procfs_subr.c revision 1.70
1 /* $NetBSD: procfs_subr.c,v 1.70 2006/10/25 18:59:52 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 1.70 2006/10/25 18:59:52 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 89 #include <miscfs/procfs/procfs.h> 90 91 void procfs_hashins(struct pfsnode *); 92 void procfs_hashrem(struct pfsnode *); 93 struct vnode *procfs_hashget(pid_t, pfstype, int, struct mount *); 94 95 LIST_HEAD(pfs_hashhead, pfsnode) *pfs_hashtbl; 96 u_long pfs_ihash; /* size of hash table - 1 */ 97 #define PFSPIDHASH(pid) ((pid) & pfs_ihash) 98 99 struct lock pfs_hashlock; 100 struct simplelock pfs_hash_slock; 101 102 #define ISSET(t, f) ((t) & (f)) 103 104 /* 105 * allocate a pfsnode/vnode pair. the vnode is 106 * referenced, and locked. 107 * 108 * the pid, pfs_type, and mount point uniquely 109 * identify a pfsnode. the mount point is needed 110 * because someone might mount this filesystem 111 * twice. 112 * 113 * all pfsnodes are maintained on a singly-linked 114 * list. new nodes are only allocated when they cannot 115 * be found on this list. entries on the list are 116 * removed when the vfs reclaim entry is called. 117 * 118 * a single lock is kept for the entire list. this is 119 * needed because the getnewvnode() function can block 120 * waiting for a vnode to become free, in which case there 121 * may be more than one process trying to get the same 122 * vnode. this lock is only taken if we are going to 123 * call getnewvnode, since the kernel itself is single-threaded. 124 * 125 * if an entry is found on the list, then call vget() to 126 * take a reference. this is done because there may be 127 * zero references to it and so it needs to removed from 128 * the vnode free list. 129 */ 130 int 131 procfs_allocvp(mp, vpp, pid, pfs_type, fd) 132 struct mount *mp; 133 struct vnode **vpp; 134 pid_t pid; 135 pfstype pfs_type; 136 int fd; 137 { 138 struct pfsnode *pfs; 139 struct vnode *vp; 140 int error; 141 142 do { 143 if ((*vpp = procfs_hashget(pid, pfs_type, fd, mp)) != NULL) 144 return (0); 145 } while (lockmgr(&pfs_hashlock, LK_EXCLUSIVE|LK_SLEEPFAIL, 0)); 146 147 if ((error = getnewvnode(VT_PROCFS, mp, procfs_vnodeop_p, &vp)) != 0) { 148 *vpp = NULL; 149 lockmgr(&pfs_hashlock, LK_RELEASE, NULL); 150 return (error); 151 } 152 153 MALLOC(pfs, void *, sizeof(struct pfsnode), M_TEMP, M_WAITOK); 154 vp->v_data = pfs; 155 156 pfs->pfs_pid = pid; 157 pfs->pfs_type = pfs_type; 158 pfs->pfs_vnode = vp; 159 pfs->pfs_flags = 0; 160 pfs->pfs_fileno = PROCFS_FILENO(pid, pfs_type, fd); 161 pfs->pfs_fd = fd; 162 163 switch (pfs_type) { 164 case PFSroot: /* /proc = dr-xr-xr-x */ 165 pfs->pfs_mode = S_IRUSR|S_IXUSR|S_IRGRP|S_IXGRP|S_IROTH|S_IXOTH; 166 vp->v_type = VDIR; 167 vp->v_flag = VROOT; 168 break; 169 170 case PFScurproc: /* /proc/curproc = lr-xr-xr-x */ 171 case PFSself: /* /proc/self = lr-xr-xr-x */ 172 case PFScwd: /* /proc/N/cwd = lr-xr-xr-x */ 173 case PFSchroot: /* /proc/N/chroot = lr-xr-xr-x */ 174 case PFSexe: /* /proc/N/exe = lr-xr-xr-x */ 175 pfs->pfs_mode = S_IRUSR|S_IXUSR|S_IRGRP|S_IXGRP|S_IROTH|S_IXOTH; 176 vp->v_type = VLNK; 177 break; 178 179 case PFSproc: /* /proc/N = dr-xr-xr-x */ 180 case PFSfd: 181 if (fd == -1) { /* /proc/N/fd = dr-xr-xr-x */ 182 pfs->pfs_mode = S_IRUSR|S_IXUSR; 183 vp->v_type = VDIR; 184 } else { /* /proc/N/fd/M = [ps-]rw------- */ 185 struct file *fp; 186 struct vnode *vxp; 187 struct proc *pown; 188 189 /* XXX can procfs_getfp() ever fail here? */ 190 if ((error = procfs_getfp(pfs, &pown, &fp)) != 0) 191 goto bad; 192 FILE_USE(fp); 193 194 pfs->pfs_mode = S_IRUSR|S_IWUSR; 195 switch (fp->f_type) { 196 case DTYPE_VNODE: 197 vxp = (struct vnode *)fp->f_data; 198 199 /* 200 * We make symlinks for directories 201 * to avoid cycles. 