procfs_subr.c revision 1.73
1 /* $NetBSD: procfs_subr.c,v 1.73 2006/11/28 17:27:09 elad Exp $ */ 2 3 /* 4 * Copyright (c) 1993 5 * The Regents of the University of California. All rights reserved. 6 * 7 * This code is derived from software contributed to Berkeley by 8 * Jan-Simon Pendry. 9 * 10 * Redistribution and use in source and binary forms, with or without 11 * modification, are permitted provided that the following conditions 12 * are met: 13 * 1. Redistributions of source code must retain the above copyright 14 * notice, this list of conditions and the following disclaimer. 15 * 2. Redistributions in binary form must reproduce the above copyright 16 * notice, this list of conditions and the following disclaimer in the 17 * documentation and/or other materials provided with the distribution. 18 * 3. Neither the name of the University nor the names of its contributors 19 * may be used to endorse or promote products derived from this software 20 * without specific prior written permission. 21 * 22 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 23 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 24 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 25 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 26 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 27 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 28 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 29 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 30 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 31 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 32 * SUCH DAMAGE. 33 * 34 * @(#)procfs_subr.c 8.6 (Berkeley) 5/14/95 35 */ 36 37 /* 38 * Copyright (c) 1994 Christopher G. Demetriou. All rights reserved. 39 * Copyright (c) 1993 Jan-Simon Pendry 40 * 41 * This code is derived from software contributed to Berkeley by 42 * Jan-Simon Pendry. 43 * 44 * Redistribution and use in source and binary forms, with or without 45 * modification, are permitted provided that the following conditions 46 * are met: 47 * 1. Redistributions of source code must retain the above copyright 48 * notice, this list of conditions and the following disclaimer. 49 * 2. Redistributions in binary form must reproduce the above copyright 50 * notice, this list of conditions and the following disclaimer in the 51 * documentation and/or other materials provided with the distribution. 52 * 3. All advertising materials mentioning features or use of this software 53 * must display the following acknowledgement: 54 * This product includes software developed by the University of 55 * California, Berkeley and its contributors. 56 * 4. Neither the name of the University nor the names of its contributors 57 * may be used to endorse or promote products derived from this software 58 * without specific prior written permission. 59 * 60 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 61 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 62 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 63 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 64 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 65 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 66 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 67 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 68 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 69 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 70 * SUCH DAMAGE. 