proc->thread stage 3.5: Add an IO_CORE flag so coda doesn't have to dig
[dragonfly.git] / sys / kern / kern_ktrace.c
CommitLineData
984263bc
MD
1/*
2 * Copyright (c) 1989, 1993
3 * The Regents of the University of California. All rights reserved.
4 *
5 * Redistribution and use in source and binary forms, with or without
6 * modification, are permitted provided that the following conditions
7 * are met:
8 * 1. Redistributions of source code must retain the above copyright
9 * notice, this list of conditions and the following disclaimer.
10 * 2. Redistributions in binary form must reproduce the above copyright
11 * notice, this list of conditions and the following disclaimer in the
12 * documentation and/or other materials provided with the distribution.
13 * 3. All advertising materials mentioning features or use of this software
14 * must display the following acknowledgement:
15 * This product includes software developed by the University of
16 * California, Berkeley and its contributors.
17 * 4. Neither the name of the University nor the names of its contributors
18 * may be used to endorse or promote products derived from this software
19 * without specific prior written permission.
20 *
21 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
22 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
23 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
24 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
25 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
26 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
27 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
28 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
29 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
30 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
31 * SUCH DAMAGE.
32 *
33 * @(#)kern_ktrace.c 8.2 (Berkeley) 9/23/93
34 * $FreeBSD: src/sys/kern/kern_ktrace.c,v 1.35.2.6 2002/07/05 22:36:38 darrenr Exp $
41c20dac 35 * $DragonFly: src/sys/kern/kern_ktrace.c,v 1.3 2003/06/23 17:55:41 dillon Exp $
984263bc
MD
36 */
37
38#include "opt_ktrace.h"
39
40#include <sys/param.h>
41#include <sys/systm.h>
42#include <sys/sysproto.h>
43#include <sys/kernel.h>
44#include <sys/proc.h>
45#include <sys/fcntl.h>
46#include <sys/lock.h>
47#include <sys/namei.h>
48#include <sys/vnode.h>
49#include <sys/ktrace.h>
50#include <sys/malloc.h>
51#include <sys/syslog.h>
52#include <sys/sysent.h>
53
54#include <vm/vm_zone.h>
55static MALLOC_DEFINE(M_KTRACE, "KTRACE", "KTRACE");
56
57#ifdef KTRACE
58static struct ktr_header *ktrgetheader __P((int type));
59static void ktrwrite __P((struct vnode *, struct ktr_header *, struct uio *));
60static int ktrcanset __P((struct proc *,struct proc *));
61static int ktrsetchildren __P((struct proc *,struct proc *,int,int,struct vnode *));
62static int ktrops __P((struct proc *,struct proc *,int,int,struct vnode *));
63
64
65static struct ktr_header *
66ktrgetheader(type)
67 int type;
68{