202 */ 203 if (vxp->v_type == VDIR) 204 goto symlink; 205 vp->v_type = vxp->v_type; 206 break; 207 case DTYPE_PIPE: 208 vp->v_type = VFIFO; 209 break; 210 case DTYPE_SOCKET: 211 vp->v_type = VSOCK; 212 break; 213 case DTYPE_KQUEUE: 214 case DTYPE_MISC: 215 symlink: 216 pfs->pfs_mode = S_IRUSR|S_IXUSR|S_IRGRP| 217 S_IXGRP|S_IROTH|S_IXOTH; 218 vp->v_type = VLNK; 219 break; 220 default: 221 error = EOPNOTSUPP; 222 FILE_UNUSE(fp, proc_representative_lwp(pown)); 223 goto bad; 224 } 225 FILE_UNUSE(fp, proc_representative_lwp(pown)); 226 } 227 break; 228 229 case PFSfile: /* /proc/N/file = -rw------- */ 230 case PFSmem: /* /proc/N/mem = -rw------- */ 231 case PFSregs: /* /proc/N/regs = -rw------- */ 232 case PFSfpregs: /* /proc/N/fpregs = -rw------- */ 233 pfs->pfs_mode = S_IRUSR|S_IWUSR; 234 vp->v_type = VREG; 235 break; 236 237 case PFSctl: /* /proc/N/ctl = --w------ */ 238 case PFSnote: /* /proc/N/note = --w------ */ 239 case PFSnotepg: /* /proc/N/notepg = --w------ */ 240 pfs->pfs_mode = S_IWUSR; 241 vp->v_type = VREG; 242 break; 243 244 case PFSmap: /* /proc/N/map = -r--r--r-- */ 245 case PFSmaps: /* /proc/N/maps = -r--r--r-- */ 246 case PFSstatus: /* /proc/N/status = -r--r--r-- */ 247 case PFSstat: /* /proc/N/stat = -r--r--r-- */ 248 case PFScmdline: /* /proc/N/cmdline = -r--r--r-- */ 249 case PFSmeminfo: /* /proc/meminfo = -r--r--r-- */ 250 case PFSdevices: /* /proc/devices = -r--r--r-- */ 251 case PFScpuinfo: /* /proc/cpuinfo = -r--r--r-- */ 252 case PFSuptime: /* /proc/uptime = -r--r--r-- */ 253 case PFSmounts: /* /proc/mounts = -r--r--r-- */ 254 pfs->pfs_mode = S_IRUSR|S_IRGRP|S_IROTH; 255 vp->v_type = VREG; 256 break; 257 258 #ifdef __HAVE_PROCFS_MACHDEP 259 PROCFS_MACHDEP_NODETYPE_CASES 260 procfs_machdep_allocvp(vp); 261 break; 262 #endif 263 264 default: 265 panic("procfs_allocvp"); 266 } 267 268 procfs_hashins(pfs); 269 uvm_vnp_setsize(vp, 0); 270 lockmgr(&pfs_hashlock, LK_RELEASE, NULL); 271 272 *vpp = vp; 273 return (0); 274 275 bad: 276 lockmgr(&pfs_hashlock, LK_RELEASE, NULL); 277 FREE(pfs, M_TEMP); 278 ungetnewvnode(vp); 279 return (error); 280 } 281 282 int 283 procfs_freevp(vp) 284 struct vnode *vp; 285 { 286 struct pfsnode *pfs = VTOPFS(vp); 287 288 procfs_hashrem(pfs); 289 290 FREE(vp->v_data, M_TEMP); 291 vp->v_data = 0; 292 return (0); 293 } 294 295 int 296 procfs_rw(v) 297 void *v; 298 { 299 struct vop_read_args *ap = v; 300 struct vnode *vp = ap->a_vp; 301 struct uio *uio = ap->a_uio; 302 struct lwp *curl; 303 struct lwp *l; 304 struct pfsnode *pfs = VTOPFS(vp); 305 struct proc *p; 306 307 if (uio->uio_offset < 0) 308 return EINVAL; 309 p = PFIND(pfs->pfs_pid); 310 if (p == 0) 311 return ESRCH; 312 /* 313 * Do not allow init to be modified while in secure mode; it 314 * could be duped into changing the security level. 