2 * Copyright (c) 1989, 1993
3 * The Regents of the University of California. All rights reserved.
5 * Redistribution and use in source and binary forms, with or without
6 * modification, are permitted provided that the following conditions
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.
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
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 $
37 #include "opt_ktrace.h"
39 #include <sys/param.h>
40 #include <sys/systm.h>
41 #include <sys/sysproto.h>
42 #include <sys/kernel.h>
44 #include <sys/fcntl.h>
46 #include <sys/nlookup.h>
47 #include <sys/vnode.h>
48 #include <sys/ktrace.h>
49 #include <sys/malloc.h>
50 #include <sys/syslog.h>
51 #include <sys/sysent.h>
53 #include <vm/vm_zone.h>
55 #include <sys/mplock2.h>
57 static MALLOC_DEFINE(M_KTRACE, "KTRACE", "KTRACE");
60 static struct ktr_header *ktrgetheader (int type);
61 static void ktrwrite (struct lwp *, struct ktr_header *, struct uio *);
62 static int ktrcanset (struct thread *,struct proc *);
63 static int ktrsetchildren (struct thread *,struct proc *,int,int, ktrace_node_t);
64 static int ktrops (struct thread *,struct proc *,int,int, ktrace_node_t);
69 static struct ktr_header *
70 ktrgetheader(int type)
72 struct ktr_header *kth;
73 struct proc *p = curproc; /* XXX */
74 struct lwp *lp = curthread->td_lwp;
76 kth = kmalloc(sizeof(struct ktr_header), M_KTRACE, M_WAITOK);
78 /* XXX threaded flag is a hack at the moment */
79 kth->ktr_flags = (p->p_nthreads > 1) ? KTRH_THREADED : 0;
80 microtime(&kth->ktr_time);
81 kth->ktr_pid = p->p_pid;
82 kth->ktr_tid = lp->lwp_tid;
83 bcopy(p->p_comm, kth->ktr_comm, MAXCOMLEN + 1);
88 ktrsyscall(struct lwp *lp, int code, int narg, register_t args[])
90 struct ktr_header *kth;
91 struct ktr_syscall *ktp;
96 len = offsetof(struct ktr_syscall, ktr_args) +
97 (narg * sizeof(register_t));
100 * Setting the active bit prevents a ktrace recursion from the
101 * ktracing op itself.
103 lp->lwp_traceflag |= KTRFAC_ACTIVE;
104 kth = ktrgetheader(KTR_SYSCALL);
105 ktp = kmalloc(len, M_KTRACE, M_WAITOK);
106 ktp->ktr_code = code;
107 ktp->ktr_narg = narg;
108 argp = &ktp->ktr_args[0];
109 for (i = 0; i < narg; i++)
111 kth->ktr_buf = (caddr_t)ktp;
113 ktrwrite(lp, kth, NULL);
114 kfree(ktp, M_KTRACE);
115 kfree(kth, M_KTRACE);
116 lp->lwp_traceflag &= ~KTRFAC_ACTIVE;
120 ktrsysret(struct lwp *lp, int code, int error, register_t retval)
122 struct ktr_header *kth;
123 struct ktr_sysret ktp;
125 lp->lwp_traceflag |= KTRFAC_ACTIVE;
126 kth = ktrgetheader(KTR_SYSRET);
128 ktp.ktr_error = error;
130 ktp.ktr_retval = retval; /* what about val2 ? */
134 kth->ktr_buf = (caddr_t)&ktp;
135 kth->ktr_len = sizeof(struct ktr_sysret);
137 ktrwrite(lp, kth, NULL);
138 kfree(kth, M_KTRACE);
139 lp->lwp_traceflag &= ~KTRFAC_ACTIVE;
143 ktrnamei(struct lwp *lp, char *path)
145 struct ktr_header *kth;
147 lp->lwp_traceflag |= KTRFAC_ACTIVE;
148 kth = ktrgetheader(KTR_NAMEI);
149 kth->ktr_len = strlen(path);
152 ktrwrite(lp, kth, NULL);
153 kfree(kth, M_KTRACE);
154 lp->lwp_traceflag &= ~KTRFAC_ACTIVE;
158 ktrgenio(struct lwp *lp, int fd, enum uio_rw rw, struct uio *uio, int error)
160 struct ktr_header *kth;
161 struct ktr_genio ktg;
165 lp->lwp_traceflag |= KTRFAC_ACTIVE;
166 kth = ktrgetheader(KTR_GENIO);
169 kth->ktr_buf = (caddr_t)&ktg;
170 kth->ktr_len = sizeof(struct ktr_genio);
172 uio->uio_rw = UIO_WRITE;
174 ktrwrite(lp, kth, uio);
175 kfree(kth, M_KTRACE);
176 lp->lwp_traceflag &= ~KTRFAC_ACTIVE;
180 ktrpsig(struct lwp *lp, int sig, sig_t action, sigset_t *mask, int code)
182 struct ktr_header *kth;
185 lp->lwp_traceflag |= KTRFAC_ACTIVE;
186 kth = ktrgetheader(KTR_PSIG);
