1 /* $NetBSD: procfs_vnops.c,v 1.166 2008/03/21 21:55:00 ad Exp $ */ 2 3 /*- 4 * Copyright (c) 2006, 2007, 2008 The NetBSD Foundation, Inc. 5 * All rights reserved. 6 * 7 * This code is derived from software contributed to The NetBSD Foundation 8 * by Andrew Doran. 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. All advertising materials mentioning features or use of this software 19 * must display the following acknowledgement: 20 * This product includes software developed by the NetBSD 21 * Foundation, Inc. and its contributors. 22 * 4. Neither the name of The NetBSD Foundation nor the names of its 23 * contributors may be used to endorse or promote products derived 24 * from this software without specific prior written permission. 25 * 26 * THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS 27 * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED 28 * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR 29 * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE FOUNDATION OR CONTRIBUTORS 30 * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR 31 * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF 32 * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS 33 * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN 34 * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) 35 * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE 36 * POSSIBILITY OF SUCH DAMAGE. 37 */ 38 39 /* 40 * Copyright (c) 1993, 1995 41 * The Regents of the University of California. All rights reserved. 42 * 43 * This code is derived from software contributed to Berkeley by 44 * Jan-Simon Pendry. 45 * 46 * Redistribution and use in source and binary forms, with or without 47 * modification, are permitted provided that the following conditions 48 * are met: 49 * 1. Redistributions of source code must retain the above copyright 50 * notice, this list of conditions and the following disclaimer. 51 * 2. Redistributions in binary form must reproduce the above copyright 52 * notice, this list of conditions and the following disclaimer in the 53 * documentation and/or other materials provided with the distribution. 54 * 3. Neither the name of the University nor the names of its contributors 55 * may be used to endorse or promote products derived from this software 56 * without specific prior written permission. 57 * 58 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 59 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 60 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 61 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 62 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 63 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 64 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 65 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 66 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 67 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 68 * SUCH DAMAGE. 69 * 70 * @(#)procfs_vnops.c 8.18 (Berkeley) 5/21/95 71 */ 72 73 /* 74 * Copyright (c) 1993 Jan-Simon Pendry 75 * 76 * This code is derived from software contributed to Berkeley by 77 * Jan-Simon Pendry. 78 * 79 * Redistribution and use in source and binary forms, with or without 80 * modification, are permitted provided that the following conditions 81 * are met: 82 * 1. Redistributions of source code must retain the above copyright 83 * notice, this list of conditions and the following disclaimer. 84 * 2. Redistributions in binary form must reproduce the above copyright 85 * notice, this list of conditions and the following disclaimer in the 86 * documentation and/or other materials provided with the distribution. 87 * 3. All advertising materials mentioning features or use of this software 88 * must display the following acknowledgement: 89 * This product includes software developed by the University of 90 * California, Berkeley and its contributors. 91 * 4. Neither the name of the University nor the names of its contributors 92 * may be used to endorse or promote products derived from this software 93 * without specific prior written permission. 94 * 95 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 96 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 97 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 98 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 99 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 100 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 101 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 102 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 103 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 104 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 105 * SUCH DAMAGE. 