kern_ktrace.c revision 1.39 1 /* $NetBSD: kern_ktrace.c,v 1.39 2000/04/19 19:14:17 thorpej Exp $ */
2
3 /*
4 * Copyright (c) 1989, 1993
5 * The Regents of the University of California. All rights reserved.
6 *
7 * Redistribution and use in source and binary forms, with or without
8 * modification, are permitted provided that the following conditions
9 * are met:
10 * 1. Redistributions of source code must retain the above copyright
11 * notice, this list of conditions and the following disclaimer.
12 * 2. Redistributions in binary form must reproduce the above copyright
13 * notice, this list of conditions and the following disclaimer in the
14 * documentation and/or other materials provided with the distribution.
15 * 3. All advertising materials mentioning features or use of this software
16 * must display the following acknowledgement:
17 * This product includes software developed by the University of
18 * California, Berkeley and its contributors.
19 * 4. Neither the name of the University nor the names of its contributors
20 * may be used to endorse or promote products derived from this software
21 * without specific prior written permission.
22 *
23 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
24 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
25 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
26 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
27 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
28 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
29 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
30 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
31 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
32 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
33 * SUCH DAMAGE.
34 *
35 * @(#)kern_ktrace.c 8.5 (Berkeley) 5/14/95
36 */
37
38 #include "opt_ktrace.h"
39
40 #ifdef KTRACE
41
42 #include <sys/param.h>
43 #include <sys/systm.h>
44 #include <sys/proc.h>
45 #include <sys/file.h>
46 #include <sys/namei.h>
47 #include <sys/vnode.h>
48 #include <sys/ktrace.h>
49 #include <sys/malloc.h>
50 #include <sys/syslog.h>
51 #include <sys/filedesc.h>
52
53 #include <sys/mount.h>
54 #include <sys/syscallargs.h>
55
56 void ktrinitheader __P((struct ktr_header *, struct proc *, int));
57 int ktrops __P((struct proc *, struct proc *, int, int, void *));
58 int ktrsetchildren __P((struct proc *, struct proc *, int, int, void *));
59 int ktrwrite __P((struct proc *, void *, struct ktr_header *));
60 int ktrcanset __P((struct proc *, struct proc *));
61
62 void
63 ktrderef(p)
64 struct proc *p;
65 {
66 if (p->p_tracep == NULL)
67 return;
68
69 if (p->p_traceflag & KTRFAC_FD) {
70 struct file *fp = p->p_tracep;
71
72 FILE_USE(fp);
73 closef(fp, NULL);
74 } else {
75 struct vnode *vp = p->p_tracep;
76
77 vrele(vp);
78 }
79 p->p_tracep = NULL;
80 p->p_traceflag = 0;
81 }
82
83 void
84 ktradref(p)
85 struct proc *p;
86 {
87 if (p->p_traceflag & KTRFAC_FD) {
88 struct file *fp = p->p_tracep;
89
90 fp->f_count++;
91 } else {
92 struct vnode *vp = p->p_tracep;
93
94 VREF(vp);
95 }
96 }
97
98 void
99 ktrinitheader(kth, p, type)
100 struct ktr_header *kth;
101 struct proc *p;
102 int type;
103 {
104
105 memset(kth, 0, sizeof(*kth));
106 kth->ktr_type = type;
107 microtime(&kth->ktr_time);
108 kth->ktr_pid = p->p_pid;
109 memcpy(kth->ktr_comm, p->p_comm, MAXCOMLEN);
110 }
111
112 void
113 ktrsyscall(v, code, argsize, args)
114 void *v;
115 register_t code;
116 size_t argsize;
117 register_t args[];
118 {
119 struct ktr_header kth;
120 struct ktr_syscall *ktp;