71 * 72 * @(#)procfs_subr.c 8.6 (Berkeley) 5/14/95 73 */ 74 75 #include <sys/cdefs.h> 76 __KERNEL_RCSID(0, "$NetBSD: procfs_subr.c,v 1.73 2006/11/28 17:27:09 elad Exp $"); 77 78 #include <sys/param.h> 79 #include <sys/systm.h> 80 #include <sys/time.h> 81 #include <sys/kernel.h> 82 #include <sys/proc.h> 83 #include <sys/vnode.h> 84 #include <sys/malloc.h> 85 #include <sys/stat.h> 86 #include <sys/file.h> 87 #include <sys/filedesc.h> 88 #include <sys/kauth.h> 89 90 #include <miscfs/procfs/procfs.h> 91 92 void procfs_hashins(struct pfsnode *); 93 void procfs_hashrem(struct pfsnode *); 94 struct vnode *procfs_hashget(pid_t, pfstype, int, struct mount *); 95 96 LIST_HEAD(pfs_hashhead, pfsnode) *pfs_hashtbl; 97 u_long pfs_ihash; /* size of hash table - 1 */ 98 #define PFSPIDHASH(pid) ((pid) & pfs_ihash) 99 100 struct lock pfs_hashlock; 101 struct simplelock pfs_hash_slock; 102 103 #define ISSET(t, f) ((t) & (f)) 104 105 /* 106 * allocate a pfsnode/vnode pair. the vnode is 107 * referenced, and locked. 108 * 109 * the pid, pfs_type, and mount point uniquely 110 * identify a pfsnode. the mount point is needed 111 * because someone might mount this filesystem 112 * twice. 113 * 114 * all pfsnodes are maintained on a singly-linked 115 * list. new nodes are only allocated when they cannot 116 * be found on this list. entries on the list are 117 * removed when the vfs reclaim entry is called. 118 * 119 * a single lock is kept for the entire list. this is 120 * needed because the getnewvnode() function can block 121 * waiting for a vnode to become free, in which case there 122 * may be more than one process trying to get the same 123 * vnode. this lock is only taken if we are going to 124 * call getnewvnode, since the kernel itself is single-threaded. 125 * 126 * if an entry is found on the list, then call vget() to 127 * take a reference. this is done because there may be 128 * zero references to it and so it needs to removed from 129 * the vnode free list. 130 */ 131 int 132 procfs_allocvp(mp, vpp, pid, pfs_type, fd) 133 struct mount *mp; 134 struct vnode **vpp; 135 pid_t pid; 136 pfstype pfs_type; 137 int fd; 138 { 139 struct pfsnode *pfs; 140 struct vnode *vp; 141 int error; 142 143 do { 144 if ((*vpp = procfs_hashget(pid, pfs_type, fd, mp)) != NULL) 145 return (0); 146 } while (lockmgr(&pfs_hashlock, LK_EXCLUSIVE|LK_SLEEPFAIL, 0)); 147 148 if ((error = getnewvnode(VT_PROCFS, mp, procfs_vnodeop_p, &vp)) != 0) { 149 *vpp = NULL; 150 lockmgr(&pfs_hashlock, LK_RELEASE, NULL); 151 return (error); 152 } 153 154 MALLOC(pfs, void *, sizeof(struct pfsnode), M_TEMP, M_WAITOK); 155 vp->v_data = pfs; 156 157 pfs->pfs_pid = pid; 158 pfs->pfs_type = pfs_type; 159 pfs->pfs_vnode = vp; 160 pfs->pfs_flags = 0; 161 pfs->pfs_fileno = PROCFS_FILENO(pid, pfs_type, fd); 162 pfs->pfs_fd = fd; 163 164 switch (pfs_type) { 165 case PFSroot: /* /proc = dr-xr-xr-x */ 166 pfs->pfs_mode = S_IRUSR|S_IXUSR|S_IRGRP|S_IXGRP|S_IROTH|S_IXOTH; 167 vp->v_type = VDIR; 168 vp->v_flag = VROOT; 169 break; 170 171 case PFScurproc: /* /proc/curproc = lr-xr-xr-x */ 172 case PFSself: /* /proc/self = lr-xr-xr-x */ 173 case PFScwd: /* /proc/N/cwd = lr-xr-xr-x */ 174 case PFSchroot: /* /proc/N/chroot = lr-xr-xr-x */ 175 case PFSexe: /* /proc/N/exe = lr-xr-xr-x */ 176 pfs->pfs_mode = S_IRUSR|S_IXUSR|S_IRGRP|S_IXGRP|S_IROTH|S_IXOTH; 177 vp->v_type = VLNK; 178 break; 179 180 case PFSproc: /* /proc/N = dr-xr-xr-x */ 181 case PFSfd: 182 if (fd == -1) { /* /proc/N/fd = dr-xr-xr-x */ 183 pfs->pfs_mode = S_IRUSR|S_IXUSR; 184 vp->v_type = VDIR; 185 } else { /* /proc/N/fd/M = [ps-]rw------- */ 186 struct file *fp; 187 struct vnode *vxp; 188 struct proc *pown; 189 190 /* XXX can procfs_getfp() ever fail here? */ 191 if ((error = procfs_getfp(pfs, &pown, &fp)) != 0) 192 goto bad; 193 FILE_USE(fp); 194 195 pfs->pfs_mode = S_IRUSR|S_IWUSR; 196 switch (fp->f_type) { 197 case DTYPE_VNODE: 198 vxp = (struct vnode *)fp->f_data; 199 200 /* 201 * We make symlinks for directories 202 * to avoid cycles. 