSSIBILITY OF SUCH DAMAGE.
29 */
30
31 #include <sys/cdefs.h>
32 __FBSDID("$FreeBSD: releng/8.1/sys/security/audit/audit_worker.c 191990 2009-05-11 15:33:26Z attilio $");
33
34 #include <sys/param.h>
35 #include <sys/condvar.h>
36 #include <sys/conf.h>
37 #include <sys/file.h>
38 #include <sys/filedesc.h>
39 #include <sys/fcntl.h>
40 #include <sys/ipc.h>
41 #include <sys/kernel.h>
42 #include <sys/kthread.h>
43 #include <sys/malloc.h>
44 #include <sys/mount.h>
45 #include <sys/namei.h>
46 #include <sys/proc.h>
47 #include <sys/queue.h>
48 #include <sys/socket.h>
49 #include <sys/socketvar.h>
50 #include <sys/protosw.h>
51 #include <sys/domain.h>
52 #include <sys/sx.h>
53 #include <sys/sysproto.h>
54 #include <sys/sysent.h>
55 #include <sys/systm.h>
56 #include <sys/ucred.h>
57 #include <sys/uio.h>
58 #include <sys/un.h>
59 #include <sys/unistd.h>
60 #include <sys/vnode.h>
61
62 #include <bsm/audit.h>
63 #include <bsm/audit_internal.h>
64 #include <bsm/audit_kevents.h>
65
66 #include <netinet/in.h>
67 #include <netinet/in_pcb.h>
68
69 #include <security/audit/audit.h>
70 #include <security/audit/audit_private.h>
71
72 #include <vm/uma.h>
73
74 /*
75 * Worker thread that will schedule disk I/O, etc.
76 */
77 static struct proc *audit_thread;
78
79 /*
80 * audit_cred and audit_vp are the stored credential and vnode to use for
81 * active audit trail. They are protected by the audit worker lock, which
82 * will be held across all I/O and all rotation to prevent them from being
83 * replaced (rotated) while in use. The audit_file_rotate_wait flag is set
84 * when the kernel has delivered a trigger to auditd to rotate the trail, and
85 * is cleared when the next rotation takes place. It is also protected by
86 * the audit worker lock.
87 */
88 static int audit_file_rotate_wait;
89 static struct ucred *audit_cred;
90 static struct vnode *audit_vp;
91 static struct sx audit_worker_lock;
92
93 #define AUDIT_WORKER_LOCK_INIT() sx_init(&audit_worker_lock, \
94 "audit_worker_lock");
95 #define AUDIT_WORKER_LOCK_ASSERT() sx_assert(&audit_worker_lock, \
96 SA_XLOCKED)
97 #define AUDIT_WORKER_LOCK() sx_xlock(&audit_worker_lock)
98 #define AUDIT_WORKER_UNLOCK() sx_xunlock(&audit_worker_lock)
99
100 /*
101 * Write an audit record to a file, performed as the last stage after both
102 * preselection and BSM conversion. Both space management and write failures
103 * are handled in this function.
104 *
105 * No attempt is made to deal with possible failure to deliver a trigger to
106 * the audit daemon, since the message is asynchronous anyway.
107 */
108 static void
109 audit_record_write(struct vnode *vp, struct ucred *cred, void *data,
110 size_t len)
111 {
112 static struct timeval last_lowspace_trigger;
113 static struct timeval last_fail;
114 static int cur_lowspace_trigger;
115 struct statfs *mnt_stat;
116 int error, vfslocked;
117 static int cur_fail;
118 struct vattr vattr;
119 long temp;
120
121 AUDIT_WORKER_LOCK_ASSERT();
122
123 if (vp == NULL)
124 return;
125
126 mnt_stat = &vp->v_mount->mnt_stat;
127 vfslocked = VFS_LOCK_GIANT(vp->v_mount);
128
129 /*
130 * First, gather statistics on the audit log file and file system so
131 * that we know how we're doing on space. Consider failure of these
132 * operations to indicate a future inability to write to the file.
133 */
134 error = VFS_STATFS(vp->v_mount, mnt_stat);
135 if (error)
136 goto fail;
137 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY);
138 error = VOP_GETATTR(vp, &vattr, cred);
139 VOP_UNLOCK(vp, 0);
140 if (error)
141 goto fail;
142 audit_fstat.af_currsz = vattr.va_size;
143
144 /*
145 * We handle four different space-related limits:
146 *
147 * - A fixed (hard) limit on the minimum free blocks we require on
148 * the file system, and results in record loss, a trigger, and
149 * possible fail stop due to violating invariants.
