procfs_subr.c revision 1.74
1 /* $NetBSD: procfs_subr.c,v 1.74 2006/12/24 17:37:35 christos 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 1.74 2006/12/24 17:37:35 christos 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 vp->v_flag = VROOT; 167 /*FALLTHROUGH*/ 168 case PFSproc: /* /proc/N = dr-xr-xr-x */ 169 pfs->pfs_mode = S_IRUSR|S_IXUSR|S_IRGRP|S_IXGRP|S_IROTH|S_IXOTH; 170 vp->v_type = VDIR; 171 break; 172 173 case PFScurproc: /* /proc/curproc = lr-xr-xr-x */ 174 case PFSself: /* /proc/self = lr-xr-xr-x */ 175 case PFScwd: /* /proc/N/cwd = lr-xr-xr-x */ 176 case PFSchroot: /* /proc/N/chroot = lr-xr-xr-x */ 177 case PFSexe: /* /proc/N/exe = lr-xr-xr-x */ 178 pfs->pfs_mode = S_IRUSR|S_IXUSR|S_IRGRP|S_IXGRP|S_IROTH|S_IXOTH; 179 vp->v_type = VLNK; 180 break; 181 182 case PFSfd: 183 if (fd == -1) { /* /proc/N/fd = dr-xr-xr-x */ 184 pfs->pfs_mode = S_IRUSR|S_IXUSR; 185 vp->v_type = VDIR; 186 } else { /* /proc/N/fd/M = [ps-]rw------- */ 187 struct file *fp; 188 struct vnode *vxp; 189 struct proc *pown; 190 191 /* XXX can procfs_getfp() ever fail here? */ 192 if ((error = procfs_getfp(pfs, &pown, &fp)) != 0) 193 goto bad; 194 FILE_USE(fp); 195 196 pfs->pfs_mode = S_IRUSR|S_IWUSR; 197 switch (fp->f_type) { 198 case DTYPE_VNODE: 199 vxp = (struct vnode *)fp->f_data; 200 201 /* 202 * We make symlinks for directories 203 * to avoid cycles. 204 */ 205 if (vxp->v_type == VDIR) 206 goto symlink; 207 vp->v_type = vxp->v_type; 208 break; 209 case DTYPE_PIPE: 210 vp->v_type = VFIFO; 211 break; 212 case DTYPE_SOCKET: 213 vp->v_type = VSOCK; 214 break; 215 case DTYPE_KQUEUE: 216 case DTYPE_MISC: 217 symlink: 218 pfs->pfs_mode = S_IRUSR|S_IXUSR|S_IRGRP| 219 S_IXGRP|S_IROTH|S_IXOTH; 220 vp->v_type = VLNK; 221 break; 222 default: 223 error = EOPNOTSUPP; 224 FILE_UNUSE(fp, proc_representative_lwp(pown)); 225 goto bad; 226 } 227 FILE_UNUSE(fp, proc_representative_lwp(pown)); 228 } 229 break; 230 231 case PFSfile: /* /proc/N/file = -rw------- */ 232 case PFSmem: /* /proc/N/mem = -rw------- */ 233 case PFSregs: /* /proc/N/regs = -rw------- */ 234 case PFSfpregs: /* /proc/N/fpregs = -rw------- */ 235 pfs->pfs_mode = S_IRUSR|S_IWUSR; 236 vp->v_type = VREG; 237 break; 238 239 case PFSctl: /* /proc/N/ctl = --w------ */ 240 case PFSnote: /* /proc/N/note = --w------ */ 241 case PFSnotepg: /* /proc/N/notepg = --w------ */ 242 pfs->pfs_mode = S_IWUSR; 243 vp->v_type = VREG; 244 break; 245 246 case PFSmap: /* /proc/N/map = -r--r--r-- */ 247 case PFSmaps: /* /proc/N/maps = -r--r--r-- */ 248 case PFSstatus: /* /proc/N/status = -r--r--r-- */ 249 case PFSstat: /* /proc/N/stat = -r--r--r-- */ 250 case PFScmdline: /* /proc/N/cmdline = -r--r--r-- */ 251 case PFSemul: /* /proc/N/emul = -r--r--r-- */ 252 case PFSmeminfo: /* /proc/meminfo = -r--r--r-- */ 253 case PFSdevices: /* /proc/devices = -r--r--r-- */ 254 case PFScpuinfo: /* /proc/cpuinfo = -r--r--r-- */ 255 case PFSuptime: /* /proc/uptime = -r--r--r-- */ 256 case PFSmounts: /* /proc/mounts = -r--r--r-- */ 257 pfs->pfs_mode = S_IRUSR|S_IRGRP|S_IROTH; 258 vp->v_type = VREG; 259 break; 260 261 #ifdef __HAVE_PROCFS_MACHDEP 262 PROCFS_MACHDEP_NODETYPE_CASES 263 procfs_machdep_allocvp(vp); 264 break; 265 #endif 266 267 default: 268 panic("procfs_allocvp"); 269 } 270 271 procfs_hashins(pfs); 272 uvm_vnp_setsize(vp, 0); 273 lockmgr(&pfs_hashlock, LK_RELEASE, NULL); 274 275 *vpp = vp; 276 return (0); 277 278 bad: 279 lockmgr(&pfs_hashlock, LK_RELEASE, NULL); 280 FREE(pfs, M_TEMP); 281 ungetnewvnode(vp); 282 return (error); 283 } 284 285 int 286 procfs_freevp(vp) 287 struct vnode *vp; 288 { 289 struct pfsnode *pfs = VTOPFS(vp); 290 291 procfs_hashrem(pfs); 292 293 FREE(vp->v_data, M_TEMP); 294 vp->v_data = 0; 295 return (0); 296 } 297 298 int 299 procfs_rw(v) 300 void *v; 301 { 302 struct vop_read_args *ap = v; 303 struct vnode *vp = ap->a_vp; 304 struct uio *uio = ap->a_uio; 305 struct lwp *curl; 306 struct lwp *l; 307 struct pfsnode *pfs = VTOPFS(vp); 308 struct proc *p; 309 int error; 310 311 if (uio->uio_offset < 0) 312 return EINVAL; 313 p = PFIND(pfs->pfs_pid); 314 if (p == 0) 315 return ESRCH; 316 317 if (ISSET(p->p_flag, P_INEXEC)) 318 return (EAGAIN); 319 320 curl = curlwp; 321 322 /* 323 * Do not allow init to be modified while in secure mode; it 324 * could be duped into changing the security level. 325 */ 326 #define M2K(m) ((m) == UIO_READ ? KAUTH_REQ_PROCESS_CANPROCFS_READ : \ 327 KAUTH_REQ_PROCESS_CANPROCFS_WRITE) 328 error = kauth_authorize_process(curl->l_cred, KAUTH_PROCESS_CANPROCFS, 329 p, pfs, KAUTH_ARG(M2K(uio->uio_rw)), NULL); 330 if (error) 331 return (error); 332 #undef M2K 333 334 /* XXX NJWLWP 335 * The entire procfs interface needs work to be useful to 336 * a process with multiple LWPs. For the moment, we'll 337 * just kluge this and fail on others. 338 */ 339 l = proc_representative_lwp(p); 340 341 switch (pfs->pfs_type) { 342 case PFSnote: 343 case PFSnotepg: 344 return (procfs_donote(curl, p, pfs, uio)); 345 346 case PFSregs: 347 return (procfs_doregs(curl, l, pfs, uio)); 348 349 case PFSfpregs: 350 return (procfs_dofpregs(curl, l, pfs, uio)); 351 352 case PFSctl: 353 return (procfs_doctl(curl, l, pfs, uio)); 354 355 case PFSstatus: 356 return (procfs_dostatus(curl, l, pfs, uio)); 357 358 case PFSstat: 359 return (procfs_do_pid_stat(curl, l, pfs, uio)); 360 361 case PFSmap: 362 return (procfs_domap(curl, p, pfs, uio, 0)); 363 364 case PFSmaps: 365 return (procfs_domap(curl, p, pfs, uio, 1)); 366 367 case PFSmem: 368 return (procfs_domem(curl, l, pfs, uio)); 369 370 case PFScmdline: 371 return (procfs_docmdline(curl, p, pfs, uio)); 372 373 case PFSmeminfo: 374 return (procfs_domeminfo(curl, p, pfs, uio)); 375 376 case PFSdevices: 377 