203 */ 204 if (vxp->v_type == VDIR) 205 goto symlink; 206 vp->v_type = vxp->v_type; 207 break; 208 case DTYPE_PIPE: 209 vp->v_type = VFIFO; 210 break; 211 case DTYPE_SOCKET: 212 vp->v_type = VSOCK; 213 break; 214 case DTYPE_KQUEUE: 215 case DTYPE_MISC: 216 symlink: 217 pfs->pfs_mode = S_IRUSR|S_IXUSR|S_IRGRP| 218 S_IXGRP|S_IROTH|S_IXOTH; 219 vp->v_type = VLNK; 220 break; 221 default: 222 error = EOPNOTSUPP; 223 FILE_UNUSE(fp, proc_representative_lwp(pown)); 224 goto bad; 225 } 226 FILE_UNUSE(fp, proc_representative_lwp(pown)); 227 } 228 break; 229 230 case PFSfile: /* /proc/N/file = -rw------- */ 231 case PFSmem: /* /proc/N/mem = -rw------- */ 232 case PFSregs: /* /proc/N/regs = -rw------- */ 233 case PFSfpregs: /* /proc/N/fpregs = -rw------- */ 234 pfs->pfs_mode = S_IRUSR|S_IWUSR; 235 vp->v_type = VREG; 236 break; 237 238 case PFSctl: /* /proc/N/ctl = --w------ */ 239 case PFSnote: /* /proc/N/note = --w------ */ 240 case PFSnotepg: /* /proc/N/notepg = --w------ */ 241 pfs->pfs_mode = S_IWUSR; 242 vp->v_type = VREG; 243 break; 244 245 case PFSmap: /* /proc/N/map = -r--r--r-- */ 246 case PFSmaps: /* /proc/N/maps = -r--r--r-- */ 247 case PFSstatus: /* /proc/N/status = -r--r--r-- */ 248 case PFSstat: /* /proc/N/stat = -r--r--r-- */ 249 case PFScmdline: /* /proc/N/cmdline = -r--r--r-- */ 250 case PFSemul: /* /proc/N/emul = -r--r--r-- */ 251 case PFSmeminfo: /* /proc/meminfo = -r--r--r-- */ 252 case PFSdevices: /* /proc/devices = -r--r--r-- */ 253 case PFScpuinfo: /* /proc/cpuinfo = -r--r--r-- */ 254 case PFSuptime: /* /proc/uptime = -r--r--r-- */ 255 case PFSmounts: /* /proc/mounts = -r--r--r-- */ 256 pfs->pfs_mode = S_IRUSR|S_IRGRP|S_IROTH; 257 vp->v_type = VREG; 258 break; 259 260 #ifdef __HAVE_PROCFS_MACHDEP 261 PROCFS_MACHDEP_NODETYPE_CASES 262 procfs_machdep_allocvp(vp); 263 break; 264 #endif 265 266 default: 267 panic("procfs_allocvp"); 268 } 269 270 procfs_hashins(pfs); 271 uvm_vnp_setsize(vp, 0); 272 lockmgr(&pfs_hashlock, LK_RELEASE, NULL); 273 274 *vpp = vp; 275 return (0); 276 277 bad: 278 lockmgr(&pfs_hashlock, LK_RELEASE, NULL); 279 FREE(pfs, M_TEMP); 280 ungetnewvnode(vp); 281 return (error); 282 } 283 284 int 285 procfs_freevp(vp) 286 struct vnode *vp; 287 { 288 struct pfsnode *pfs = VTOPFS(vp); 289 290 procfs_hashrem(pfs); 291 292 FREE(vp->v_data, M_TEMP); 293 vp->v_data = 0; 294 return (0); 295 } 296 297 int 298 procfs_rw(v) 299 void *v; 300 { 301 struct vop_read_args *ap = v; 302 struct vnode *vp = ap->a_vp; 303 struct uio *uio = ap->a_uio; 304 struct lwp *curl; 305 struct lwp *l; 306 struct pfsnode *pfs = VTOPFS(vp); 307 struct proc *p; 308 int error; 309 310 if (uio->uio_offset < 0) 311 return EINVAL; 312 p = PFIND(pfs->pfs_pid); 313 if (p == 0) 314 return ESRCH; 315 316 if (ISSET(p->p_flag, P_INEXEC)) 317 return (EAGAIN); 318 319 curl = curlwp; 320 321 /* 322 * Do not allow init to be modified while in secure mode; it 323 * could be duped into changing the security level. 