121 struct proc *p = curproc; /* XXX */
122 register_t *argp;
123 size_t len = sizeof(struct ktr_syscall) + argsize;
124 int i;
125
126 p->p_traceflag |= KTRFAC_ACTIVE;
127 ktrinitheader(&kth, p, KTR_SYSCALL);
128 ktp = malloc(len, M_TEMP, M_WAITOK);
129 ktp->ktr_code = code;
130 ktp->ktr_argsize = argsize;
131 argp = (register_t *)((char *)ktp + sizeof(struct ktr_syscall));
132 for (i = 0; i < (argsize / sizeof(*argp)); i++)
133 *argp++ = args[i];
134 kth.ktr_buf = (caddr_t)ktp;
135 kth.ktr_len = len;
136 (void) ktrwrite(p, v, &kth);
137 free(ktp, M_TEMP);
138 p->p_traceflag &= ~KTRFAC_ACTIVE;
139 }
140
141 void
142 ktrsysret(v, code, error, retval)
143 void *v;
144 register_t code;
145 int error;
146 register_t retval;
147 {
148 struct ktr_header kth;
149 struct ktr_sysret ktp;
150 struct proc *p = curproc; /* XXX */
151
152 p->p_traceflag |= KTRFAC_ACTIVE;
153 ktrinitheader(&kth, p, KTR_SYSRET);
154 ktp.ktr_code = code;
155 ktp.ktr_eosys = 0; /* XXX unused */
156 ktp.ktr_error = error;
157 ktp.ktr_retval = retval; /* what about val2 ? */
158
159 kth.ktr_buf = (caddr_t)&ktp;
160 kth.ktr_len = sizeof(struct ktr_sysret);
161
162 (void) ktrwrite(p, v, &kth);
163 p->p_traceflag &= ~KTRFAC_ACTIVE;
164 }
165
166 void
167 ktrnamei(v, path)
168 void *v;
169 char *path;
170 {
171 struct ktr_header kth;
172 struct proc *p = curproc; /* XXX */
173
174 p->p_traceflag |= KTRFAC_ACTIVE;
175 ktrinitheader(&kth, p, KTR_NAMEI);
176 kth.ktr_len = strlen(path);
177 kth.ktr_buf = path;
178
179 (void) ktrwrite(p, v, &kth);
180 p->p_traceflag &= ~KTRFAC_ACTIVE;
181 }
182
183 void
184 ktremul(v, p, emul)
185 void *v;
186 struct proc *p;
187 char *emul;
188 {
189 struct ktr_header kth;
190
191 p->p_traceflag |= KTRFAC_ACTIVE;
192 ktrinitheader(&kth, p, KTR_EMUL);
193 kth.ktr_len = strlen(emul);
194 kth.ktr_buf = emul;
195
196 (void) ktrwrite(p, v, &kth);
197 p->p_traceflag &= ~KTRFAC_ACTIVE;
198 }
199
200 void
201 ktrgenio(v, fd, rw, iov, len, error)
202 void *v;
203 int fd;
204 enum uio_rw rw;
205 struct iovec *iov;
206 int len, error;
207 {
208 struct ktr_header kth;
209 struct ktr_genio *ktp;
210 caddr_t cp;
211 int resid = len, cnt;
212 struct proc *p = curproc; /* XXX */
213 int buflen;
214
215 if (error)
216 return;
217
218 p->p_traceflag |= KTRFAC_ACTIVE;
219
220 buflen = min(PAGE_SIZE, len + sizeof(struct ktr_genio));
221
222 ktrinitheader(&kth, p, KTR_GENIO);
223 ktp = malloc(buflen, M_TEMP, M_WAITOK);
224 ktp->ktr_fd = fd;
225 ktp->ktr_rw = rw;
226
227 kth.ktr_buf = (caddr_t)ktp;
228
229 cp = (caddr_t)((char *)ktp + sizeof(struct ktr_genio));
230 buflen -= sizeof(struct ktr_genio);
231
232 while (resid > 0) {
233 if (p->p_schedflags & PSCHED_SHOULDYIELD)
234 preempt(NULL);
235
236 cnt = min(iov->iov_len, buflen);
237 if (cnt > resid)
238 cnt = resid;
239 if (copyin(iov->iov_base, cp, cnt))
240 break;
241
242 kth.ktr_len = cnt + sizeof(struct ktr_genio);
243
244 if (ktrwrite(p, v, &kth) != 0)
245 break;
246
247 iov->iov_base = (caddr_t)iov->iov_base + cnt;
248 iov->iov_len -= cnt;
249
250 if (iov->iov_len == 0)
251 iov++;
252
253 resid -= cnt;
254 }
255
256 free(ktp, M_TEMP);
257 p->p_traceflag &= ~KTRFAC_ACTIVE;
258 }
259
260 void
261 ktrpsig(v, sig, action, mask, code)
262 void *v;
263 int sig;
264 sig_t action;
265 sigset_t *mask;
266 int code;
267 {
268 struct ktr_header kth;
269 struct ktr_psig kp;
270 struct proc *p = curproc; /* XXX */
271
272 p->p_traceflag |= KTRFAC_ACTIVE;
273 ktrinitheader(&kth, p, KTR_PSIG);
274 kp.signo = (char)sig;
275 kp.action = action;
276 kp.mask = *mask;