150 *
151 * - An administrative (soft) limit, which when fallen below, results
152 * in the kernel notifying the audit daemon of low space.
153 *
154 * - An audit trail size limit, which when gone above, results in the
155 * kernel notifying the audit daemon that rotation is desired.
156 *
157 * - The total depth of the kernel audit record exceeding free space,
158 * which can lead to possible fail stop (with drain), in order to
159 * prevent violating invariants. Failure here doesn't halt
160 * immediately, but prevents new records from being generated.
161 *
162 * Possibly, the last of these should be handled differently, always
163 * allowing a full queue to be lost, rather than trying to prevent
164 * loss.
165 *
166 * First, handle the hard limit, which generates a trigger and may
167 * fail stop. This is handled in the same manner as ENOSPC from
168 * VOP_WRITE, and results in record loss.
169 */
170 if (mnt_stat->f_bfree < AUDIT_HARD_LIMIT_FREE_BLOCKS) {
171 error = ENOSPC;
172 goto fail_enospc;
173 }
174
175 /*
176 * Second, handle falling below the soft limit, if defined; we send
177 * the daemon a trigger and continue processing the record. Triggers
178 * are limited to 1/sec.
179 */
180 if (audit_qctrl.aq_minfree != 0) {
181 temp = mnt_stat->f_blocks / (100 / audit_qctrl.aq_minfree);
182 if (mnt_stat->f_bfree < temp) {
183 if (ppsratecheck(&last_lowspace_trigger,
184 &cur_lowspace_trigger, 1)) {
185 (void)audit_send_trigger(
186 AUDIT_TRIGGER_LOW_SPACE);
187 printf("Warning: disk space low (< %d%% free) "
188 "on audit log file-system\n",
189 audit_qctrl.aq_minfree);
190 }
191 }
192 }
193
194 /*
195 * If the current file is getting full, generate a rotation trigger
196 * to the daemon. This is only approximate, which is fine as more
197 * records may be generated before the daemon rotates the file.
198 */
199 if ((audit_fstat.af_filesz != 0) && (audit_file_rotate_wait == 0) &&
200 (vattr.va_size >= audit_fstat.af_filesz)) {
201 AUDIT_WORKER_LOCK_ASSERT();
202
203 audit_file_rotate_wait = 1;
204 (void)audit_send_trigger(AUDIT_TRIGGER_ROTATE_KERNEL);
205 }
206
207 /*
208 * If the estimated amount of audit data in the audit event queue
209 * (plus records allocated but not yet queued) has reached the amount
210 * of free space on the disk, then we need to go into an audit fail
211 * stop state, in which we do not permit the allocation/committing of
212 * any new audit records. We continue to process records but don't
213 * allow any activities that might generate new records. In the
214 * future, we might want to detect when space is available again and
215 * allow operation to continue, but this behavior is sufficient to
216 * meet fail stop requirements in CAPP.
217 */
218 if (audit_fail_stop) {
219 if ((unsigned long)((audit_q_len + audit_pre_q_len + 1) *
220 MAX_AUDIT_RECORD_SIZE) / mnt_stat->f_bsize >=
221 (unsigned long)(mnt_stat->f_bfree)) {
222 if (ppsratecheck(&last_fail, &cur_fail, 1))
223 printf("audit_record_write: free space "
224 "below size of audit queue, failing "
225 "stop\n");
226 audit_in_failure = 1;
227 } else if (audit_in_failure) {
228 /*
229 * Note: if we want to handle recovery, this is the
230 * spot to do it: unset audit_in_failure, and issue a
231 * wakeup on the cv.
232 */
233 }
234 }
235
236 error = vn_rdwr(UIO_WRITE, vp, data, len, (off_t)0, UIO_SYSSPACE,
237 IO_APPEND|IO_UNIT, cred, NULL, NULL, curthread);
238 if (error == ENOSPC)
239 goto fail_enospc;
240 else if (error)
241 goto fail;
242
243 /*
244 * Catch completion of a queue drain here; if we're draining and the
245 * queue is now empty, fail stop. That audit_fail_stop is implicitly
246 * true, since audit_in_failure can only be set of audit_fail_stop is
247 * set.
248 *
249 * Note: if we handle recovery from audit_in_failure, then we need to
250 * make panic here conditional.