106 * 107 * @(#)procfs_vnops.c 8.18 (Berkeley) 5/21/95 108 */ 109 110 /* 111 * procfs vnode interface 112 */ 113 114 #include <sys/cdefs.h> 115 __KERNEL_RCSID(0, "$NetBSD: procfs_vnops.c,v 1.166 2008/03/21 21:55:00 ad Exp $"); 116 117 #include <sys/param.h> 118 #include <sys/systm.h> 119 #include <sys/time.h> 120 #include <sys/kernel.h> 121 #include <sys/file.h> 122 #include <sys/filedesc.h> 123 #include <sys/proc.h> 124 #include <sys/vnode.h> 125 #include <sys/namei.h> 126 #include <sys/malloc.h> 127 #include <sys/mount.h> 128 #include <sys/dirent.h> 129 #include <sys/resourcevar.h> 130 #include <sys/stat.h> 131 #include <sys/ptrace.h> 132 #include <sys/kauth.h> 133 134 #include <uvm/uvm_extern.h> /* for PAGE_SIZE */ 135 136 #include <machine/reg.h> 137 138 #include <miscfs/genfs/genfs.h> 139 #include <miscfs/procfs/procfs.h> 140 141 /* 142 * Vnode Operations. 143 * 144 */ 145 146 static int procfs_validfile_linux(struct lwp *, struct mount *); 147 static int procfs_root_readdir_callback(struct proc *, void *); 148 static void procfs_dir(pfstype, struct lwp *, struct proc *, char **, char *, 149 size_t); 150 151 /* 152 * This is a list of the valid names in the 153 * process-specific sub-directories. It is 154 * used in procfs_lookup and procfs_readdir 155 */ 156 static const struct proc_target { 157 u_char pt_type; 158 u_char pt_namlen; 159 const char *pt_name; 160 pfstype pt_pfstype; 161 int (*pt_valid)(struct lwp *, struct mount *); 162 } proc_targets[] = { 163 #define N(s) sizeof(s)-1, s 164 /* name type validp */ 165 { DT_DIR, N("."), PFSproc, NULL }, 166 { DT_DIR, N(".."), PFSroot, NULL }, 167 { DT_DIR, N("fd"), PFSfd, NULL }, 168 { DT_REG, N("file"), PFSfile, procfs_validfile }, 169 { DT_REG, N("mem"), PFSmem, NULL }, 170 { DT_REG, N("regs"), PFSregs, procfs_validregs }, 171 { DT_REG, N("fpregs"), PFSfpregs, procfs_validfpregs }, 172 { DT_REG, N("ctl"), PFSctl, NULL }, 173 { DT_REG, N("stat"), PFSstat, procfs_validfile_linux }, 174 { DT_REG, N("status"), PFSstatus, NULL }, 175 { DT_REG, N("note"), PFSnote, NULL }, 176 { DT_REG, N("notepg"), PFSnotepg, NULL }, 177 { DT_REG, N("map"), PFSmap, procfs_validmap }, 178 { DT_REG, N("maps"), PFSmaps, procfs_validmap }, 179 { DT_REG, N("cmdline"), PFScmdline, NULL }, 180 { DT_REG, N("exe"), PFSexe, procfs_validfile }, 181 { DT_LNK, N("cwd"), PFScwd, NULL }, 182 { DT_LNK, N("root"), PFSchroot, NULL }, 183 { DT_LNK, N("emul"), PFSemul, NULL }, 184 { DT_REG, N("statm"), PFSstatm, procfs_validfile_linux }, 185 #ifdef __HAVE_PROCFS_MACHDEP 186 PROCFS_MACHDEP_NODETYPE_DEFNS 187 #endif 188 #undef N 189 }; 190 static const int nproc_targets = sizeof(proc_targets) / sizeof(proc_targets[0]); 191 192 /* 193 * List of files in the root directory. Note: the validate function will 194 * be called with p == NULL for these ones. 195 */ 196 static const struct proc_target proc_root_targets[] = { 197 #define N(s) sizeof(s)-1, s 198 /* name type validp */ 199 { DT_REG, N("meminfo"), PFSmeminfo, procfs_validfile_linux }, 200 { DT_REG, N("cpuinfo"), PFScpuinfo, procfs_validfile_linux }, 201 { DT_REG, N("uptime"), PFSuptime, procfs_validfile_linux }, 202 { DT_REG, N("mounts"), PFSmounts, procfs_validfile_linux }, 203 { DT_REG, N("devices"), PFSdevices, procfs_validfile_linux }, 204 { DT_REG, N("stat"), PFScpustat, procfs_validfile_linux }, 205 { DT_REG, N("loadavg"), PFSloadavg, procfs_validfile_linux }, 206 #undef N 207 }; 208 static const int nproc_root_targets = 209 sizeof(proc_root_targets) / sizeof(proc_root_targets[0]); 210 211 int procfs_lookup(void *); 212 #define procfs_create genfs_eopnotsupp 213 #define procfs_mknod genfs_eopnotsupp 214 int procfs_open(void *); 215 int procfs_close(void *); 216 int procfs_access(void *); 217 int procfs_getattr(void *); 218 int procfs_setattr(void *); 219 #define procfs_read procfs_rw 220 #define procfs_write procfs_rw 221 #define procfs_fcntl genfs_fcntl 222 #define procfs_ioctl genfs_enoioctl 223 #define procfs_poll genfs_poll 224 #define procfs_revoke genfs_revoke 225 #define procfs_fsync genfs_nullop 226 #define procfs_seek genfs_nullop 227 #define procfs_remove genfs_eopnotsupp 228 int procfs_link(void *); 229 #define procfs_rename genfs_eopnotsupp 230 #define procfs_mkdir genfs_eopnotsupp 231 #define procfs_rmdir genfs_eopnotsupp 232 int procfs_symlink(void *); 233 int procfs_readdir(void *); 234 int procfs_readlink(void *); 235 #define procfs_abortop genfs_abortop 236 int procfs_inactive(void *); 237 int procfs_reclaim(void *); 238 #define procfs_lock genfs_lock 239 #define procfs_unlock genfs_unlock 240 #define procfs_bmap genfs_badop 241 #define procfs_strategy genfs_badop 242 int procfs_print(void *); 243 int procfs_pathconf(void *); 244 #define procfs_islocked genfs_islocked 245 #define procfs_advlock genfs_einval 246 #define procfs_bwrite genfs_eopnotsupp 247 #define procfs_putpages genfs_null_putpages 248 249 static int atoi(const char *, size_t); 250 251 /* 252 * procfs vnode operations. 