277 kp.code = code;
278 kth.ktr_buf = (caddr_t)&kp;
279 kth.ktr_len = sizeof(struct ktr_psig);
280
281 (void) ktrwrite(p, v, &kth);
282 p->p_traceflag &= ~KTRFAC_ACTIVE;
283 }
284
285 void
286 ktrcsw(v, out, user)
287 void *v;
288 int out, user;
289 {
290 struct ktr_header kth;
291 struct ktr_csw kc;
292 struct proc *p = curproc; /* XXX */
293
294 p->p_traceflag |= KTRFAC_ACTIVE;
295 ktrinitheader(&kth, p, KTR_CSW);
296 kc.out = out;
297 kc.user = user;
298 kth.ktr_buf = (caddr_t)&kc;
299 kth.ktr_len = sizeof(struct ktr_csw);
300
301 (void) ktrwrite(p, v, &kth);
302 p->p_traceflag &= ~KTRFAC_ACTIVE;
303 }
304
305 /* Interface and common routines */
306
307 /*
308 * ktrace system call
309 */
310 /* ARGSUSED */
311 int
312 sys_fktrace(curp, v, retval)
313 struct proc *curp;
314 void *v;
315 register_t *retval;
316 {
317 struct sys_fktrace_args /* {
318 syscallarg(int) fd;
319 syscallarg(int) ops;
320 syscallarg(int) facs;
321 syscallarg(int) pid;
322 } */ *uap = v;
323 struct file *fp = NULL;
324 struct proc *p;
325 struct filedesc *fdp = curp->p_fd;
326 struct pgrp *pg;
327 int facs;
328 int ops;
329 int descend;
330 int ret = 0;
331 int error = 0;
332
333 if (((u_int)SCARG(uap, fd)) >= fdp->fd_nfiles ||
334 (fp = fdp->fd_ofiles[SCARG(uap, fd)]) == NULL ||
335 (fp->f_flag & FWRITE) == 0)
336 return (EBADF);
337
338 ops = KTROP(SCARG(uap, ops)) | KTRFLAG_FD;
339 descend = SCARG(uap, ops) & KTRFLAG_DESCEND;
340 facs = SCARG(uap, facs) & ~((unsigned) KTRFAC_ROOT);
341 curp->p_traceflag |= KTRFAC_ACTIVE;
342
343 /*
344 * Clear all uses of the tracefile
345 */
346 if (KTROP(ops) == KTROP_CLEARFILE) {
347 proclist_lock_read();
348 for (p = LIST_FIRST(&allproc); p != NULL;
349 p = LIST_NEXT(p, p_list)) {
350 if (p->p_tracep == fp) {
351 if (ktrcanset(curp, p))
352 ktrderef(p);
353 else
354 error = EPERM;
355 }
356 }
357 proclist_unlock_read();
358 goto done;
359 }
360 /*
361 * need something to (un)trace (XXX - why is this here?)
362 */
363 if (!facs) {
364 error = EINVAL;
365 goto done;
366 }
367 /*
368 * do it
369 */
370 if (SCARG(uap, pid) < 0) {
371 /*
372 * by process group
373 */
374 pg = pgfind(-SCARG(uap, pid));
375 if (pg == NULL) {
376 error = ESRCH;
377 goto done;
378 }
379 for (p = LIST_FIRST(&pg->pg_members); p != NULL;
380 p = LIST_NEXT(p, p_pglist)) {
381 if (descend)
382 ret |= ktrsetchildren(curp, p, ops, facs, fp);
383 else
384 ret |= ktrops(curp, p, ops, facs, fp);
385 }
386
387 } else {
388 /*
389 * by pid
390 */
391 p = pfind(SCARG(uap, pid));
392 if (p == NULL) {
393 error = ESRCH;
394 goto done;
395 }
396 if (descend)
397 ret |= ktrsetchildren(curp, p, ops, facs, fp);
398 else
399 ret |= ktrops(curp, p, ops, facs, fp);
400 }
401 if (!ret)
402 error = EPERM;
403 done:
404 curp->p_traceflag &= ~KTRFAC_ACTIVE;
405 return (error);
406 }
407
408 /*
409 * ktrace system call
410 */
411 /* ARGSUSED */
412 int
413 sys_ktrace(curp, v, retval)
414 struct proc *curp;
415 void *v;
416 register_t *retval;
417 {
418 struct sys_ktrace_args /* {
419 syscallarg(const char *) fname;
420 syscallarg(int) ops;
421 syscallarg(int) facs;
422 syscallarg(int) pid;
423 } */ *uap = v;
424 struct vnode *vp = NULL;
425 struct proc *p;
426 struct pgrp *pg;
427 int facs = SCARG(uap, facs) & ~((unsigned) KTRFAC_ROOT);
428 int ops = KTROP(SCARG(uap, ops));
429 int descend = SCARG(uap, ops) & KTRFLAG_DESCEND;
430 int ret = 0;
431 int error = 0;
432 struct nameidata nd;
433
434 curp->p_traceflag |= KTRFAC_ACTIVE;
435 if (ops != KTROP_CLEAR) {
436 /*
437 * an operation which requires a file argument.