251 */
252 if (audit_in_failure) {
253 if (audit_q_len == 0 && audit_pre_q_len == 0) {
254 VOP_LOCK(vp, LK_EXCLUSIVE | LK_RETRY);
255 (void)VOP_FSYNC(vp, MNT_WAIT, curthread);
256 VOP_UNLOCK(vp, 0);
257 panic("Audit store overflow; record queue drained.");
258 }
259 }
260
261 VFS_UNLOCK_GIANT(vfslocked);
262 return;
263
264 fail_enospc:
265 /*
266 * ENOSPC is considered a special case with respect to failures, as
267 * this can reflect either our preemptive detection of insufficient
268 * space, or ENOSPC returned by the vnode write call.
269 */
270 if (audit_fail_stop) {
271 VOP_LOCK(vp, LK_EXCLUSIVE | LK_RETRY);
272 (void)VOP_FSYNC(vp, MNT_WAIT, curthread);
273 VOP_UNLOCK(vp, 0);
274 panic("Audit log space exhausted and fail-stop set.");
275 }
276 (void)audit_send_trigger(AUDIT_TRIGGER_NO_SPACE);
277 audit_suspended = 1;
278
279 /* FALLTHROUGH */
280 fail:
281 /*
282 * We have failed to write to the file, so the current record is
283 * lost, which may require an immediate system halt.
284 */
285 if (audit_panic_on_write_fail) {
286 VOP_LOCK(vp, LK_EXCLUSIVE | LK_RETRY);
287 (void)VOP_FSYNC(vp, MNT_WAIT, curthread);
288 VOP_UNLOCK(vp, 0);
289 panic("audit_worker: write error %d\n", error);
290 } else if (ppsratecheck(&last_fail, &cur_fail, 1))
291 printf("audit_worker: write error %d\n", error);
292 VFS_UNLOCK_GIANT(vfslocked);
293 }
294
295 /*
296 * Given a kernel audit record, process as required. Kernel audit records
297 * are converted to one, or possibly two, BSM records, depending on whether
298 * there is a user audit record present also. Kernel records need be
299 * converted to BSM before they can be written out. Both types will be
300 * written to disk, and audit pipes.
301 */
302 static void
303 audit_worker_process_record(struct kaudit_record *ar)
304 {
305 struct au_record *bsm;
306 au_class_t class;
307 au_event_t event;
308 au_id_t auid;
309 int error, sorf;
310 int locked;
311
312 /*
313 * We hold the audit worker lock over both writes, if there are two,
314 * so that the two records won't be split across a rotation and end
315 * up in two different trail files.
316 */
317 if (((ar->k_ar_commit & AR_COMMIT_USER) &&
318 (ar->k_ar_commit & AR_PRESELECT_USER_TRAIL)) ||
319 (ar->k_ar_commit & AR_PRESELECT_TRAIL)) {
320 AUDIT_WORKER_LOCK();
321 locked = 1;
322 } else
323 locked = 0;
324
325 /*
326 * First, handle the user record, if any: commit to the system trail
327 * and audit pipes as selected.
328 */
329 if ((ar->k_ar_commit & AR_COMMIT_USER) &&
330 (ar->k_ar_commit & AR_PRESELECT_USER_TRAIL)) {
331 AUDIT_WORKER_LOCK_ASSERT();
332 audit_record_write(audit_vp, audit_cred, ar->k_udata,
333 ar->k_ulen);
334 }
335
336 if ((ar->k_ar_commit & AR_COMMIT_USER) &&
337 (ar->k_ar_commit & AR_PRESELECT_USER_PIPE))
338 audit_pipe_submit_user(ar->k_udata, ar->k_ulen);
339
340 if (!(ar->k_ar_commit & AR_COMMIT_KERNEL) ||
341 ((ar->k_ar_commit & AR_PRESELECT_PIPE) == 0 &&
342 (ar->k_ar_commit & AR_PRESELECT_TRAIL) == 0))
343 goto out;
344
345 auid = ar->k_ar.ar_subj_auid;
346 event = ar->k_ar.ar_event;
347 class = au_event_class(event);
348 if (ar->k_ar.ar_errno == 0)
349 sorf = AU_PRS_SUCCESS;
350 else
351 sorf = AU_PRS_FAILURE;
352
353 error = kaudit_to_bsm(ar, &bsm);
354 switch (error) {
355 case BSM_NOAUDIT:
356 goto out;
357
358 case BSM_FAILURE:
359 printf("audit_worker_process_record: BSM_FAILURE\n");
360 goto out;
361
362 case BSM_SUCCESS:
363 break;
364
365 default:
366 panic("kaudit_to_bsm returned %d", error);
367 }
368
369 if (ar->k_ar_commit & AR_PRESELECT_TRAIL) {
370 AUDIT_WORKER_LOCK_ASSERT();
371 audit_record_write(audit_vp, audit_cred, bsm->data, bsm->len);
372 }
373
374 if (ar->k_ar_commit & AR_PRESELECT_PIPE)
375 audit_pipe_submit(auid, event, class, sorf,
376 ar->k_ar_commit & AR_PRESELECT_TRAIL, bsm->data,
377 bsm->len);
378
379 kau_free(bsm);
380 out:
381 if (locked)
382 AUDIT_WORKER_UNLOCK();
383 }
384
385 /*
386 * The audit_worker thread is responsible for watching the event queue,
387 * dequeueing records, converting them to BSM format, and committing them to
388 * disk. In order to minimize lock thrashing, records are dequeued in sets
389 * to a thread-local work queue.
