kern_ktrace.c revision 70707
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 70707 2001-01-06 09:34:20Z alfred $ 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#include <sys/sysent.h> 52 53static MALLOC_DEFINE(M_KTRACE, "KTRACE", "KTRACE"); 54 55#ifdef KTRACE 56static struct ktr_header *ktrgetheader __P((int type)); 57static void ktrwrite __P((struct vnode *, struct ktr_header *, struct uio *)); 58static int ktrcanset __P((struct proc *,struct proc *)); 59static int ktrsetchildren __P((struct proc *,struct proc *,int,int,struct vnode *)); 60static int ktrops __P((struct proc *,struct proc *,int,int,struct vnode *)); 61 62 63static struct ktr_header * 64ktrgetheader(type) 65 int type; 66{ 67 register struct ktr_header *kth; 68 struct proc *p = curproc; /* XXX */ 69 70 MALLOC(kth, struct ktr_header *, sizeof (struct ktr_header), 71 M_KTRACE, M_WAITOK); 72 kth->ktr_type = type; 73 microtime(&kth->ktr_time); 74 kth->ktr_pid = p->p_pid; 75 bcopy(p->p_comm, kth->ktr_comm, MAXCOMLEN + 1); 76 return (kth); 77} 78 79void 80ktrsyscall(vp, code, narg, args) 81 struct vnode *vp; 82 int code, narg; 83 register_t args[]; 84{ 85 struct ktr_header *kth; 86 struct ktr_syscall *ktp; 87 register int len = offsetof(struct ktr_syscall, ktr_args) + 88 (narg * sizeof(register_t)); 89 struct proc *p = curproc; /* XXX */ 90 register_t *argp; 91 int i; 92 93 p->p_traceflag |= KTRFAC_ACTIVE; 94 kth = ktrgetheader(KTR_SYSCALL); 95 MALLOC(ktp, struct ktr_syscall *, len, M_KTRACE, M_WAITOK); 96 ktp->ktr_code = code; 97 ktp->ktr_narg = narg; 98 argp = &ktp->ktr_args[0]; 99 for (i = 0; i < narg; i++) 100 *argp++ = args[i]; 101 kth->ktr_buffer = (caddr_t)ktp; 102 kth->ktr_len = len; 103 ktrwrite(vp, kth, NULL); 104 FREE(ktp, M_KTRACE); 105 FREE(kth, M_KTRACE); 106 p->p_traceflag &= ~KTRFAC_ACTIVE; 107} 108 109void 110ktrsysret(vp, code, error, retval) 111 struct vnode *vp; 112 int code, error; 113 register_t retval; 114{ 115 struct ktr_header *kth; 116 struct ktr_sysret ktp; 117 struct proc *p = curproc; /* XXX */ 118 119 p->p_traceflag |= KTRFAC_ACTIVE; 120 kth = ktrgetheader(KTR_SYSRET); 121 ktp.ktr_code = code; 122 ktp.ktr_error = error; 123 ktp.ktr_retval = retval; /* what about val2 ? */ 124 125 kth->ktr_buffer = (caddr_t)&ktp; 126 kth->ktr_len = sizeof(struct ktr_sysret); 127 128 ktrwrite(vp, kth, NULL); 129 FREE(kth, M_KTRACE); 130 p->p_traceflag &= ~KTRFAC_ACTIVE; 131} 132 133void 134ktrnamei(vp, path) 135 struct vnode *vp; 136 char *path; 137{ 138 struct ktr_header *kth; 139 struct proc *p = curproc; /* XXX */ 140 141 p->p_traceflag |= KTRFAC_ACTIVE; 142 kth = ktrgetheader(KTR_NAMEI); 143 kth->ktr_len = strlen(path); 144 kth->ktr_buffer = path; 145 146 ktrwrite(vp, kth, NULL); 147 FREE(kth, M_KTRACE); 148 p->p_traceflag &= ~KTRFAC_ACTIVE; 149} 150 151void 152ktrgenio(vp, fd, rw, uio, error) 153 struct vnode *vp; 154 int fd; 155 enum uio_rw rw; 156 struct uio *uio; 157 int error; 158{ 159 struct ktr_header *kth; 160 struct ktr_genio ktg; 161 struct proc *p = curproc; /* XXX */ 162 163 