187 kp.signo = (char)sig;
191 kth->ktr_buf = (caddr_t)&kp;
192 kth->ktr_len = sizeof (struct ktr_psig);
194 ktrwrite(lp, kth, NULL);
195 kfree(kth, M_KTRACE);
196 lp->lwp_traceflag &= ~KTRFAC_ACTIVE;
200 ktrcsw(struct lwp *lp, int out, int user)
202 struct ktr_header *kth;
205 lp->lwp_traceflag |= KTRFAC_ACTIVE;
206 kth = ktrgetheader(KTR_CSW);
209 kth->ktr_buf = (caddr_t)&kc;
210 kth->ktr_len = sizeof (struct ktr_csw);
212 ktrwrite(lp, kth, NULL);
213 kfree(kth, M_KTRACE);
214 lp->lwp_traceflag &= ~KTRFAC_ACTIVE;
218 /* Interface and common routines */
224 struct ktrace_clear_info {
225 ktrace_node_t tracenode;
230 static int ktrace_clear_callback(struct proc *p, void *data);
238 sys_ktrace(struct ktrace_args *uap)
241 struct ktrace_clear_info info;
242 struct thread *td = curthread;
243 struct proc *curp = td->td_proc;
246 int facs = uap->facs & ~KTRFAC_ROOT;
247 int ops = KTROP(uap->ops);
248 int descend = uap->ops & KTRFLAG_DESCEND;
251 struct nlookupdata nd;
252 ktrace_node_t tracenode = NULL;
255 curp->p_traceflag |= KTRFAC_ACTIVE;
256 if (ops != KTROP_CLEAR) {
258 * an operation which requires a file argument.
260 error = nlookup_init(&nd, uap->fname,
261 UIO_USERSPACE, NLC_LOCKVP);
263 error = vn_open(&nd, NULL, FREAD|FWRITE|O_NOFOLLOW, 0);
264 if (error == 0 && nd.nl_open_vp->v_type != VREG)
267 curp->p_traceflag &= ~KTRFAC_ACTIVE;
271 tracenode = kmalloc(sizeof(struct ktrace_node), M_KTRACE,
273 tracenode->kn_vp = nd.nl_open_vp;
274 tracenode->kn_refs = 1;
275 nd.nl_open_vp = NULL;
277 vn_unlock(tracenode->kn_vp);
280 * Clear all uses of the tracefile. Not the most efficient operation
283 if (ops == KTROP_CLEARFILE) {
284 info.tracenode = tracenode;
287 allproc_scan(ktrace_clear_callback, &info);
292 * need something to (un)trace (XXX - why is this here?)
303 * By process group. Process group is referenced, preventing
306 pg = pgfind(-uap->pid);
311 lwkt_gettoken(&pg->pg_token);
312 LIST_FOREACH(p, &pg->pg_members, p_pglist) {
315 ret |= ktrsetchildren(td, p, ops, facs, tracenode);
317 ret |= ktrops(td, p, ops, facs, tracenode);
320 lwkt_reltoken(&pg->pg_token);
332 ret |= ktrsetchildren(td, p, ops, facs, tracenode);
334 ret |= ktrops(td, p, ops, facs, tracenode);
341 ktrdestroy(&tracenode);
342 curp->p_traceflag &= ~KTRFAC_ACTIVE;
353 * NOTE: NOT MPSAFE (yet)
356 ktrace_clear_callback(struct proc *p, void *data)
358 struct ktrace_clear_info *info = data;
360 if (p->p_tracenode) {
361 if (info->rootclear) {
362 if (p->p_tracenode == info->tracenode) {
363 ktrdestroy(&p->p_tracenode);
367 if (p->p_tracenode->kn_vp == info->tracenode->kn_vp) {
368 if (ktrcanset(curthread, p)) {
369 ktrdestroy(&p->p_tracenode);
388 sys_utrace(struct utrace_args *uap)
391 struct ktr_header *kth;
392 struct thread *td = curthread; /* XXX */
395 if (!KTRPOINT(td, KTR_USER))
397 if (uap->len > KTR_USER_MAXLEN)
399 td->td_lwp->lwp_traceflag |= KTRFAC_ACTIVE;
400 kth = ktrgetheader(KTR_USER);
401 cp = kmalloc(uap->len, M_KTRACE, M_WAITOK);
402 if (!copyin(uap->addr, cp, uap->len)) {
404 kth->ktr_len = uap->len;
405 ktrwrite(td->td_lwp, kth, NULL);
407 kfree(kth, M_KTRACE);
409 td->td_lwp->lwp_traceflag &= ~KTRFAC_ACTIVE;
418 ktrdestroy(struct ktrace_node **tracenodep)
420 ktrace_node_t tracenode;
422 if ((tracenode = *tracenodep) != NULL) {
424 KKASSERT(tracenode->kn_refs > 0);
425 if (atomic_fetchadd_int(&tracenode->kn_refs, -1) == 1) {
426 vn_close(tracenode->kn_vp, FREAD|FWRITE);
427 tracenode->kn_vp = NULL;
428 kfree(tracenode, M_KTRACE);
434 * This allows a process to inherit a ref on a tracenode and is also used
435 * as a temporary ref to prevent a tracenode from being destroyed out from
436 * under an active operation.
