sys/kern/kern_ktrace.c

   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 $
  35  * $DragonFly: src/sys/kern/kern_ktrace.c,v 1.23 2006/05/17 18:30:20 dillon Exp $
  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/nlookup.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>
  55 static MALLOC_DEFINE(M_KTRACE, "KTRACE", "KTRACE");
  56
  57 #ifdef KTRACE
  58 static struct ktr_header *ktrgetheader (int type);
  59 static void ktrwrite (struct proc *, struct ktr_header *, struct uio *);
  60 static int ktrcanset (struct proc *,struct proc *);
  61 static int ktrsetchildren (struct proc *,struct proc *,int,int,struct vnode *);
  62 static int ktrops (struct proc *,struct proc *,int,int,struct vnode *);
  63
  64
  65 static struct ktr_header *
  66 ktrgetheader(int type)
  67 {
  68         struct ktr_header *kth;
  69         struct proc *p = curproc;       /* XXX */
  70
  71         MALLOC(kth, struct ktr_header *, sizeof (struct ktr_header),
  72                 M_KTRACE, M_WAITOK);
  73         kth->ktr_type = type;
  74         microtime(&kth->ktr_time);
  75         kth->ktr_pid = p->p_pid;
  76         bcopy(p->p_comm, kth->ktr_comm, MAXCOMLEN + 1);
  77         return (kth);
  78 }
  79
  80 void
  81 ktrsyscall(struct proc *p, int code, int narg, register_t args[])
  82 {
  83         struct  ktr_header *kth;
  84         struct  ktr_syscall *ktp;
  85         int len;
  86         register_t *argp;
  87         int i;
  88
  89         len = offsetof(struct ktr_syscall, ktr_args) +
  90               (narg * sizeof(register_t));
  91         p->p_traceflag |= KTRFAC_ACTIVE;
  92         kth = ktrgetheader(KTR_SYSCALL);
  93         MALLOC(ktp, struct ktr_syscall *, len, M_KTRACE, M_WAITOK);
  94         ktp->ktr_code = code;
  95         ktp->ktr_narg = narg;
  96         argp = &ktp->ktr_args[0];
  97         for (i = 0; i < narg; i++)
  98                 *argp++ = args[i];
  99         kth->ktr_buf = (caddr_t)ktp;
 100         kth->ktr_len = len;
 101         ktrwrite(p, kth, NULL);
 102         FREE(ktp, M_KTRACE);
 103         FREE(kth, M_KTRACE);
 104         p->p_traceflag &= ~KTRFAC_ACTIVE;
 105 }
 106
 107 void
 108 ktrsysret(struct proc *p, int code, int error, register_t retval)
 109 {
 110         struct ktr_header *kth;
 111         struct ktr_sysret ktp;
 112
 113         p->p_traceflag |= KTRFAC_ACTIVE;
 114         kth = ktrgetheader(KTR_SYSRET);
 115         ktp.ktr_code = code;
 116         ktp.ktr_error = error;
 117         ktp.ktr_retval = retval;                /* what about val2 ? */
 118
 119         kth->ktr_buf = (caddr_t)&ktp;
 120         kth->ktr_len = sizeof(struct ktr_sysret);
 121
 122         ktrwrite(p, kth, NULL);
 123         FREE(kth, M_KTRACE);
 124         p->p_traceflag &= ~KTRFAC_ACTIVE;
 125 }
 126
 127 void
 128 ktrnamei(struct proc *p, char *path)
 129 {
 130         struct ktr_header *kth;
 131
 132         p->p_traceflag |= KTRFAC_ACTIVE;
 133         kth = ktrgetheader(KTR_NAMEI);
 134         kth->ktr_len = strlen(path);
 135         kth->ktr_buf = path;
 136
 137         ktrwrite(p, kth, NULL);
 138         FREE(kth, M_KTRACE);
 139         p->p_traceflag &= ~KTRFAC_ACTIVE;
 140 }
 141
 142 void
 143 ktrgenio(struct proc *p, int fd, enum uio_rw rw, struct uio *uio, int error)
 144 {
 145         struct ktr_header *kth;
 146         struct ktr_genio ktg;
 147
 148         if (error)
 149                 return;
 150         /*
 151          * don't let p_tracep get ripped out from under us
 152          */
 153         p->p_traceflag |= KTRFAC_ACTIVE;
 154         kth = ktrgetheader(KTR_GENIO);
 155         ktg.ktr_fd = fd;
 156         ktg.ktr_rw = rw;
 157         kth->ktr_buf = (caddr_t)&ktg;
 158         kth->ktr_len = sizeof(struct ktr_genio);
 159         uio->uio_offset = 0;
