kern_ktrace.c revision 54655
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: head/sys/kern/kern_ktrace.c 54655 1999-12-15 23:02:35Z eivind $
35 */
36
37#include "opt_ktrace.h"
38
39#include <sys/param.h>
40#include <sys/systm.h>
41#include <sys/sysproto.h>
42#include <sys/kernel.h>
43#include <sys/proc.h>
44#include <sys/fcntl.h>
45#include <sys/lock.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
52#include <vm/vm_zone.h>
53
54#include <stddef.h>
55
56static MALLOC_DEFINE(M_KTRACE, "KTRACE", "KTRACE");
57
58#ifdef KTRACE
59static struct ktr_header *ktrgetheader __P((int type));
60static void ktrwrite __P((struct vnode *, struct ktr_header *));
61static int ktrcanset __P((struct proc *,struct proc *));
62static int ktrsetchildren __P((struct proc *,struct proc *,int,int,struct vnode *));
63static int ktrops __P((struct proc *,struct proc *,int,int,struct vnode *));
64
65
66static struct ktr_header *
67ktrgetheader(type)
68	int type;
69{
70	register struct ktr_header *kth;
71	struct proc *p = curproc;	/* XXX */
72
73	MALLOC(kth, struct ktr_header *, sizeof (struct ktr_header),
74		M_KTRACE, M_WAITOK);
75	kth->ktr_type = type;
76	microtime(&kth->ktr_time);
77	kth->ktr_pid = p->p_pid;
78	bcopy(p->p_comm, kth->ktr_comm, MAXCOMLEN + 1);
79	return (kth);
80}
81
82void
83ktrsyscall(vp, code, narg, args)
84	struct vnode *vp;
85	int code, narg;
86	register_t args[];
87{
88	struct	ktr_header *kth;
89	struct	ktr_syscall *ktp;
90	register int len = offsetof(struct ktr_syscall, ktr_args) +
91	    (narg * sizeof(register_t));
92	struct proc *p = curproc;	/* XXX */
93	register_t *argp;
94	int i;
95
96	p->p_traceflag |= KTRFAC_ACTIVE;
97	kth = ktrgetheader(KTR_SYSCALL);
98	MALLOC(ktp, struct ktr_syscall *, len, M_KTRACE, M_WAITOK);
99	ktp->ktr_code = code;
100	ktp->ktr_narg = narg;
101	argp = &ktp->ktr_args[0];
102	for (i = 0; i < narg; i++)
103		*argp++ = args[i];
104	kth->ktr_buf = (caddr_t)ktp;
105	kth->ktr_len = len;
106	ktrwrite(vp, kth);
107	FREE(ktp, M_KTRACE);
108	FREE(kth, M_KTRACE);
109	p->p_traceflag &= ~KTRFAC_ACTIVE;
110}
111
112void
113ktrsysret(vp, code, error, retval)
114	struct vnode *vp;
115	int code, error;
116	register_t retval;
117{
118	struct ktr_header *kth;
119	struct ktr_sysret ktp;
120	struct proc *p = curproc;	/* XXX */
121
122	p->p_traceflag |= KTRFAC_ACTIVE;
123	kth = ktrgetheader(KTR_SYSRET);
124	ktp.ktr_code = code;
125	ktp.ktr_error = error;
126	ktp.ktr_retval = retval;		/* what about val2 ? */
127
128	kth->ktr_buf = (caddr_t)&ktp;
129	kth->ktr_len = sizeof(struct ktr_sysret);
130
131	ktrwrite(vp, kth);
132	FREE(kth, M_KTRACE);
133	p->p_traceflag &= ~KTRFAC_ACTIVE;
134}
135
136void
137ktrnamei(vp, path)
138	struct vnode *vp;
139	char *path;
140{
141	struct ktr_header *kth;
142	struct proc *p = curproc;	/* XXX */
143
144	p->p_traceflag |= KTRFAC_ACTIVE;
145	kth = ktrgetheader(KTR_NAMEI);
146	kth->ktr_len = strlen(path);
147	kth->ktr_buf = path;