if (error) 164 return; 165 p->p_traceflag |= KTRFAC_ACTIVE; 166 kth = ktrgetheader(KTR_GENIO); 167 ktg.ktr_fd = fd; 168 ktg.ktr_rw = rw; 169 kth->ktr_buffer = (caddr_t)&ktg; 170 kth->ktr_len = sizeof(struct ktr_genio); 171 uio->uio_offset = 0; 172 uio->uio_rw = UIO_WRITE; 173 174 ktrwrite(vp, kth, uio); 175 FREE(kth, M_KTRACE); 176 p->p_traceflag &= ~KTRFAC_ACTIVE; 177} 178 179void 180ktrpsig(vp, sig, action, mask, code) 181 struct vnode *vp; 182 int sig; 183 sig_t action; 184 sigset_t *mask; 185 int code; 186{ 187 struct ktr_header *kth; 188 struct ktr_psig kp; 189 struct proc *p = curproc; /* XXX */ 190 191 p->p_traceflag |= KTRFAC_ACTIVE; 192 kth = ktrgetheader(KTR_PSIG); 193 kp.signo = (char)sig; 194 kp.action = action; 195 kp.mask = *mask; 196 kp.code = code; 197 kth->ktr_buffer = (caddr_t)&kp; 198 kth->ktr_len = sizeof (struct ktr_psig); 199 200 ktrwrite(vp, kth, NULL); 201 FREE(kth, M_KTRACE); 202 p->p_traceflag &= ~KTRFAC_ACTIVE; 203} 204 205void 206ktrcsw(vp, out, user) 207 struct vnode *vp; 208 int out, user; 209{ 210 struct ktr_header *kth; 211 struct ktr_csw kc; 212 struct proc *p = curproc; /* XXX */ 213 214 p->p_traceflag |= KTRFAC_ACTIVE; 215 kth = ktrgetheader(KTR_CSW); 216 kc.out = out; 217 kc.user = user; 218 kth->ktr_buffer = (caddr_t)&kc; 219 kth->ktr_len = sizeof (struct ktr_csw); 220 221 ktrwrite(vp, kth, NULL); 222 FREE(kth, M_KTRACE); 223 p->p_traceflag &= ~KTRFAC_ACTIVE; 224} 225#endif 226 227/* Interface and common routines */ 228 229/* 230 * ktrace system call 231 */ 232#ifndef _SYS_SYSPROTO_H_ 233struct ktrace_args { 234 char *fname; 235 int ops; 236 int facs; 237 int pid; 238}; 239#endif 240/* ARGSUSED */ 241int 242ktrace(curp, uap) 243 struct proc *curp; 244 register struct ktrace_args *uap; 245{ 246#ifdef KTRACE 247 register struct vnode *vp = NULL; 248 register struct proc *p; 249 struct pgrp *pg; 250 int facs = uap->facs & ~KTRFAC_ROOT; 251 int ops = KTROP(uap->ops); 252 int descend = uap->ops & KTRFLAG_DESCEND; 253 int ret = 0; 254 int flags, error = 0; 255 struct nameidata nd; 256 257 curp->p_traceflag |= KTRFAC_ACTIVE; 258 if (ops != KTROP_CLEAR) { 259 /* 260 * an operation which requires a file argument. 261 */ 262 NDINIT(&nd, LOOKUP, NOFOLLOW, UIO_USERSPACE, uap->fname, curp); 263 flags = FREAD | FWRITE | O_NOFOLLOW; 264 error = vn_open(&nd, &flags, 0); 265 if (error) { 266 curp->p_traceflag &= ~KTRFAC_ACTIVE; 267 return (error); 268 } 269 NDFREE(&nd, NDF_ONLY_PNBUF); 270 vp = nd.ni_vp; 271 VOP_UNLOCK(vp, 0, curp); 272 if (vp->v_type != VREG) { 273 (void) vn_close(vp, FREAD|FWRITE, curp->p_ucred, curp); 274 curp->p_traceflag &= ~KTRFAC_ACTIVE; 275 return (EACCES); 276 } 277 } 278 /* 279 * Clear all uses of the tracefile 280 */ 281 if (ops == KTROP_CLEARFILE) { 282 ALLPROC_LOCK(AP_SHARED); 283 LIST_FOREACH(p, &allproc, p_list) { 284 if (p->p_tracep == vp) { 285 if (ktrcanset(curp, p)) { 286 p->p_tracep = NULL; 287 p->p_traceflag = 0; 288 (void) vn_close(vp, FREAD|FWRITE, 289 p->p_ucred, p); 290 } else 291 error = EPERM; 292 } 293 } 294 