 160         uio->uio_rw = UIO_WRITE;
 161
 162         ktrwrite(p, kth, uio);
 163         FREE(kth, M_KTRACE);
 164         p->p_traceflag &= ~KTRFAC_ACTIVE;
 165 }
 166
 167 void
 168 ktrpsig(struct proc *p, int sig, sig_t action, sigset_t *mask, int code)
 169 {
 170         struct ktr_header *kth;
 171         struct ktr_psig kp;
 172
 173         p->p_traceflag |= KTRFAC_ACTIVE;
 174         kth = ktrgetheader(KTR_PSIG);
 175         kp.signo = (char)sig;
 176         kp.action = action;
 177         kp.mask = *mask;
 178         kp.code = code;
 179         kth->ktr_buf = (caddr_t)&kp;
 180         kth->ktr_len = sizeof (struct ktr_psig);
 181
 182         ktrwrite(p, kth, NULL);
 183         FREE(kth, M_KTRACE);
 184         p->p_traceflag &= ~KTRFAC_ACTIVE;
 185 }
 186
 187 void
 188 ktrcsw(struct proc *p, int out, int user)
 189 {
 190         struct ktr_header *kth;
 191         struct  ktr_csw kc;
 192
 193         p->p_traceflag |= KTRFAC_ACTIVE;
 194         kth = ktrgetheader(KTR_CSW);
 195         kc.out = out;
 196         kc.user = user;
 197         kth->ktr_buf = (caddr_t)&kc;
 198         kth->ktr_len = sizeof (struct ktr_csw);
 199
 200         ktrwrite(p, kth, NULL);
 201         FREE(kth, M_KTRACE);
 202         p->p_traceflag &= ~KTRFAC_ACTIVE;
 203 }
 204 #endif
 205
 206 /* Interface and common routines */
 207
 208 /*
 209  * ktrace system call
 210  */
 211 /* ARGSUSED */
 212 int
 213 ktrace(struct ktrace_args *uap)
 214 {
 215 #ifdef KTRACE
 216         struct thread *td = curthread;
 217         struct proc *curp = td->td_proc;
 218         struct vnode *vp = NULL;
 219         struct proc *p;
 220         struct pgrp *pg;
 221         int facs = uap->facs & ~KTRFAC_ROOT;
 222         int ops = KTROP(uap->ops);
 223         int descend = uap->ops & KTRFLAG_DESCEND;
 224         int ret = 0;
 225         int error = 0;
 226         struct nlookupdata nd;
 227
 228         curp->p_traceflag |= KTRFAC_ACTIVE;
 229         if (ops != KTROP_CLEAR) {
 230                 /*
 231                  * an operation which requires a file argument.
 232                  */
 233                 error = nlookup_init(&nd, uap->fname,
 234                                         UIO_USERSPACE, NLC_LOCKVP);
 235                 if (error == 0)
 236                         error = vn_open(&nd, NULL, FREAD|FWRITE|O_NOFOLLOW, 0);
 237                 if (error == 0 && nd.nl_open_vp->v_type != VREG)
 238                         error = EACCES;
 239                 if (error) {
 240                         curp->p_traceflag &= ~KTRFAC_ACTIVE;
 241                         nlookup_done(&nd);
 242                         return (error);
 243                 }
 244                 vp = nd.nl_open_vp;
 245                 nd.nl_open_vp = NULL;
 246                 nlookup_done(&nd);
 247                 VOP_UNLOCK(vp, 0);
 248         }
 249         /*
 250          * Clear all uses of the tracefile.  XXX umm, what happens to the
 251          * loop if vn_close() blocks?
 252          */
 253         if (ops == KTROP_CLEARFILE) {
 254                 FOREACH_PROC_IN_SYSTEM(p) {
 255                         if (p->p_tracep == vp) {
 256                                 if (ktrcanset(curp, p) && p->p_tracep == vp) {
 257                                         p->p_tracep = NULL;
 258                                         p->p_traceflag = 0;
 259                                         vn_close(vp, FREAD|FWRITE);
 260                                 } else {
 261                                         error = EPERM;
 262                                 }
 263                         }
 264                 }
 265                 goto done;
 266         }
 267         /*
 268          * need something to (un)trace (XXX - why is this here?)