324 */ 325 #define M2K(m) ((m) == UIO_READ ? KAUTH_REQ_PROCESS_CANPROCFS_READ : \ 326 KAUTH_REQ_PROCESS_CANPROCFS_WRITE) 327 error = kauth_authorize_process(curl->l_cred, KAUTH_PROCESS_CANPROCFS, 328 p, pfs, KAUTH_ARG(M2K(uio->uio_rw)), NULL); 329 if (error) 330 return (error); 331 #undef M2K 332 333 /* XXX NJWLWP 334 * The entire procfs interface needs work to be useful to 335 * a process with multiple LWPs. For the moment, we'll 336 * just kluge this and fail on others. 337 */ 338 l = proc_representative_lwp(p); 339 340 switch (pfs->pfs_type) { 341 case PFSnote: 342 case PFSnotepg: 343 return (procfs_donote(curl, p, pfs, uio)); 344 345 case PFSregs: 346 return (procfs_doregs(curl, l, pfs, uio)); 347 348 case PFSfpregs: 349 return (procfs_dofpregs(curl, l, pfs, uio)); 350 351 case PFSctl: 352 return (procfs_doctl(curl, l, pfs, uio)); 353 354 case PFSstatus: 355 return (procfs_dostatus(curl, l, pfs, uio)); 356 357 case PFSstat: 358 return (procfs_do_pid_stat(curl, l, pfs, uio)); 359 360 case PFSmap: 361 return (procfs_domap(curl, p, pfs, uio, 0)); 362 363 case PFSmaps: 364 return (procfs_domap(curl, p, pfs, uio, 1)); 365 366 case PFSmem: 367 return (procfs_domem(curl, l, pfs, uio)); 368 369 case PFScmdline: 370 return (procfs_docmdline(curl, p, pfs, uio)); 371 372 case PFSmeminfo: 373 return (procfs_domeminfo(curl, p, pfs, uio)); 374 375 case PFSdevices: 376 return (procfs_dodevices(curl, p, pfs, uio)); 377 378 case PFScpuinfo: 379 return (procfs_docpuinfo(curl, p, pfs, uio)); 380 381 case PFSfd: 382 return (procfs_dofd(curl, p, pfs, uio)); 383 384 case PFSuptime: 385 return (procfs_douptime(curl, p, pfs, uio)); 386 387 case PFSmounts: 388 return (procfs_domounts(curl, p, pfs, uio)); 389 390 case PFSemul: 391 return procfs_doemul(curl, p, pfs, uio); 392 393 #ifdef __HAVE_PROCFS_MACHDEP 394 PROCFS_MACHDEP_NODETYPE_CASES 395 return (procfs_machdep_rw(curl, l, pfs, uio)); 396 #endif 397 398 default: 399 return (EOPNOTSUPP); 400 } 401 } 402 403 /* 404 * Get a string from userland into (bf). Strip a trailing 405 * nl character (to allow easy access from the shell). 406 * The buffer should be *buflenp + 1 chars long. vfs_getuserstr 407 * will automatically add a nul char at the end. 408 * 409 * Returns 0 on success or the following errors 410 * 411 * EINVAL: file offset is non-zero. 412 * EMSGSIZE: message is longer than kernel buffer 413 * EFAULT: user i/o buffer is not addressable 414 */ 415 int 416 vfs_getuserstr(uio, bf, buflenp) 417 struct uio *uio; 418 char *bf; 419 int *buflenp; 420 { 421 int xlen; 422 int error; 423 424 if (uio->uio_offset != 0) 425 return (EINVAL); 426 427 xlen = *buflenp; 428 429 /* must be able to read the whole string in one go */ 430 if (xlen < uio->uio_resid) 431 return (EMSGSIZE); 432 xlen = uio->uio_resid; 433 434 if ((error = uiomove(bf, xlen, uio)) != 0) 435 return (error); 436 437 /* allow multiple writes without seeks */ 438 uio->uio_offset = 0; 439 440 /* cleanup string and remove trailing newline */ 441 bf[xlen] = '\0'; 442 xlen = strlen(bf); 443 if (xlen > 0 && bf[xlen-1] == '\n') 444 bf[--xlen] = '\0'; 445 *buflenp = xlen; 446 447 return (0); 448 } 449 450 const vfs_namemap_t * 451 vfs_findname(nm, bf, buflen) 452 const vfs_namemap_t *nm; 453 const char *bf; 454 int buflen; 455 { 456 457 for (; nm->nm_name; nm++) 458 if (memcmp(bf, nm->nm_name, buflen+1) == 0) 459 return (nm); 460 461 return (0); 462 } 463 464 /* 465 * Initialize pfsnode hash table. 