69 register struct ktr_header *kth;
70 struct proc *p = curproc; /* XXX */
71
72 MALLOC(kth, struct ktr_header *, sizeof (struct ktr_header),
73 M_KTRACE, M_WAITOK);
74 kth->ktr_type = type;
75 microtime(&kth->ktr_time);
76 kth->ktr_pid = p->p_pid;
77 bcopy(p->p_comm, kth->ktr_comm, MAXCOMLEN + 1);
78 return (kth);
79}
80
81void
82ktrsyscall(vp, code, narg, args)
83 struct vnode *vp;
84 int code, narg;
85 register_t args[];
86{
87 struct ktr_header *kth;
88 struct ktr_syscall *ktp;
89 register int len = offsetof(struct ktr_syscall, ktr_args) +
90 (narg * sizeof(register_t));
91 struct proc *p = curproc; /* XXX */
92 register_t *argp;
93 int i;
94
95 p->p_traceflag |= KTRFAC_ACTIVE;
96 kth = ktrgetheader(KTR_SYSCALL);
97 MALLOC(ktp, struct ktr_syscall *, len, M_KTRACE, M_WAITOK);
98 ktp->ktr_code = code;
99 ktp->ktr_narg = narg;
100 argp = &ktp->ktr_args[0];
101 for (i = 0; i < narg; i++)
102 *argp++ = args[i];
103 kth->ktr_buf = (caddr_t)ktp;
104 kth->ktr_len = len;
105 ktrwrite(vp, kth, NULL);
106 FREE(ktp, M_KTRACE);
107 FREE(kth, M_KTRACE);
108 p->p_traceflag &= ~KTRFAC_ACTIVE;
109}
110
111void
112ktrsysret(vp, code, error, retval)
113 struct vnode *vp;
114 int code, error;
115 register_t retval;
116{
117 struct ktr_header *kth;
118 struct ktr_sysret ktp;
119 struct proc *p = curproc; /* XXX */
120
121 p->p_traceflag |= KTRFAC_ACTIVE;
122 kth = ktrgetheader(KTR_SYSRET);
123 ktp.ktr_code = code;
124 ktp.ktr_error = error;
125 ktp.ktr_retval = retval; /* what about val2 ? */
126
127 kth->ktr_buf = (caddr_t)&ktp;
128 kth->ktr_len = sizeof(struct ktr_sysret);
129
130 ktrwrite(vp, kth, NULL);
131 FREE(kth, M_KTRACE);
132 p->p_traceflag &= ~KTRFAC_ACTIVE;
133}
134
135void
136ktrnamei(vp, path)
137 struct vnode *vp;
138 char *path;
139{
140 struct ktr_header *kth;
141 struct proc *p = curproc; /* XXX */
142
143 /*
144 * don't let vp get ripped out from under us
145 */
146 if (vp)
147 VREF(vp);
148 p->p_traceflag |= KTRFAC_ACTIVE;
149 kth = ktrgetheader(KTR_NAMEI);
150 kth->ktr_len = strlen(path);
151 kth->ktr_buf = path;
152
153 ktrwrite(vp, kth, NULL);
154 if (vp)
155 vrele(vp);
156 FREE(kth, M_KTRACE);
157 p->p_traceflag &= ~KTRFAC_ACTIVE;
158}
159
160void
161ktrgenio(vp, fd, rw, uio, error)
162 struct vnode *vp;
163 int fd;
164 enum uio_rw rw;
165 struct uio *uio;
166 int error;
167{
168 struct ktr_header *kth;
169 struct ktr_genio ktg;
170 struct proc *p = curproc; /* XXX */
171
172 if (error)
173 return;
174 /*
175 * don't let p_tracep get ripped out from under us
176 */
177 if (vp)
178 VREF(vp);
179 p->p_traceflag |= KTRFAC_ACTIVE;
180 kth = ktrgetheader(KTR_GENIO);
181 ktg.ktr_fd = fd;
182 ktg.ktr_rw = rw;
183 kth->ktr_buf = (caddr_t)&ktg;
184 kth->ktr_len = sizeof(struct ktr_genio);
185 uio->uio_offset = 0;
186 uio->uio_rw = UIO_WRITE;
187
188 ktrwrite(vp, kth, uio);
189 if (vp)
190 vrele(vp);
191 FREE(kth, M_KTRACE);
192 p->p_traceflag &= ~KTRFAC_ACTIVE;
193}
194
195void
196ktrpsig(vp, sig, action, mask, code)