 269          */
 270         if (!facs) {
 271                 error = EINVAL;
 272                 goto done;
 273         }
 274         /*
 275          * do it
 276          */
 277         if (uap->pid < 0) {
 278                 /*
 279                  * by process group
 280                  */
 281                 pg = pgfind(-uap->pid);
 282                 if (pg == NULL) {
 283                         error = ESRCH;
 284                         goto done;
 285                 }
 286                 LIST_FOREACH(p, &pg->pg_members, p_pglist) {
 287                         if (descend)
 288                                 ret |= ktrsetchildren(curp, p, ops, facs, vp);
 289                         else
 290                                 ret |= ktrops(curp, p, ops, facs, vp);
 291                 }
 292         } else {
 293                 /*
 294                  * by pid
 295                  */
 296                 p = pfind(uap->pid);
 297                 if (p == NULL) {
 298                         error = ESRCH;
 299                         goto done;
 300                 }
 301                 if (descend)
 302                         ret |= ktrsetchildren(curp, p, ops, facs, vp);
 303                 else
 304                         ret |= ktrops(curp, p, ops, facs, vp);
 305         }
 306         if (!ret)
 307                 error = EPERM;
 308 done:
 309         if (vp != NULL)
 310                 vn_close(vp, FWRITE);
 311         curp->p_traceflag &= ~KTRFAC_ACTIVE;
 312         return (error);
 313 #else
 314         return ENOSYS;
 315 #endif
 316 }
 317
 318 /*
 319  * utrace system call
 320  */
 321 /* ARGSUSED */
 322 int
 323 utrace(struct utrace_args *uap)
 324 {
 325 #ifdef KTRACE
 326         struct ktr_header *kth;
 327         struct thread *td = curthread;  /* XXX */
 328         struct proc *p = td->td_proc;
 329         caddr_t cp;
 330
 331         if (!KTRPOINT(td, KTR_USER))
 332                 return (0);
 333         if (uap->len > KTR_USER_MAXLEN)
 334                 return (EINVAL);
 335         p->p_traceflag |= KTRFAC_ACTIVE;
 336         /*
 337          * don't let p_tracep get ripped out from under us while we are
 338          * writing.
 339          */
 340         kth = ktrgetheader(KTR_USER);
 341         MALLOC(cp, caddr_t, uap->len, M_KTRACE, M_WAITOK);
 342         if (!copyin(uap->addr, cp, uap->len)) {
 343                 kth->ktr_buf = cp;
 344                 kth->ktr_len = uap->len;
 345                 ktrwrite(p, kth, NULL);
 346         }
 347         FREE(kth, M_KTRACE);
 348         FREE(cp, M_KTRACE);
 349         p->p_traceflag &= ~KTRFAC_ACTIVE;
 350
 351         return (0);
 352 #else
 353         return (ENOSYS);
 354 #endif
 355 }
 356
 357 #ifdef KTRACE
 358 static int
 359 ktrops(struct proc *curp, struct proc *p, int ops, int facs, struct vnode *vp)
 360 {
 361
 362         if (!ktrcanset(curp, p))
 363                 return (0);
 364         if (ops == KTROP_SET) {
 365                 if (p->p_tracep != vp) {
 366                         struct vnode *vtmp;
 367
 368                         /*
 369                          * if trace file already in use, relinquish
 370                          */
 371                         vref(vp);
 372                         while ((vtmp = p->p_tracep) != NULL) {
 373                                 p->p_tracep = NULL;
 374                                 vrele(vtmp);
 375                         }
 376                         p->p_tracep = vp;
 377                 }
 378                 p->p_traceflag |= facs;
 379                 if (curp->p_ucred->cr_uid == 0)
 380                         p->p_traceflag |= KTRFAC_ROOT;
 381         } else {
 382                 /* KTROP_CLEAR */
 383                 if (((p->p_traceflag &= ~facs) & KTRFAC_MASK) == 0) {
 384                         struct vnode *vtmp;
 385
 386                         /* no more tracing */
 387                         p->p_traceflag = 0;
 388                         if ((vtmp = p->p_tracep) != NULL) {
 389                                 p->p_tracep = NULL;
 390                                 vrele(vtmp);
 391                         }
 392                 }
 393         }
 394
 395         return (1);
 396 }
 397
 398 static int
 399 ktrsetchildren(struct proc *curp, struct proc *top, int ops, int facs,
 400                struct vnode *vp)
 401 {
 402         struct proc *p;
 403         int ret = 0;
 404
 405         p = top;
 406         for (;;) {
 407                 ret |= ktrops(curp, p, ops, facs, vp);
 408                 /*
 409                  * If this process has children, descend to them next,
 410                  * otherwise do any siblings, and if done with this level,
 411                  * follow back up the tree (but not past top).