466 */ 467 void 468 procfs_hashinit() 469 { 470 lockinit(&pfs_hashlock, PINOD, "pfs_hashlock", 0, 0); 471 pfs_hashtbl = hashinit(desiredvnodes / 4, HASH_LIST, M_UFSMNT, 472 M_WAITOK, &pfs_ihash); 473 simple_lock_init(&pfs_hash_slock); 474 } 475 476 void 477 procfs_hashreinit() 478 { 479 struct pfsnode *pp; 480 struct pfs_hashhead *oldhash, *hash; 481 u_long i, oldmask, mask, val; 482 483 hash = hashinit(desiredvnodes / 4, HASH_LIST, M_UFSMNT, M_WAITOK, 484 &mask); 485 486 simple_lock(&pfs_hash_slock); 487 oldhash = pfs_hashtbl; 488 oldmask = pfs_ihash; 489 pfs_hashtbl = hash; 490 pfs_ihash = mask; 491 for (i = 0; i <= oldmask; i++) { 492 while ((pp = LIST_FIRST(&oldhash[i])) != NULL) { 493 LIST_REMOVE(pp, pfs_hash); 494 val = PFSPIDHASH(pp->pfs_pid); 495 LIST_INSERT_HEAD(&hash[val], pp, pfs_hash); 496 } 497 } 498 simple_unlock(&pfs_hash_slock); 499 hashdone(oldhash, M_UFSMNT); 500 } 501 502 /* 503 * Free pfsnode hash table. 504 */ 505 void 506 procfs_hashdone() 507 { 508 hashdone(pfs_hashtbl, M_UFSMNT); 509 } 510 511 struct vnode * 512 procfs_hashget(pid, type, fd, mp) 513 pid_t pid; 514 pfstype type; 515 int fd; 516 struct mount *mp; 517 { 518 struct pfs_hashhead *ppp; 519 struct pfsnode *pp; 520 struct vnode *vp; 521 522 loop: 523 simple_lock(&pfs_hash_slock); 524 ppp = &pfs_hashtbl[PFSPIDHASH(pid)]; 525 LIST_FOREACH(pp, ppp, pfs_hash) { 526 vp = PFSTOV(pp); 527 if (pid == pp->pfs_pid && pp->pfs_type == type && 528 pp->pfs_fd == fd && vp->v_mount == mp) { 529 simple_lock(&vp->v_interlock); 530 simple_unlock(&pfs_hash_slock); 531 if (vget(vp, LK_EXCLUSIVE | LK_INTERLOCK)) 532 goto loop; 533 return (vp); 534 } 535 } 536 simple_unlock(&pfs_hash_slock); 537 return (NULL); 538 } 539 540 /* 541 * Insert the pfsnode into the hash table and lock it. 542 */ 543 void 544 procfs_hashins(pp) 545 struct pfsnode *pp; 546 { 547 struct pfs_hashhead *ppp; 548 549 /* lock the pfsnode, then put it on the appropriate hash list */ 550 lockmgr(&pp->pfs_vnode->v_lock, LK_EXCLUSIVE, (struct simplelock *)0); 551 552 simple_lock(&pfs_hash_slock); 553 ppp = &pfs_hashtbl[PFSPIDHASH(pp->pfs_pid)]; 554 LIST_INSERT_HEAD(ppp, pp, pfs_hash); 555 simple_unlock(&pfs_hash_slock); 556 } 557 558 /* 559 * Remove the pfsnode from the hash table. 560 */ 561 void 562 procfs_hashrem(pp) 563 struct pfsnode *pp; 564 { 565 simple_lock(&pfs_hash_slock); 566 LIST_REMOVE(pp, pfs_hash); 567 simple_unlock(&pfs_hash_slock); 568 } 569 570 void 571 procfs_revoke_vnodes(p, arg) 572 struct proc *p; 573 void *arg; 574 { 575 struct pfsnode *pfs, *pnext; 576 struct vnode *vp; 577 struct mount *mp = (struct mount *)arg; 578 struct pfs_hashhead *ppp; 579 580 if (!(p->p_flag & P_SUGID)) 581 return; 582 583 ppp = &pfs_hashtbl[PFSPIDHASH(p->p_pid)]; 584 for (pfs = LIST_FIRST(ppp); pfs; pfs = pnext) { 585 vp = PFSTOV(pfs); 586 pnext = LIST_NEXT(pfs, pfs_hash); 587 if (vp->v_usecount > 0 && pfs->pfs_pid == p->p_pid && 588 vp->v_mount == mp) 589 VOP_REVOKE(vp, REVOKEALL); 590 } 591 } 592 593 int 594 procfs_getfp(pfs, pown, fp) 595 struct pfsnode *pfs; 596 struct proc **pown; 597 struct file **fp; 598 { 599 struct proc *p = PFIND(pfs->pfs_pid); 600 601 if (p == NULL) 602 return ESRCH; 603 604 if (pfs->pfs_fd == -1) 605 return EINVAL; 606 607 if ((*fp = fd_getfile(p->p_fd, pfs->pfs_fd)) == NULL) 608 return EBADF; 609 610 *pown = p; 611 return 0; 612 } 613 614 int 615 procfs_doemul(struct lwp *curl, struct proc *p, 616 struct pfsnode *pfs, struct uio *uio) 617 { 618 const char *ename = p->p_emul->e_name; 619 return uiomove_frombuf(__UNCONST(ename), strlen(ename), uio); 620 } 621
Indexes created Tue Sep 23 13:09:54 GMT 2025