ALLPROC_LOCK(AP_RELEASE); 295 goto done; 296 } 297 /* 298 * need something to (un)trace (XXX - why is this here?) 299 */ 300 if (!facs) { 301 error = EINVAL; 302 goto done; 303 } 304 /* 305 * do it 306 */ 307 if (uap->pid < 0) { 308 /* 309 * by process group 310 */ 311 pg = pgfind(-uap->pid); 312 if (pg == NULL) { 313 error = ESRCH; 314 goto done; 315 } 316 LIST_FOREACH(p, &pg->pg_members, p_pglist) 317 if (descend) 318 ret |= ktrsetchildren(curp, p, ops, facs, vp); 319 else 320 ret |= ktrops(curp, p, ops, facs, vp); 321 } else { 322 /* 323 * by pid 324 */ 325 p = pfind(uap->pid); 326 if (p == NULL) { 327 error = ESRCH; 328 goto done; 329 } 330 if (descend) 331 ret |= ktrsetchildren(curp, p, ops, facs, vp); 332 else 333 ret |= ktrops(curp, p, ops, facs, vp); 334 } 335 if (!ret) 336 error = EPERM; 337done: 338 if (vp != NULL) 339 (void) vn_close(vp, FWRITE, curp->p_ucred, curp); 340 curp->p_traceflag &= ~KTRFAC_ACTIVE; 341 return (error); 342#else 343 return ENOSYS; 344#endif 345} 346 347/* 348 * utrace system call 349 */ 350/* ARGSUSED */ 351int 352utrace(curp, uap) 353 struct proc *curp; 354 register struct utrace_args *uap; 355{ 356#ifdef KTRACE 357 struct ktr_header *kth; 358 struct proc *p = curproc; /* XXX */ 359 register caddr_t cp; 360 361 if (!KTRPOINT(p, KTR_USER)) 362 return (0); 363 if (SCARG(uap, len) > KTR_USER_MAXLEN) 364 return (EINVAL); 365 p->p_traceflag |= KTRFAC_ACTIVE; 366 kth = ktrgetheader(KTR_USER); 367 MALLOC(cp, caddr_t, uap->len, M_KTRACE, M_WAITOK); 368 if (!copyin(uap->addr, cp, uap->len)) { 369 kth->ktr_buffer = cp; 370 kth->ktr_len = uap->len; 371 ktrwrite(p->p_tracep, kth, NULL); 372 } 373 FREE(kth, M_KTRACE); 374 FREE(cp, M_KTRACE); 375 p->p_traceflag &= ~KTRFAC_ACTIVE; 376 377 return (0); 378#else 379 return (ENOSYS); 380#endif 381} 382 383#ifdef KTRACE 384static int 385ktrops(curp, p, ops, facs, vp) 386 struct proc *p, *curp; 387 int ops, facs; 388 struct vnode *vp; 389{ 390 391 if (!ktrcanset(curp, p)) 392 return (0); 393 if (ops == KTROP_SET) { 394 if (p->p_tracep != vp) { 395 /* 396 * if trace file already in use, relinquish 397 */ 398 if (p->p_tracep != NULL) 399 vrele(p->p_tracep); 400 VREF(vp); 401 p->p_tracep = vp; 402 } 403 p->p_traceflag |= facs; 404 if (curp->p_ucred->cr_uid == 0) 405 p->p_traceflag |= KTRFAC_ROOT; 406 } else { 407 /* KTROP_CLEAR */ 408 if (((p->p_traceflag &= ~facs) & KTRFAC_MASK) == 0) { 409 /* no more tracing */ 410 p->p_traceflag = 0; 411 if (p->p_tracep != NULL) { 412 vrele(p->p_tracep); 413 p->p_tracep = NULL; 414 } 415 } 416 } 417 418 return (1); 419} 420 421static int 422ktrsetchildren(curp, top, ops, facs, vp) 423 struct proc *curp, *top; 424 int ops, facs; 425 struct vnode *vp; 426{ 427 register struct proc *p; 428 register int ret = 0; 429 430 p = top; 431 PROCTREE_LOCK(PT_SHARED); 432 for (;;) { 433 ret |= ktrops(curp, p, ops, facs, vp); 434 /* 435 * If this process has children, descend to them next, 436 * otherwise do any siblings, and if done with this level, 437 * follow back up the tree (but not past top). 