315 */ 316 if (uio->uio_rw == UIO_WRITE && p == initproc && securelevel > -1) 317 return EPERM; 318 319 curl = curlwp; 320 321 /* XXX NJWLWP 322 * The entire procfs interface needs work to be useful to 323 * a process with multiple LWPs. For the moment, we'll 324 * just kluge this and fail on others. 325 */ 326 l = proc_representative_lwp(p); 327 328 switch (pfs->pfs_type) { 329 case PFSnote: 330 case PFSnotepg: 331 return (procfs_donote(curl, p, pfs, uio)); 332 333 case PFSregs: 334 return (procfs_doregs(curl, l, pfs, uio)); 335 336 case PFSfpregs: 337 return (procfs_dofpregs(curl, l, pfs, uio)); 338 339 case PFSctl: 340 return (procfs_doctl(curl, l, pfs, uio)); 341 342 case PFSstatus: 343 return (procfs_dostatus(curl, l, pfs, uio)); 344 345 case PFSstat: 346 return (procfs_do_pid_stat(curl, l, pfs, uio)); 347 348 case PFSmap: 349 return (procfs_domap(curl, p, pfs, uio, 0)); 350 351 case PFSmaps: 352 return (procfs_domap(curl, p, pfs, uio, 1)); 353 354 case PFSmem: 355 return (procfs_domem(curl, l, pfs, uio)); 356 357 case PFScmdline: 358 return (procfs_docmdline(curl, p, pfs, uio)); 359 360 case PFSmeminfo: 361 return (procfs_domeminfo(curl, p, pfs, uio)); 362 363 case PFSdevices: 364 return (procfs_dodevices(curl, p, pfs, uio)); 365 366 case PFScpuinfo: 367 return (procfs_docpuinfo(curl, p, pfs, uio)); 368 369 case PFSfd: 370 return (procfs_dofd(curl, p, pfs, uio)); 371 372 case PFSuptime: 373 return (procfs_douptime(curl, p, pfs, uio)); 374 375 case PFSmounts: 376 return (procfs_domounts(curl, p, pfs, uio)); 377 378 #ifdef __HAVE_PROCFS_MACHDEP 379 PROCFS_MACHDEP_NODETYPE_CASES 380 return (procfs_machdep_rw(curl, l, pfs, uio)); 381 #endif 382 383 default: 384 return (EOPNOTSUPP); 385 } 386 } 387 388 /* 389 * Get a string from userland into (bf). Strip a trailing 390 * nl character (to allow easy access from the shell). 391 * The buffer should be *buflenp + 1 chars long. vfs_getuserstr 392 * will automatically add a nul char at the end. 393 * 394 * Returns 0 on success or the following errors 395 * 396 * EINVAL: file offset is non-zero. 397 * EMSGSIZE: message is longer than kernel buffer 398 * EFAULT: user i/o buffer is not addressable 399 */ 400 int 401 vfs_getuserstr(uio, bf, buflenp) 402 struct uio *uio; 403 char *bf; 404 int *buflenp; 405 { 406 int xlen; 407 int error; 408 409 if (uio->uio_offset != 0) 410 return (EINVAL); 411 412 xlen = *buflenp; 413 414 /* must be able to read the whole string in one go */ 415 if (xlen < uio->uio_resid) 416 return (EMSGSIZE); 417 xlen = uio->uio_resid; 418 419 if ((error = uiomove(bf, xlen, uio)) != 0) 420 return (error); 421 422 /* allow multiple writes without seeks */ 423 uio->uio_offset = 0; 424 425 /* cleanup string and remove trailing newline */ 426 bf[xlen] = '\0'; 427 xlen = strlen(bf); 428 if (xlen > 0 && bf[xlen-1] == '\n') 429 bf[--xlen] = '\0'; 430 *buflenp = xlen; 431 432 return (0); 433 } 434 435 const vfs_namemap_t * 436 vfs_findname(nm, bf, buflen) 437 const vfs_namemap_t *nm; 438 const char *bf; 439 int buflen; 440 { 441 442 for (; nm->nm_name; nm++) 443 if (memcmp(bf, nm->nm_name, buflen+1) == 0) 444 return (nm); 445 446 return (0); 447 } 448 449 /* 450 * Initialize pfsnode hash table. 