148
149	ktrwrite(vp, kth);
150	FREE(kth, M_KTRACE);
151	p->p_traceflag &= ~KTRFAC_ACTIVE;
152}
153
154void
155ktrgenio(vp, fd, rw, iov, len, error)
156	struct vnode *vp;
157	int fd;
158	enum uio_rw rw;
159	register struct iovec *iov;
160	int len, error;
161{
162	struct ktr_header *kth;
163	register struct ktr_genio *ktp;
164	register caddr_t cp;
165	register int resid = len, cnt;
166	struct proc *p = curproc;	/* XXX */
167
168	if (error)
169		return;
170	p->p_traceflag |= KTRFAC_ACTIVE;
171	kth = ktrgetheader(KTR_GENIO);
172	MALLOC(ktp, struct ktr_genio *, sizeof(struct ktr_genio) + len,
173		M_KTRACE, M_WAITOK);
174	ktp->ktr_fd = fd;
175	ktp->ktr_rw = rw;
176	cp = (caddr_t)((char *)ktp + sizeof (struct ktr_genio));
177	while (resid > 0) {
178		if ((cnt = iov->iov_len) > resid)
179			cnt = resid;
180		if (copyin(iov->iov_base, cp, (unsigned)cnt))
181			goto done;
182		cp += cnt;
183		resid -= cnt;
184		iov++;
185	}
186	kth->ktr_buf = (caddr_t)ktp;
187	kth->ktr_len = sizeof (struct ktr_genio) + len;
188
189	ktrwrite(vp, kth);
190done:
191	FREE(kth, M_KTRACE);
192	FREE(ktp, M_KTRACE);
193	p->p_traceflag &= ~KTRFAC_ACTIVE;
194}
195
196void
197ktrpsig(vp, sig, action, mask, code)
198	struct vnode *vp;
199	int sig;
200	sig_t action;
201	sigset_t *mask;
202	int code;
203{
204	struct ktr_header *kth;
205	struct ktr_psig	kp;
206	struct proc *p = curproc;	/* XXX */
207
208	p->p_traceflag |= KTRFAC_ACTIVE;
209	kth = ktrgetheader(KTR_PSIG);
210	kp.signo = (char)sig;
211	kp.action = action;
212	kp.mask = *mask;
213	kp.code = code;
214	kth->ktr_buf = (caddr_t)&kp;
215	kth->ktr_len = sizeof (struct ktr_psig);
216
217	ktrwrite(vp, kth);
218	FREE(kth, M_KTRACE);
219	p->p_traceflag &= ~KTRFAC_ACTIVE;
220}
221
222void
223ktrcsw(vp, out, user)
224	struct vnode *vp;
225	int out, user;
226{
227	struct ktr_header *kth;
228	struct	ktr_csw kc;
229	struct proc *p = curproc;	/* XXX */
230
231	p->p_traceflag |= KTRFAC_ACTIVE;
232	kth = ktrgetheader(KTR_CSW);
233	kc.out = out;
234	kc.user = user;
235	kth->ktr_buf = (caddr_t)&kc;
236	kth->ktr_len = sizeof (struct ktr_csw);
237
238	ktrwrite(vp, kth);
239	FREE(kth, M_KTRACE);
240	p->p_traceflag &= ~KTRFAC_ACTIVE;
241}
242#endif
243
244/* Interface and common routines */
245
246/*
247 * ktrace system call
248 */
249#ifndef _SYS_SYSPROTO_H_
250struct ktrace_args {
251	char	*fname;
252	int	ops;
253	int	facs;
254	int	pid;
255};
256#endif
257/* ARGSUSED */
258int
259ktrace(curp, uap)
260	struct proc *curp;
261	register struct ktrace_args *uap;
262{
263#ifdef KTRACE
264	register struct vnode *vp = NULL;
265	register struct proc *p;
266	struct pgrp *pg;
267	int facs = uap->facs & ~KTRFAC_ROOT;
268	int ops = KTROP(uap->ops);
269	int descend = uap->ops & KTRFLAG_DESCEND;
270	int ret = 0;
271	int error = 0;
272	struct nameidata nd;
273
274	curp->p_traceflag |= KTRFAC_ACTIVE;
275	if (ops != KTROP_CLEAR) {
276		/*
277		 * an operation which requires a file argument.