 412                  */
 413                 if (!LIST_EMPTY(&p->p_children))
 414                         p = LIST_FIRST(&p->p_children);
 415                 else for (;;) {
 416                         if (p == top)
 417                                 return (ret);
 418                         if (LIST_NEXT(p, p_sibling)) {
 419                                 p = LIST_NEXT(p, p_sibling);
 420                                 break;
 421                         }
 422                         p = p->p_pptr;
 423                 }
 424         }
 425         /*NOTREACHED*/
 426 }
 427
 428 static void
 429 ktrwrite(struct proc *p, struct ktr_header *kth, struct uio *uio)
 430 {
 431         struct uio auio;
 432         struct iovec aiov[2];
 433         struct vnode *vp;
 434         int error;
 435
 436         /*
 437          * Since multiple processes may be trying to ktrace, we must vref
 438          * the vnode to prevent it from being ripped out from under us in
 439          * case another process gets a write error while ktracing and removes
 440          * the reference from the proc.
 441          *
 442          * XXX not MP safe
 443          */
 444         if ((vp = p->p_tracep) == NULL)
 445                 return;
 446         vref(vp);
 447         auio.uio_iov = &aiov[0];
 448         auio.uio_offset = 0;
 449         auio.uio_segflg = UIO_SYSSPACE;
 450         auio.uio_rw = UIO_WRITE;
 451         aiov[0].iov_base = (caddr_t)kth;
 452         aiov[0].iov_len = sizeof(struct ktr_header);
 453         auio.uio_resid = sizeof(struct ktr_header);
 454         auio.uio_iovcnt = 1;
 455         auio.uio_td = curthread;
 456         if (kth->ktr_len > 0) {
 457                 auio.uio_iovcnt++;
 458                 aiov[1].iov_base = kth->ktr_buf;
 459                 aiov[1].iov_len = kth->ktr_len;
 460                 auio.uio_resid += kth->ktr_len;
 461                 if (uio != NULL)
 462                         kth->ktr_len += uio->uio_resid;
 463         }
 464         vn_lock(vp, LK_EXCLUSIVE | LK_RETRY);
 465         error = VOP_WRITE(vp, &auio, IO_UNIT | IO_APPEND, p->p_ucred);
 466         if (error == 0 && uio != NULL) {
 467                 error = VOP_WRITE(vp, uio, IO_UNIT | IO_APPEND, p->p_ucred);
 468         }
 469         VOP_UNLOCK(vp, 0);
 470         vrele(vp);
 471         if (!error)
 472                 return;
 473         /*
 474          * If error encountered, give up tracing on this vnode.  XXX what
 475          * happens to the loop if vrele() blocks?
 476          */
 477         log(LOG_NOTICE, "ktrace write failed, errno %d, tracing stopped\n",
 478             error);
 479         FOREACH_PROC_IN_SYSTEM(p) {
 480                 if (p->p_tracep == vp) {
 481                         p->p_tracep = NULL;
 482                         p->p_traceflag = 0;
 483                         vrele(vp);
 484                 }
 485         }
 486 }
 487
 488 /*
 489  * Return true if caller has permission to set the ktracing state
 490  * of target.  Essentially, the target can't possess any
 491  * more permissions than the caller.  KTRFAC_ROOT signifies that
 492  * root previously set the tracing status on the target process, and
 493  * so, only root may further change it.
 494  *
 495  * TODO: check groups.  use caller effective gid.
 496  */
 497 static int
 498 ktrcanset(struct proc *callp, struct proc *targetp)
 499 {
 500         struct ucred *caller = callp->p_ucred;
 501         struct ucred *target = targetp->p_ucred;
 502
 503         if (!PRISON_CHECK(caller, target))
 504                 return (0);
 505         if ((caller->cr_uid == target->cr_ruid &&
 506              target->cr_ruid == target->cr_svuid &&
 507              caller->cr_rgid == target->cr_rgid &&      /* XXX */
 508              target->cr_rgid == target->cr_svgid &&
 509              (targetp->p_traceflag & KTRFAC_ROOT) == 0 &&
 510              (targetp->p_flag & P_SUGID) == 0) ||
 511              caller->cr_uid == 0)
 512                 return (1);
 513
 514         return (0);
 515 }
 516
 517 #endif /* KTRACE */