390 *
391 * Note: this means that the effect bound on the size of the pending record
392 * queue is 2x the length of the global queue.
393 */
394 static void
395 audit_worker(void *arg)
396 {
397 struct kaudit_queue ar_worklist;
398 struct kaudit_record *ar;
399 int lowater_signal;
400
401 TAILQ_INIT(&ar_worklist);
402 mtx_lock(&audit_mtx);
403 while (1) {
404 mtx_assert(&audit_mtx, MA_OWNED);
405
406 /*
407 * Wait for a record.
408 */
409 while (TAILQ_EMPTY(&audit_q))
410 cv_wait(&audit_worker_cv, &audit_mtx);
411
412 /*
413 * If there are records in the global audit record queue,
414 * transfer them to a thread-local queue and process them
415 * one by one. If we cross the low watermark threshold,
416 * signal any waiting processes that they may wake up and
417 * continue generating records.
418 */
419 lowater_signal = 0;
420 while ((ar = TAILQ_FIRST(&audit_q))) {
421 TAILQ_REMOVE(&audit_q, ar, k_q);
422 audit_q_len--;
423 if (audit_q_len == audit_qctrl.aq_lowater)
424 lowater_signal++;
425 TAILQ_INSERT_TAIL(&ar_worklist, ar, k_q);
426 }
427 if (lowater_signal)
428 cv_broadcast(&audit_watermark_cv);
429
430 mtx_unlock(&audit_mtx);
431 while ((ar = TAILQ_FIRST(&ar_worklist))) {
432 TAILQ_REMOVE(&ar_worklist, ar, k_q);
433 audit_worker_process_record(ar);
434 audit_free(ar);
435 }
436 mtx_lock(&audit_mtx);
437 }
438 }
439
440 /*
441 * audit_rotate_vnode() is called by a user or kernel thread to configure or
442 * de-configure auditing on a vnode. The arguments are the replacement
443 * credential (referenced) and vnode (referenced and opened) to substitute
444 * for the current credential and vnode, if any. If either is set to NULL,
445 * both should be NULL, and this is used to indicate that audit is being
446 * disabled. Any previous cred/vnode will be closed and freed. We re-enable
447 * generating rotation requests to auditd.
448 */
449 void
450 audit_rotate_vnode(struct ucred *cred, struct vnode *vp)
451 {
452 struct ucred *old_audit_cred;
453 struct vnode *old_audit_vp;
454 int vfslocked;
455
456 KASSERT((cred != NULL && vp != NULL) || (cred == NULL && vp == NULL),
457 ("audit_rotate_vnode: cred %p vp %p", cred, vp));
458
459 /*
460 * Rotate the vnode/cred, and clear the rotate flag so that we will
461 * send a rotate trigger if the new file fills.
462 */
463 AUDIT_WORKER_LOCK();
464 old_audit_cred = audit_cred;
465 old_audit_vp = audit_vp;
466 audit_cred = cred;
467 audit_vp = vp;
468 audit_file_rotate_wait = 0;
469 audit_enabled = (audit_vp != NULL);
470 AUDIT_WORKER_UNLOCK();
471
472 /*
473 * If there was an old vnode/credential, close and free.
474 */
475 if (old_audit_vp != NULL) {
476 vfslocked = VFS_LOCK_GIANT(old_audit_vp->v_mount);
477 vn_close(old_audit_vp, AUDIT_CLOSE_FLAGS, old_audit_cred,
478 curthread);
479 VFS_UNLOCK_GIANT(vfslocked);
480 crfree(old_audit_cred);
481 }
482 }
483
484 void
485 audit_worker_init(void)
486 {
487 int error;
488
489 AUDIT_WORKER_LOCK_INIT();
490 error = kproc_create(audit_worker, NULL, &audit_thread, RFHIGHPID,
491 0, "audit");
492 if (error)
493 panic("audit_worker_init: kproc_create returned %d", error);
494 }
Cache object: 5f6bd83ea51946fcb5feeebbb5a337ab
|