278		 */
279		NDINIT(&nd, LOOKUP, NOFOLLOW, UIO_USERSPACE, uap->fname, curp);
280		error = vn_open(&nd, FREAD|FWRITE|O_NOFOLLOW, 0);
281		if (error) {
282			curp->p_traceflag &= ~KTRFAC_ACTIVE;
283			return (error);
284		}
285		NDFREE(&nd, NDF_ONLY_PNBUF);
286		vp = nd.ni_vp;
287		VOP_UNLOCK(vp, 0, curp);
288		if (vp->v_type != VREG) {
289			(void) vn_close(vp, FREAD|FWRITE, curp->p_ucred, curp);
290			curp->p_traceflag &= ~KTRFAC_ACTIVE;
291			return (EACCES);
292		}
293	}
294	/*
295	 * Clear all uses of the tracefile
296	 */
297	if (ops == KTROP_CLEARFILE) {
298		LIST_FOREACH(p, &allproc, p_list) {
299			if (p->p_tracep == vp) {
300				if (ktrcanset(curp, p)) {
301					p->p_tracep = NULL;
302					p->p_traceflag = 0;
303					(void) vn_close(vp, FREAD|FWRITE,
304						p->p_ucred, p);
305				} else
306					error = EPERM;
307			}
308		}
309		goto done;
310	}
311	/*
312	 * need something to (un)trace (XXX - why is this here?)
313	 */
314	if (!facs) {
315		error = EINVAL;
316		goto done;
317	}
318	/*
319	 * do it
320	 */
321	if (uap->pid < 0) {
322		/*
323		 * by process group
324		 */
325		pg = pgfind(-uap->pid);
326		if (pg == NULL) {
327			error = ESRCH;
328			goto done;
329		}
330		LIST_FOREACH(p, &pg->pg_members, p_pglist)
331			if (descend)
332				ret |= ktrsetchildren(curp, p, ops, facs, vp);
333			else
334				ret |= ktrops(curp, p, ops, facs, vp);
335
336	} else {
337		/*
338		 * by pid
339		 */
340		p = pfind(uap->pid);
341		if (p == NULL) {
342			error = ESRCH;
343			goto done;
344		}
345		if (descend)
346			ret |= ktrsetchildren(curp, p, ops, facs, vp);
347		else
348			ret |= ktrops(curp, p, ops, facs, vp);
349	}
350	if (!ret)
351		error = EPERM;
352done:
353	if (vp != NULL)
354		(void) vn_close(vp, FWRITE, curp->p_ucred, curp);
355	curp->p_traceflag &= ~KTRFAC_ACTIVE;
356	return (error);
357#else
358	return ENOSYS;
359#endif
360}
361
362/*
363 * utrace system call
364 */
365/* ARGSUSED */
366int
367utrace(curp, uap)
368	struct proc *curp;
369	register struct utrace_args *uap;
370{
371#ifdef KTRACE
372	struct ktr_header *kth;
373	struct proc *p = curproc;	/* XXX */
374	register caddr_t cp;
375
376	if (!KTRPOINT(p, KTR_USER))
377		return (0);
378	p->p_traceflag |= KTRFAC_ACTIVE;
379	kth = ktrgetheader(KTR_USER);
380	MALLOC(cp, caddr_t, uap->len, M_KTRACE, M_WAITOK);
381	if (!copyin(uap->addr, cp, uap->len)) {
382		kth->ktr_buf = cp;
383		kth->ktr_len = uap->len;
384		ktrwrite(p->p_tracep, kth);
385	}
386	FREE(kth, M_KTRACE);
387	FREE(cp, M_KTRACE);
388	p->p_traceflag &= ~KTRFAC_ACTIVE;
389
390	return (0);
391#else
392	return (ENOSYS);
393#endif
394}
395
396#ifdef KTRACE
397static int
398ktrops(curp, p, ops, facs, vp)
399	struct proc *p, *curp;
400	int ops, facs;
401	struct vnode *vp;
402{
403
404	if (!ktrcanset(curp, p))
405		return (0);
406	if (ops == KTROP_SET) {
407		if (p->p_tracep != vp) {
408			/*
409			 * if trace file already in use, relinquish
410			 */
411			if (p->p_tracep != NULL)
412				vrele(p->p_tracep);
413			VREF(vp);
414			p->p_tracep = vp;
415		}
416		p->p_traceflag |= facs;
417		if (curp->p_ucred->cr_uid == 0)
418			p->p_traceflag |= KTRFAC_ROOT;
419	} else {
420		/* KTROP_CLEAR */
421		if (((p->p_traceflag &= ~facs) & KTRFAC_MASK) == 0) {
422			/* no more tracing */
423			p->p_traceflag = 0;
424			if (p->p_tracep != NULL) {
425				vrele(p->p_tracep);
426				p->p_tracep = NULL;
427			}
428		}
429	}
430
431	return (1);
432}
433
434static int
435ktrsetchildren(curp, top, ops, facs, vp)
436	struct proc *curp, *top;
437	int ops, facs;
438	struct vnode *vp;
439{
440	register struct proc *p;
441	register int ret = 0;
442
443	p = top;
444	for (;;) {
445		ret |= ktrops(curp, p, ops, facs, vp);
446		/*
447		 * If this process has children, descend to them next,
448		 * otherwise do any siblings, and if done with this level,
449		 * follow back up the tree (but not past top).