197 struct vnode *vp;
198 int sig;
199 sig_t action;
200 sigset_t *mask;
201 int code;
202{
203 struct ktr_header *kth;
204 struct ktr_psig kp;
205 struct proc *p = curproc; /* XXX */
206
207 /*
208 * don't let vp get ripped out from under us
209 */
210 if (vp)
211 VREF(vp);
212 p->p_traceflag |= KTRFAC_ACTIVE;
213 kth = ktrgetheader(KTR_PSIG);
214 kp.signo = (char)sig;
215 kp.action = action;
216 kp.mask = *mask;
217 kp.code = code;
218 kth->ktr_buf = (caddr_t)&kp;
219 kth->ktr_len = sizeof (struct ktr_psig);
220
221 ktrwrite(vp, kth, NULL);
222 if (vp)
223 vrele(vp);
224 FREE(kth, M_KTRACE);
225 p->p_traceflag &= ~KTRFAC_ACTIVE;
226}
227
228void
229ktrcsw(vp, out, user)
230 struct vnode *vp;
231 int out, user;
232{
233 struct ktr_header *kth;
234 struct ktr_csw kc;
235 struct proc *p = curproc; /* XXX */
236
237 /*
238 * don't let vp get ripped out from under us
239 */
240 if (vp)
241 VREF(vp);
242 p->p_traceflag |= KTRFAC_ACTIVE;
243 kth = ktrgetheader(KTR_CSW);
244 kc.out = out;
245 kc.user = user;
246 kth->ktr_buf = (caddr_t)&kc;
247 kth->ktr_len = sizeof (struct ktr_csw);
248
249 ktrwrite(vp, kth, NULL);
250 if (vp)
251 vrele(vp);
252 FREE(kth, M_KTRACE);
253 p->p_traceflag &= ~KTRFAC_ACTIVE;
254}
255#endif
256
257/* Interface and common routines */
258
259/*
260 * ktrace system call
261 */
262#ifndef _SYS_SYSPROTO_H_
263struct ktrace_args {
264 char *fname;
265 int ops;
266 int facs;
267 int pid;
268};
269#endif
270/* ARGSUSED */
271int
41c20dac 272ktrace(struct ktrace_args *uap)
984263bc
MD
273{
274#ifdef KTRACE
41c20dac
MD
275 struct proc *curp = curproc;
276 struct vnode *vp = NULL;
277 struct proc *p;
984263bc
MD
278 struct pgrp *pg;
279 int facs = uap->facs & ~KTRFAC_ROOT;
280 int ops = KTROP(uap->ops);
281 int descend = uap->ops & KTRFLAG_DESCEND;
282 int ret = 0;
283 int error = 0;
284 struct nameidata nd;
285
286 curp->p_traceflag |= KTRFAC_ACTIVE;
287 if (ops != KTROP_CLEAR) {
288 /*
289 * an operation which requires a file argument.
290 */
291 NDINIT(&nd, LOOKUP, NOFOLLOW, UIO_USERSPACE, uap->fname, curp);
292 error = vn_open(&nd, FREAD|FWRITE|O_NOFOLLOW, 0);
293 if (error) {
294 curp->p_traceflag &= ~KTRFAC_ACTIVE;
295 return (error);
296 }
297 NDFREE(&nd, NDF_ONLY_PNBUF);
298 vp = nd.ni_vp;
299 VOP_UNLOCK(vp, 0, curp);
300 if (vp->v_type != VREG) {
301 (void) vn_close(vp, FREAD|FWRITE, curp->p_ucred, curp);
302 curp->p_traceflag &= ~KTRFAC_ACTIVE;
303 return (EACCES);
304 }
305 }
306 /*
307 * Clear all uses of the tracefile. XXX umm, what happens to the
308 * loop if vn_close() blocks?
309 */
310 if (ops == KTROP_CLEARFILE) {
311 LIST_FOREACH(p, &allproc, p_list) {
312 if (p->p_tracep == vp) {
313 if (ktrcanset(curp, p) && p->p_tracep == vp) {
314 p->p_tracep = NULL;
315 p->p_traceflag = 0;
316 (void) vn_close(vp, FREAD|FWRITE,
317 p->p_ucred, p);
318 } else {
319 error = EPERM;
320 }
321 }
322 }
323 goto done;
324 }
325 /*
326 * need something to (un)trace (XXX - why is this here?)