439 ktrinherit(ktrace_node_t tracenode)
442 KKASSERT(tracenode->kn_refs > 0);
443 atomic_add_int(&tracenode->kn_refs, 1);
450 ktrops(struct thread *td, struct proc *p, int ops, int facs,
451 ktrace_node_t tracenode)
453 ktrace_node_t oldnode;
455 if (!ktrcanset(td, p))
457 if (ops == KTROP_SET) {
458 if ((oldnode = p->p_tracenode) != tracenode) {
459 p->p_tracenode = ktrinherit(tracenode);
460 ktrdestroy(&oldnode);
462 p->p_traceflag |= facs;
463 if (td->td_ucred->cr_uid == 0)
464 p->p_traceflag |= KTRFAC_ROOT;
467 if (((p->p_traceflag &= ~facs) & KTRFAC_MASK) == 0) {
468 /* no more tracing */
470 ktrdestroy(&p->p_tracenode);
478 ktrsetchildren(struct thread *td, struct proc *top, int ops, int facs,
479 ktrace_node_t tracenode)
487 lwkt_gettoken(&p->p_token);
490 ret |= ktrops(td, p, ops, facs, tracenode);
493 * If this process has children, descend to them next,
494 * otherwise do any siblings, and if done with this level,
495 * follow back up the tree (but not past top).
497 if ((np = LIST_FIRST(&p->p_children)) != NULL) {
503 if ((np = LIST_NEXT(p, p_sibling)) != NULL) {
509 * recurse up to parent, set p in our inner
510 * loop when doing this. np can be NULL if
511 * we race a reparenting to init (thus 'top'
512 * is skipped past and never encountered).
518 lwkt_reltoken(&p->p_token);
521 lwkt_gettoken(&p->p_token);
524 lwkt_reltoken(&p->p_token);
529 /* Already held, but we need the token too */
530 lwkt_gettoken(&p->p_token);
536 ktrwrite(struct lwp *lp, struct ktr_header *kth, struct uio *uio)
538 struct ktrace_clear_info info;
540 struct iovec aiov[2];
542 ktrace_node_t tracenode;
545 * We have to ref our tracenode to prevent it from being ripped out
546 * from under us while we are trying to use it. p_tracenode can
547 * go away at any time if another process gets a write error.
551 if (lp->lwp_proc->p_tracenode == NULL)
553 tracenode = ktrinherit(lp->lwp_proc->p_tracenode);
554 auio.uio_iov = &aiov[0];
556 auio.uio_segflg = UIO_SYSSPACE;
557 auio.uio_rw = UIO_WRITE;
558 aiov[0].iov_base = (caddr_t)kth;
559 aiov[0].iov_len = sizeof(struct ktr_header);
560 auio.uio_resid = sizeof(struct ktr_header);
562 auio.uio_td = curthread;
563 if (kth->ktr_len > 0) {
565 aiov[1].iov_base = kth->ktr_buf;
566 aiov[1].iov_len = kth->ktr_len;
567 auio.uio_resid += kth->ktr_len;
569 kth->ktr_len += uio->uio_resid;
571 vn_lock(tracenode->kn_vp, LK_EXCLUSIVE | LK_RETRY);
572 error = VOP_WRITE(tracenode->kn_vp, &auio,
573 IO_UNIT | IO_APPEND, lp->lwp_thread->td_ucred);
574 if (error == 0 && uio != NULL) {
575 error = VOP_WRITE(tracenode->kn_vp, uio,
576 IO_UNIT | IO_APPEND, lp->lwp_thread->td_ucred);
578 vn_unlock(tracenode->kn_vp);
581 * If an error occured, give up tracing on all processes
582 * using this tracenode. This is not MP safe but is
586 "ktrace write failed, errno %d, tracing stopped\n", error);
587 info.tracenode = tracenode;
590 allproc_scan(ktrace_clear_callback, &info);
592 ktrdestroy(&tracenode);
596 * Return true if caller has permission to set the ktracing state
597 * of target. Essentially, the target can't possess any
598 * more permissions than the caller. KTRFAC_ROOT signifies that
599 * root previously set the tracing status on the target process, and
600 * so, only root may further change it.
602 * TODO: check groups. use caller effective gid.
605 ktrcanset(struct thread *calltd, struct proc *targetp)
607 struct ucred *caller = calltd->td_ucred;
608 struct ucred *target = targetp->p_ucred;
610 if (!PRISON_CHECK(caller, target))
612 if ((caller->cr_uid == target->cr_ruid &&
613 target->cr_ruid == target->cr_svuid &&
614 caller->cr_rgid == target->cr_rgid && /* XXX */
615 target->cr_rgid == target->cr_svgid &&
616 (targetp->p_traceflag & KTRFAC_ROOT) == 0 &&
617 (targetp->p_flags & P_SUGID) == 0) ||