450		 */
451		if (!LIST_EMPTY(&p->p_children))
452			p = LIST_FIRST(&p->p_children);
453		else for (;;) {
454			if (p == top)
455				return (ret);
456			if (LIST_NEXT(p, p_sibling)) {
457				p = LIST_NEXT(p, p_sibling);
458				break;
459			}
460			p = p->p_pptr;
461		}
462	}
463	/*NOTREACHED*/
464}
465
466static void
467ktrwrite(vp, kth)
468	struct vnode *vp;
469	register struct ktr_header *kth;
470{
471	struct uio auio;
472	struct iovec aiov[2];
473	register struct proc *p = curproc;	/* XXX */
474	int error;
475
476	if (vp == NULL)
477		return;
478	auio.uio_iov = &aiov[0];
479	auio.uio_offset = 0;
480	auio.uio_segflg = UIO_SYSSPACE;
481	auio.uio_rw = UIO_WRITE;
482	aiov[0].iov_base = (caddr_t)kth;
483	aiov[0].iov_len = sizeof(struct ktr_header);
484	auio.uio_resid = sizeof(struct ktr_header);
485	auio.uio_iovcnt = 1;
486	auio.uio_procp = curproc;
487	if (kth->ktr_len > 0) {
488		auio.uio_iovcnt++;
489		aiov[1].iov_base = kth->ktr_buf;
490		aiov[1].iov_len = kth->ktr_len;
491		auio.uio_resid += kth->ktr_len;
492	}
493	vn_lock(vp, LK_EXCLUSIVE | LK_RETRY, p);
494	error = VOP_WRITE(vp, &auio, IO_UNIT|IO_APPEND, p->p_ucred);
495	VOP_UNLOCK(vp, 0, p);
496	if (!error)
497		return;
498	/*
499	 * If error encountered, give up tracing on this vnode.
500	 */
501	log(LOG_NOTICE, "ktrace write failed, errno %d, tracing stopped\n",
502	    error);
503	LIST_FOREACH(p, &allproc, p_list) {
504		if (p->p_tracep == vp) {
505			p->p_tracep = NULL;
506			p->p_traceflag = 0;
507			vrele(vp);
508		}
509	}
510}
511
512/*
513 * Return true if caller has permission to set the ktracing state
514 * of target.  Essentially, the target can't possess any
515 * more permissions than the caller.  KTRFAC_ROOT signifies that
516 * root previously set the tracing status on the target process, and
517 * so, only root may further change it.
518 *
519 * TODO: check groups.  use caller effective gid.
520 */
521static int
522ktrcanset(callp, targetp)
523	struct proc *callp, *targetp;
524{
525	register struct pcred *caller = callp->p_cred;
526	register struct pcred *target = targetp->p_cred;
527
528	if (!PRISON_CHECK(callp, targetp))
529		return (0);
530	if ((caller->pc_ucred->cr_uid == target->p_ruid &&
531	     target->p_ruid == target->p_svuid &&
532	     caller->p_rgid == target->p_rgid &&	/* XXX */
533	     target->p_rgid == target->p_svgid &&
534	     (targetp->p_traceflag & KTRFAC_ROOT) == 0) ||
535	     caller->pc_ucred->cr_uid == 0)
536		return (1);
537
538	return (0);
539}
540
541#endif /* KTRACE */
542