451 */ 452 void 453 procfs_hashinit() 454 { 455 lockinit(&pfs_hashlock, PINOD, "pfs_hashlock", 0, 0); 456 pfs_hashtbl = hashinit(desiredvnodes / 4, HASH_LIST, M_UFSMNT, 457 M_WAITOK, &pfs_ihash); 458 simple_lock_init(&pfs_hash_slock); 459 } 460 461 void 462 procfs_hashreinit() 463 { 464 struct pfsnode *pp; 465 struct pfs_hashhead *oldhash, *hash; 466 u_long i, oldmask, mask, val; 467 468 hash = hashinit(desiredvnodes / 4, HASH_LIST, M_UFSMNT, M_WAITOK, 469 &mask); 470 471 simple_lock(&pfs_hash_slock); 472 oldhash = pfs_hashtbl; 473 oldmask = pfs_ihash; 474 pfs_hashtbl = hash; 475 pfs_ihash = mask; 476 for (i = 0; i <= oldmask; i++) { 477 while ((pp = LIST_FIRST(&oldhash[i])) != NULL) { 478 LIST_REMOVE(pp, pfs_hash); 479 val = PFSPIDHASH(pp->pfs_pid); 480 LIST_INSERT_HEAD(&hash[val], pp, pfs_hash); 481 } 482 } 483 simple_unlock(&pfs_hash_slock); 484 hashdone(oldhash, M_UFSMNT); 485 } 486 487 /* 488 * Free pfsnode hash table. 489 */ 490 void 491 procfs_hashdone() 492 { 493 hashdone(pfs_hashtbl, M_UFSMNT); 494 } 495 496 struct vnode * 497 procfs_hashget(pid, type, fd, mp) 498 pid_t pid; 499 pfstype type; 500 int fd; 501 struct mount *mp; 502 { 503 struct pfs_hashhead *ppp; 504 struct pfsnode *pp; 505 struct vnode *vp; 506 507 loop: 508 simple_lock(&pfs_hash_slock); 509 ppp = &pfs_hashtbl[PFSPIDHASH(pid)]; 510 LIST_FOREACH(pp, ppp, pfs_hash) { 511 vp = PFSTOV(pp); 512 if (pid == pp->pfs_pid && pp->pfs_type == type && 513 pp->pfs_fd == fd && vp->v_mount == mp) { 514 simple_lock(&vp->v_interlock); 515 simple_unlock(&pfs_hash_slock); 516 if (vget(vp, LK_EXCLUSIVE | LK_INTERLOCK)) 517 goto loop; 518 return (vp); 519 } 520 } 521 simple_unlock(&pfs_hash_slock); 522 return (NULL); 523 } 524 525 /* 526 * Insert the pfsnode into the hash table and lock it. 527 */ 528 void 529 procfs_hashins(pp) 530 struct pfsnode *pp; 531 { 532 struct pfs_hashhead *ppp; 533 534 /* lock the pfsnode, then put it on the appropriate hash list */ 535 lockmgr(&pp->pfs_vnode->v_lock, LK_EXCLUSIVE, (struct simplelock *)0); 536 537 simple_lock(&pfs_hash_slock); 538 ppp = &pfs_hashtbl[PFSPIDHASH(pp->pfs_pid)]; 539 LIST_INSERT_HEAD(ppp, pp, pfs_hash); 540 simple_unlock(&pfs_hash_slock); 541 } 542 543 /* 544 * Remove the pfsnode from the hash table. 545 */ 546 void 547 procfs_hashrem(pp) 548 struct pfsnode *pp; 549 { 550 simple_lock(&pfs_hash_slock); 551 LIST_REMOVE(pp, pfs_hash); 552 simple_unlock(&pfs_hash_slock); 553 } 554 555 void 556 procfs_revoke_vnodes(p, arg) 557 struct proc *p; 558 void *arg; 559 { 560 struct pfsnode *pfs, *pnext; 561 struct vnode *vp; 562 struct mount *mp = (struct mount *)arg; 563 struct pfs_hashhead *ppp; 564 565 if (!(p->p_flag & P_SUGID)) 566 return; 567 568 ppp = &pfs_hashtbl[PFSPIDHASH(p->p_pid)]; 569 for (pfs = LIST_FIRST(ppp); pfs; pfs = pnext) { 570 vp = PFSTOV(pfs); 571 pnext = LIST_NEXT(pfs, pfs_hash); 572 if (vp->v_usecount > 0 && pfs->pfs_pid == p->p_pid && 573 vp->v_mount == mp) 574 VOP_REVOKE(vp, REVOKEALL); 575 } 576 } 577 578 int 579 procfs_getfp(pfs, pown, fp) 580 struct pfsnode *pfs; 581 struct proc **pown; 582 struct file **fp; 583 { 584 struct proc *p = PFIND(pfs->pfs_pid); 585 586 if (p == NULL) 587 return ESRCH; 588 589 if (pfs->pfs_fd == -1) 590 return EINVAL; 591 592 if ((*fp = fd_getfile(p->p_fd, pfs->pfs_fd)) == NULL) 593 return EBADF; 594 595 *pown = p; 596 return 0; 597 } 598
Indexes created Tue Sep 23 02:09:52 GMT 2025