return (procfs_dodevices(curl, p, pfs, uio)); 378 379 case PFScpuinfo: 380 return (procfs_docpuinfo(curl, p, pfs, uio)); 381 382 case PFSfd: 383 return (procfs_dofd(curl, p, pfs, uio)); 384 385 case PFSuptime: 386 return (procfs_douptime(curl, p, pfs, uio)); 387 388 case PFSmounts: 389 return (procfs_domounts(curl, p, pfs, uio)); 390 391 case PFSemul: 392 return procfs_doemul(curl, p, pfs, uio); 393 394 #ifdef __HAVE_PROCFS_MACHDEP 395 PROCFS_MACHDEP_NODETYPE_CASES 396 return (procfs_machdep_rw(curl, l, pfs, uio)); 397 #endif 398 399 default: 400 return (EOPNOTSUPP); 401 } 402 } 403 404 /* 405 * Get a string from userland into (bf). Strip a trailing 406 * nl character (to allow easy access from the shell). 407 * The buffer should be *buflenp + 1 chars long. vfs_getuserstr 408 * will automatically add a nul char at the end. 409 * 410 * Returns 0 on success or the following errors 411 * 412 * EINVAL: file offset is non-zero. 413 * EMSGSIZE: message is longer than kernel buffer 414 * EFAULT: user i/o buffer is not addressable 415 */ 416 int 417 vfs_getuserstr(uio, bf, buflenp) 418 struct uio *uio; 419 char *bf; 420 int *buflenp; 421 { 422 int xlen; 423 int error; 424 425 if (uio->uio_offset != 0) 426 return (EINVAL); 427 428 xlen = *buflenp; 429 430 /* must be able to read the whole string in one go */ 431 if (xlen < uio->uio_resid) 432 return (EMSGSIZE); 433 xlen = uio->uio_resid; 434 435 if ((error = uiomove(bf, xlen, uio)) != 0) 436 return (error); 437 438 /* allow multiple writes without seeks */ 439 uio->uio_offset = 0; 440 441 /* cleanup string and remove trailing newline */ 442 bf[xlen] = '\0'; 443 xlen = strlen(bf); 444 if (xlen > 0 && bf[xlen-1] == '\n') 445 bf[--xlen] = '\0'; 446 *buflenp = xlen; 447 448 return (0); 449 } 450 451 const vfs_namemap_t * 452 vfs_findname(nm, bf, buflen) 453 const vfs_namemap_t *nm; 454 const char *bf; 455 int buflen; 456 { 457 458 for (; nm->nm_name; nm++) 459 if (memcmp(bf, nm->nm_name, buflen+1) == 0) 460 return (nm); 461 462 return (0); 463 } 464 465 /* 466 * Initialize pfsnode hash table. 467 */ 468 void 469 procfs_hashinit() 470 { 471 lockinit(&pfs_hashlock, PINOD, "pfs_hashlock", 0, 0); 472 pfs_hashtbl = hashinit(desiredvnodes / 4, HASH_LIST, M_UFSMNT, 473 M_WAITOK, &pfs_ihash); 474 simple_lock_init(&pfs_hash_slock); 475 } 476 477 void 478 procfs_hashreinit() 479 { 480 struct pfsnode *pp; 481 struct pfs_hashhead *oldhash, *hash; 482 u_long i, oldmask, mask, val; 483 484 hash = hashinit(desiredvnodes / 4, HASH_LIST, M_UFSMNT, M_WAITOK, 485 &mask); 486 487 simple_lock(&pfs_hash_slock); 488 oldhash = pfs_hashtbl; 489 oldmask = pfs_ihash; 490 pfs_hashtbl = hash; 491 pfs_ihash = mask; 492 for (i = 0; i <= oldmask; i++) { 493 while ((pp = LIST_FIRST(&oldhash[i])) != NULL) { 494 LIST_REMOVE(pp, pfs_hash); 495 val = PFSPIDHASH(pp->pfs_pid); 496 LIST_INSERT_HEAD(&hash[val], pp, pfs_hash); 497 } 498 } 499 simple_unlock(&pfs_hash_slock); 500 hashdone(oldhash, M_UFSMNT); 501 } 502 503 /* 504 * Free pfsnode hash table. 