327 */
328 if (!facs) {
329 error = EINVAL;
330 goto done;
331 }
332 /*
333 * do it
334 */
335 if (uap->pid < 0) {
336 /*
337 * by process group
338 */
339 pg = pgfind(-uap->pid);
340 if (pg == NULL) {
341 error = ESRCH;
342 goto done;
343 }
344 LIST_FOREACH(p, &pg->pg_members, p_pglist)
345 if (descend)
346 ret |= ktrsetchildren(curp, p, ops, facs, vp);
347 else
348 ret |= ktrops(curp, p, ops, facs, vp);
349
350 } else {
351 /*
352 * by pid
353 */
354 p = pfind(uap->pid);
355 if (p == NULL) {
356 error = ESRCH;
357 goto done;
358 }
359 if (descend)
360 ret |= ktrsetchildren(curp, p, ops, facs, vp);
361 else
362 ret |= ktrops(curp, p, ops, facs, vp);
363 }
364 if (!ret)
365 error = EPERM;
366done:
367 if (vp != NULL)
368 (void) vn_close(vp, FWRITE, curp->p_ucred, curp);
369 curp->p_traceflag &= ~KTRFAC_ACTIVE;
370 return (error);
371#else
372 return ENOSYS;
373#endif
374}
375
376/*
377 * utrace system call
378 */
379/* ARGSUSED */
380int
41c20dac 381utrace(struct utrace_args *uap)
984263bc
MD
382{
383#ifdef KTRACE
384 struct ktr_header *kth;
385 struct proc *p = curproc; /* XXX */
386 struct vnode *vp;
387 register caddr_t cp;
388
389 if (!KTRPOINT(p, KTR_USER))
390 return (0);
391 if (SCARG(uap, len) > KTR_USER_MAXLEN)
392 return (EINVAL);
393 p->p_traceflag |= KTRFAC_ACTIVE;
394 /*
395 * don't let p_tracep get ripped out from under us while we are
396 * writing.
397 */
398 if ((vp = p->p_tracep) != NULL)
399 VREF(vp);
400 kth = ktrgetheader(KTR_USER);
401 MALLOC(cp, caddr_t, uap->len, M_KTRACE, M_WAITOK);
402 if (!copyin(uap->addr, cp, uap->len)) {
403 kth->ktr_buf = cp;
404 kth->ktr_len = uap->len;
405 ktrwrite(vp, kth, NULL);
406 }
407 if (vp)
408 vrele(vp);
409 FREE(kth, M_KTRACE);
410 FREE(cp, M_KTRACE);
411 p->p_traceflag &= ~KTRFAC_ACTIVE;
412
413 return (0);
414#else
415 return (ENOSYS);
416#endif
417}
418
419#ifdef KTRACE
420static int
421ktrops(curp, p, ops, facs, vp)
422 struct proc *p, *curp;
423 int ops, facs;
424 struct vnode *vp;
425{
426
427 if (!ktrcanset(curp, p))
428 return (0);
429 if (ops == KTROP_SET) {
430 if (p->p_tracep != vp) {
431 struct vnode *vtmp;
432
433 /*
434 * if trace file already in use, relinquish
435 */
436 VREF(vp);
437 while ((vtmp = p->p_tracep) != NULL) {
438 p->p_tracep = NULL;
439 vrele(vtmp);
440 }
441 p->p_tracep = vp;
442 }
443 p->p_traceflag |= facs;
444 if (curp->p_ucred->cr_uid == 0)
445 p->p_traceflag |= KTRFAC_ROOT;
446 } else {
447 /* KTROP_CLEAR */
448 if (((p->p_traceflag &= ~facs) & KTRFAC_MASK) == 0) {
449 struct vnode *vtmp;
450
451 /* no more tracing */
452 p->p_traceflag = 0;
453 if ((vtmp = p->p_tracep) != NULL) {
454 p->p_tracep = NULL;
455 vrele(vtmp);
456 }
457 }
458 }
459
460 return (1);
461}
462
463static int
464ktrsetchildren(curp, top, ops, facs, vp)
465 struct proc *curp, *top;
466 int ops, facs;
467 struct vnode *vp;
468{
469 register struct proc *p;
470 register int ret = 0;
471
472 p = top;
473 for (;;) {
474 ret |= ktrops(curp, p, ops, facs, vp);
475 /*
476 * If this process has children, descend to them next,
477 * otherwise do any siblings, and if done with this level,
478 * follow back up the tree (but not past top).