438 */
439 NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, SCARG(uap, fname),
440 curp);
441 if ((error = vn_open(&nd, FREAD|FWRITE, 0)) != 0) {
442 curp->p_traceflag &= ~KTRFAC_ACTIVE;
443 return (error);
444 }
445 vp = nd.ni_vp;
446 VOP_UNLOCK(vp, 0);
447 if (vp->v_type != VREG) {
448 (void) vn_close(vp, FREAD|FWRITE, curp->p_ucred, curp);
449 curp->p_traceflag &= ~KTRFAC_ACTIVE;
450 return (EACCES);
451 }
452 }
453 /*
454 * Clear all uses of the tracefile
455 */
456 if (KTROP(ops) == KTROP_CLEARFILE) {
457 proclist_lock_read();
458 for (p = LIST_FIRST(&allproc); p != NULL;
459 p = LIST_NEXT(p, p_list)) {
460 if (p->p_tracep == vp &&
461 !ktrops(curp, p, KTROP_CLEAR, ~0, vp))
462 error = EPERM;
463 }
464 proclist_unlock_read();
465 goto done;
466 }
467 /*
468 * need something to (un)trace (XXX - why is this here?)
469 */
470 if (!facs) {
471 error = EINVAL;
472 goto done;
473 }
474 /*
475 * do it
476 */
477 if (SCARG(uap, pid) < 0) {
478 /*
479 * by process group
480 */
481 pg = pgfind(-SCARG(uap, pid));
482 if (pg == NULL) {
483 error = ESRCH;
484 goto done;
485 }
486 for (p = LIST_FIRST(&pg->pg_members); p != NULL;
487 p = LIST_NEXT(p, p_pglist)) {
488 if (descend)
489 ret |= ktrsetchildren(curp, p, ops, facs, vp);
490 else
491 ret |= ktrops(curp, p, ops, facs, vp);
492 }
493
494 } else {
495 /*
496 * by pid
497 */
498 p = pfind(SCARG(uap, pid));
499 if (p == NULL) {
500 error = ESRCH;
501 goto done;
502 }
503 if (descend)
504 ret |= ktrsetchildren(curp, p, ops, facs, vp);
505 else
506 ret |= ktrops(curp, p, ops, facs, vp);
507 }
508 if (!ret)
509 error = EPERM;
510 done:
511 if (vp != NULL)
512 (void) vn_close(vp, FWRITE, curp->p_ucred, curp);
513 curp->p_traceflag &= ~KTRFAC_ACTIVE;
514 return (error);
515 }
516
517 int
518 ktrops(curp, p, ops, facs, v)
519 struct proc *p, *curp;
520 int ops, facs;
521 void *v;
522 {
523
524 if (!ktrcanset(curp, p))
525 return (0);
526 if (KTROP(ops) == KTROP_SET) {
527 if (p->p_tracep != v) {
528 /*
529 * if trace file already in use, relinquish
530 */
531 ktrderef(p);
532 if (ops & KTRFLAG_FD)
533 p->p_traceflag = KTRFAC_FD;
534 p->p_tracep = v;
535 ktradref(p);
536 }
537 p->p_traceflag |= facs;
538 if (curp->p_ucred->cr_uid == 0)
539 p->p_traceflag |= KTRFAC_ROOT;
540 } else {
541 /* KTROP_CLEAR */
542 if (((p->p_traceflag &= ~facs) & KTRFAC_MASK) == 0) {
543 /* no more tracing */
544 ktrderef(p);
545 }
546 }
547
548 /*
549 * Emit an emulation record, every time there is a ktrace
550 * change/attach request.