505 */ 506 void 507 procfs_hashdone() 508 { 509 hashdone(pfs_hashtbl, M_UFSMNT); 510 } 511 512 struct vnode * 513 procfs_hashget(pid, type, fd, mp) 514 pid_t pid; 515 pfstype type; 516 int fd; 517 struct mount *mp; 518 { 519 struct pfs_hashhead *ppp; 520 struct pfsnode *pp; 521 struct vnode *vp; 522 523 loop: 524 simple_lock(&pfs_hash_slock); 525 ppp = &pfs_hashtbl[PFSPIDHASH(pid)]; 526 LIST_FOREACH(pp, ppp, pfs_hash) { 527 vp = PFSTOV(pp); 528 if (pid == pp->pfs_pid && pp->pfs_type == type && 529 pp->pfs_fd == fd && vp->v_mount == mp) { 530 simple_lock(&vp->v_interlock); 531 simple_unlock(&pfs_hash_slock); 532 if (vget(vp, LK_EXCLUSIVE | LK_INTERLOCK)) 533 goto loop; 534 return (vp); 535 } 536 } 537 simple_unlock(&pfs_hash_slock); 538 return (NULL); 539 } 540 541 /* 542 * Insert the pfsnode into the hash table and lock it. 543 */ 544 void 545 procfs_hashins(pp) 546 struct pfsnode *pp; 547 { 548 struct pfs_hashhead *ppp; 549 550 /* lock the pfsnode, then put it on the appropriate hash list */ 551 lockmgr(&pp->pfs_vnode->v_lock, LK_EXCLUSIVE, (struct simplelock *)0); 552 553 simple_lock(&pfs_hash_slock); 554 ppp = &pfs_hashtbl[PFSPIDHASH(pp->pfs_pid)]; 555 LIST_INSERT_HEAD(ppp, pp, pfs_hash); 556 simple_unlock(&pfs_hash_slock); 557 } 558 559 /* 560 * Remove the pfsnode from the hash table. 561 */ 562 void 563 procfs_hashrem(pp) 564 struct pfsnode *pp; 565 { 566 simple_lock(&pfs_hash_slock); 567 LIST_REMOVE(pp, pfs_hash); 568 simple_unlock(&pfs_hash_slock); 569 } 570 571 void 572 procfs_revoke_vnodes(p, arg) 573 struct proc *p; 574 void *arg; 575 { 576 struct pfsnode *pfs, *pnext; 577 struct vnode *vp; 578 struct mount *mp = (struct mount *)arg; 579 struct pfs_hashhead *ppp; 580 581 if (!(p->p_flag & P_SUGID)) 582 return; 583 584 ppp = &pfs_hashtbl[PFSPIDHASH(p->p_pid)]; 585 for (pfs = LIST_FIRST(ppp); pfs; pfs = pnext) { 586 vp = PFSTOV(pfs); 587 pnext = LIST_NEXT(pfs, pfs_hash); 588 if (vp->v_usecount > 0 && pfs->pfs_pid == p->p_pid && 589 vp->v_mount == mp) 590 VOP_REVOKE(vp, REVOKEALL); 591 } 592 } 593 594 int 595 procfs_getfp(pfs, pown, fp) 596 struct pfsnode *pfs; 597 struct proc **pown; 598 struct file **fp; 599 { 600 struct proc *p = PFIND(pfs->pfs_pid); 601 602 if (p == NULL) 603 return ESRCH; 604 605 if (pfs->pfs_fd == -1) 606 return EINVAL; 607 608 if ((*fp = fd_getfile(p->p_fd, pfs->pfs_fd)) == NULL) 609 return EBADF; 610 611 *pown = p; 612 return 0; 613 } 614 615 int 616 procfs_doemul(struct lwp *curl, struct proc *p, 617 struct pfsnode *pfs, struct uio *uio) 618 { 619 const char *ename = p->p_emul->e_name; 620 return uiomove_frombuf(__UNCONST(ename), strlen(ename), uio); 621 } 622
Indexes created Tue Sep 23 02:09:52 GMT 2025