479 */
480 if (!LIST_EMPTY(&p->p_children))
481 p = LIST_FIRST(&p->p_children);
482 else for (;;) {
483 if (p == top)
484 return (ret);
485 if (LIST_NEXT(p, p_sibling)) {
486 p = LIST_NEXT(p, p_sibling);
487 break;
488 }
489 p = p->p_pptr;
490 }
491 }
492 /*NOTREACHED*/
493}
494
495static void
496ktrwrite(vp, kth, uio)
497 struct vnode *vp;
498 register struct ktr_header *kth;
499 struct uio *uio;
500{
501 struct uio auio;
502 struct iovec aiov[2];
503 register struct proc *p = curproc; /* XXX */
504 int error;
505
506 if (vp == NULL)
507 return;
508 auio.uio_iov = &aiov[0];
509 auio.uio_offset = 0;
510 auio.uio_segflg = UIO_SYSSPACE;
511 auio.uio_rw = UIO_WRITE;
512 aiov[0].iov_base = (caddr_t)kth;
513 aiov[0].iov_len = sizeof(struct ktr_header);
514 auio.uio_resid = sizeof(struct ktr_header);
515 auio.uio_iovcnt = 1;
516 auio.uio_procp = curproc;
517 if (kth->ktr_len > 0) {
518 auio.uio_iovcnt++;
519 aiov[1].iov_base = kth->ktr_buf;
520 aiov[1].iov_len = kth->ktr_len;
521 auio.uio_resid += kth->ktr_len;
522 if (uio != NULL)
523 kth->ktr_len += uio->uio_resid;
524 }
525 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY, p);
526 (void)VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
527 error = VOP_WRITE(vp, &auio, IO_UNIT | IO_APPEND, p->p_ucred);
528 if (error == 0 && uio != NULL) {
529 (void)VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE);
530 error = VOP_WRITE(vp, uio, IO_UNIT | IO_APPEND, p->p_ucred);
531 }
532 VOP_UNLOCK(vp, 0, p);
533 if (!error)
534 return;
535 /*
536 * If error encountered, give up tracing on this vnode. XXX what
537 * happens to the loop if vrele() blocks?
538 */
539 log(LOG_NOTICE, "ktrace write failed, errno %d, tracing stopped\n",
540 error);
541 LIST_FOREACH(p, &allproc, p_list) {
542 if (p->p_tracep == vp) {
543 p->p_tracep = NULL;
544 p->p_traceflag = 0;
545 vrele(vp);
546 }
547 }
548}
549
550/*
551 * Return true if caller has permission to set the ktracing state
552 * of target. Essentially, the target can't possess any
553 * more permissions than the caller. KTRFAC_ROOT signifies that
554 * root previously set the tracing status on the target process, and
555 * so, only root may further change it.
556 *
557 * TODO: check groups. use caller effective gid.
558 */
559static int
41c20dac 560ktrcanset(struct proc *callp, struct proc *targetp)
984263bc 561{
41c20dac
MD
562 struct ucred *caller = callp->p_ucred;
563 struct ucred *target = targetp->p_ucred;
984263bc 564
41c20dac 565 if (!PRISON_CHECK(caller, target))
984263bc 566 return (0);
41c20dac
MD
567 if ((caller->cr_uid == target->cr_ruid &&
568 target->cr_ruid == target->cr_svuid &&
569 caller->cr_rgid == target->cr_rgid && /* XXX */
570 target->cr_rgid == target->cr_svgid &&
984263bc
MD
571 (targetp->p_traceflag & KTRFAC_ROOT) == 0 &&
572 (targetp->p_flag & P_SUGID) == 0) ||
41c20dac 573 caller->cr_uid == 0)
984263bc
MD
574 return (1);
575
576 return (0);
577}
578
579#endif /* KTRACE */