551 */
552 if (KTRPOINT(p, KTR_EMUL))
553 ktremul(p->p_tracep, p, p->p_emul->e_name);
554
555 return (1);
556 }
557
558 int
559 ktrsetchildren(curp, top, ops, facs, v)
560 struct proc *curp, *top;
561 int ops, facs;
562 void *v;
563 {
564 struct proc *p;
565 int ret = 0;
566
567 p = top;
568 for (;;) {
569 ret |= ktrops(curp, p, ops, facs, v);
570 /*
571 * If this process has children, descend to them next,
572 * otherwise do any siblings, and if done with this level,
573 * follow back up the tree (but not past top).
574 */
575 if (LIST_FIRST(&p->p_children) != NULL)
576 p = LIST_FIRST(&p->p_children);
577 else for (;;) {
578 if (p == top)
579 return (ret);
580 if (LIST_NEXT(p, p_sibling) != NULL) {
581 p = LIST_NEXT(p, p_sibling);
582 break;
583 }
584 p = p->p_pptr;
585 }
586 }
587 /*NOTREACHED*/
588 }
589
590 int
591 ktrwrite(p, v, kth)
592 struct proc *p;
593 void *v;
594 struct ktr_header *kth;
595 {
596 struct uio auio;
597 struct iovec aiov[2];
598 int error;
599
600 if (v == NULL)
601 return (0);
602 auio.uio_iov = &aiov[0];
603 auio.uio_offset = 0;
604 auio.uio_segflg = UIO_SYSSPACE;
605 auio.uio_rw = UIO_WRITE;
606 aiov[0].iov_base = (caddr_t)kth;
607 aiov[0].iov_len = sizeof(struct ktr_header);
608 auio.uio_resid = sizeof(struct ktr_header);
609 auio.uio_iovcnt = 1;
610 auio.uio_procp = (struct proc *)0;
611 if (kth->ktr_len > 0) {
612 auio.uio_iovcnt++;
613 aiov[1].iov_base = kth->ktr_buf;
614 aiov[1].iov_len = kth->ktr_len;
615 auio.uio_resid += kth->ktr_len;
616 }
617 if (p->p_traceflag & KTRFAC_FD) {
618 struct file *fp = v;
619
620 FILE_USE(fp);
621 error = (*fp->f_ops->fo_write)(fp, &fp->f_offset, &auio,
622 fp->f_cred, FOF_UPDATE_OFFSET);
623 FILE_UNUSE(fp, NULL);
624 }
625 else {
626 struct vnode *vp = v;
627
628 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY);
629 error = VOP_WRITE(vp, &auio, IO_UNIT|IO_APPEND, p->p_ucred);
630 VOP_UNLOCK(vp, 0);
631 }
632 if (error == 0)
633 return (0);
634 /*
635 * If error encountered, give up tracing on this vnode. Don't report
636 * EPIPE as this can easily happen with fktrace()/ktruss.
637 */
638 if (error != EPIPE)
639 log(LOG_NOTICE,
640 "ktrace write failed, errno %d, tracing stopped\n",
641 error);
642 proclist_lock_read();
643 for (p = LIST_FIRST(&allproc); p != NULL; p = LIST_NEXT(p, p_list)) {
644 if (p->p_tracep == v)
645 ktrderef(p);
646 }
647 proclist_unlock_read();
648
649 return (error);
650 }
651
652 /*
653 * Return true if caller has permission to set the ktracing state
654 * of target. Essentially, the target can't possess any
655 * more permissions than the caller. KTRFAC_ROOT signifies that
656 * root previously set the tracing status on the target process, and
657 * so, only root may further change it.
658 *
659 * TODO: check groups. use caller effective gid.
660 */
661 int
662 ktrcanset(callp, targetp)
663 struct proc *callp, *targetp;
664 {
665 struct pcred *caller = callp->p_cred;
666 struct pcred *target = targetp->p_cred;
667
668 if ((caller->pc_ucred->cr_uid == target->p_ruid &&
669 target->p_ruid == target->p_svuid &&
670 caller->p_rgid == target->p_rgid && /* XXX */
671 target->p_rgid == target->p_svgid &&
672 (targetp->p_traceflag & KTRFAC_ROOT) == 0) ||
673 caller->pc_ucred->cr_uid == 0)
674 return (1);
675
676 return (0);
677 }
678
679 #endif
680