438 */ 439 if (!LIST_EMPTY(&p->p_children)) 440 p = LIST_FIRST(&p->p_children); 441 else for (;;) { 442 if (p == top) { 443 PROCTREE_LOCK(PT_RELEASE); 444 return (ret); 445 } 446 if (LIST_NEXT(p, p_sibling)) { 447 p = LIST_NEXT(p, p_sibling); 448 break; 449 } 450 p = p->p_pptr; 451 } 452 } 453 /*NOTREACHED*/ 454} 455 456static void 457ktrwrite(vp, kth, uio) 458 struct vnode *vp; 459 register struct ktr_header *kth; 460 struct uio *uio; 461{ 462 struct uio auio; 463 struct iovec aiov[2]; 464 struct proc *p = curproc; /* XXX */ 465 struct mount *mp; 466 int error; 467 468 if (vp == NULL) 469 return; 470 auio.uio_iov = &aiov[0]; 471 auio.uio_offset = 0; 472 auio.uio_segflg = UIO_SYSSPACE; 473 auio.uio_rw = UIO_WRITE; 474 aiov[0].iov_base = (caddr_t)kth; 475 aiov[0].iov_len = sizeof(struct ktr_header); 476 auio.uio_resid = sizeof(struct ktr_header); 477 auio.uio_iovcnt = 1; 478 auio.uio_procp = curproc; 479 if (kth->ktr_len > 0) { 480 auio.uio_iovcnt++; 481 aiov[1].iov_base = kth->ktr_buffer; 482 aiov[1].iov_len = kth->ktr_len; 483 auio.uio_resid += kth->ktr_len; 484 if (uio != NULL) 485 kth->ktr_len += uio->uio_resid; 486 } 487 vn_start_write(vp, &mp, V_WAIT); 488 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY, p); 489 (void)VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE); 490 error = VOP_WRITE(vp, &auio, IO_UNIT | IO_APPEND, p->p_ucred); 491 if (error == 0 && uio != NULL) { 492 (void)VOP_LEASE(vp, p, p->p_ucred, LEASE_WRITE); 493 error = VOP_WRITE(vp, uio, IO_UNIT | IO_APPEND, p->p_ucred); 494 } 495 VOP_UNLOCK(vp, 0, p); 496 vn_finished_write(mp); 497 if (!error) 498 return; 499 /* 500 * If error encountered, give up tracing on this vnode. 501 */ 502 log(LOG_NOTICE, "ktrace write failed, errno %d, tracing stopped\n", 503 error); 504 ALLPROC_LOCK(AP_SHARED); 505 LIST_FOREACH(p, &allproc, p_list) { 506 if (p->p_tracep == vp) { 507 p->p_tracep = NULL; 508 p->p_traceflag = 0; 509 vrele(vp); 510 } 511 } 512 ALLPROC_LOCK(AP_RELEASE); 513} 514 515/* 516 * Return true if caller has permission to set the ktracing state 517 * of target. Essentially, the target can't possess any 518 * more permissions than the caller. KTRFAC_ROOT signifies that 519 * root previously set the tracing status on the target process, and 520 * so, only root may further change it. 521 * 522 * XXX: These checks are stronger than for ptrace() 523 * 524 * TODO: check groups. use caller effective gid. 525 */ 526static int 527ktrcanset(callp, targetp) 528 struct proc *callp, *targetp; 529{ 530 register struct pcred *caller = callp->p_cred; 531 register struct pcred *target = targetp->p_cred; 532 533 if (!PRISON_CHECK(callp, targetp)) 534 return (0); 535 if ((caller->pc_ucred->cr_uid == target->p_ruid && 536 target->p_ruid == target->p_svuid && 537 caller->p_rgid == target->p_rgid && /* XXX */ 538 target->p_rgid == target->p_svgid && 539 (targetp->p_traceflag & KTRFAC_ROOT) == 0) || 540 caller->pc_ucred->cr_uid == 0) 541 return (1); 542 543 return (0); 544} 545 546#endif /* KTRACE */ 547