253 */ 254 int (**procfs_vnodeop_p)(void *); 255 const struct vnodeopv_entry_desc procfs_vnodeop_entries[] = { 256 { &vop_default_desc, vn_default_error }, 257 { &vop_lookup_desc, procfs_lookup }, /* lookup */ 258 { &vop_create_desc, procfs_create }, /* create */ 259 { &vop_mknod_desc, procfs_mknod }, /* mknod */ 260 { &vop_open_desc, procfs_open }, /* open */ 261 { &vop_close_desc, procfs_close }, /* close */ 262 { &vop_access_desc, procfs_access }, /* access */ 263 { &vop_getattr_desc, procfs_getattr }, /* getattr */ 264 { &vop_setattr_desc, procfs_setattr }, /* setattr */ 265 { &vop_read_desc, procfs_read }, /* read */ 266 { &vop_write_desc, procfs_write }, /* write */ 267 { &vop_fcntl_desc, procfs_fcntl }, /* fcntl */ 268 { &vop_ioctl_desc, procfs_ioctl }, /* ioctl */ 269 { &vop_poll_desc, procfs_poll }, /* poll */ 270 { &vop_revoke_desc, procfs_revoke }, /* revoke */ 271 { &vop_fsync_desc, procfs_fsync }, /* fsync */ 272 { &vop_seek_desc, procfs_seek }, /* seek */ 273 { &vop_remove_desc, procfs_remove }, /* remove */ 274 { &vop_link_desc, procfs_link }, /* link */ 275 { &vop_rename_desc, procfs_rename }, /* rename */ 276 { &vop_mkdir_desc, procfs_mkdir }, /* mkdir */ 277 { &vop_rmdir_desc, procfs_rmdir }, /* rmdir */ 278 { &vop_symlink_desc, procfs_symlink }, /* symlink */ 279 { &vop_readdir_desc, procfs_readdir }, /* readdir */ 280 { &vop_readlink_desc, procfs_readlink }, /* readlink */ 281 { &vop_abortop_desc, procfs_abortop }, /* abortop */ 282 { &vop_inactive_desc, procfs_inactive }, /* inactive */ 283 { &vop_reclaim_desc, procfs_reclaim }, /* reclaim */ 284 { &vop_lock_desc, procfs_lock }, /* lock */ 285 { &vop_unlock_desc, procfs_unlock }, /* unlock */ 286 { &vop_bmap_desc, procfs_bmap }, /* bmap */ 287 { &vop_strategy_desc, procfs_strategy }, /* strategy */ 288 { &vop_print_desc, procfs_print }, /* print */ 289 { &vop_islocked_desc, procfs_islocked }, /* islocked */ 290 { &vop_pathconf_desc, procfs_pathconf }, /* pathconf */ 291 { &vop_advlock_desc, procfs_advlock }, /* advlock */ 292 { &vop_putpages_desc, procfs_putpages }, /* putpages */ 293 { NULL, NULL } 294 }; 295 const struct vnodeopv_desc procfs_vnodeop_opv_desc = 296 { &procfs_vnodeop_p, procfs_vnodeop_entries }; 297 /* 298 * set things up for doing i/o on 299 * the pfsnode (vp). (vp) is locked 300 * on entry, and should be left locked 301 * on exit. 302 * 303 * for procfs we don't need to do anything 304 * in particular for i/o. all that is done 305 * is to support exclusive open on process 306 * memory images. 307 */ 308 int 309 procfs_open(v) 310 void *v; 311 { 312 struct vop_open_args /* { 313 struct vnode *a_vp; 314 int a_mode; 315 kauth_cred_t a_cred; 316 } */ *ap = v; 317 struct pfsnode *pfs = VTOPFS(ap->a_vp); 318 struct lwp *l1; 319 struct proc *p2; 320 int error; 321 322 if ((error = procfs_proc_lock(pfs->pfs_pid, &p2, ENOENT)) != 0) 323 return error; 324 325 l1 = curlwp; /* tracer */ 326 327 #define M2K(m) (((m) & FREAD) && ((m) & FWRITE) ? \ 328 KAUTH_REQ_PROCESS_PROCFS_RW : \ 329 (m) & FWRITE ? KAUTH_REQ_PROCESS_PROCFS_WRITE : \ 330 KAUTH_REQ_PROCESS_PROCFS_READ) 331 332 mutex_enter(&p2->p_mutex); 333 error = kauth_authorize_process(l1->l_cred, KAUTH_PROCESS_PROCFS, 334 p2, pfs, KAUTH_ARG(M2K(ap->a_mode)), NULL); 335 mutex_exit(&p2->p_mutex); 336 if (error) { 337 procfs_proc_unlock(p2); 338 return (error); 339 } 340 341 #undef M2K 342 343 switch (pfs->pfs_type) { 344 case PFSmem: 345 if (((pfs->pfs_flags & FWRITE) && (ap->a_mode & O_EXCL)) || 346 ((pfs->pfs_flags & O_EXCL) && (ap->a_mode & FWRITE))) { 347 error = EBUSY; 348 break; 349 } 350 351 if (!proc_isunder(p2, l1)) { 352 error = EPERM; 353 break; 354 } 355 356 if (ap->a_mode & FWRITE) 357 pfs->pfs_flags = ap->a_mode & (FWRITE|O_EXCL); 358 359 break; 360 361 case PFSregs: 362 case PFSfpregs: 363 if (!proc_isunder(p2, l1)) { 364 error = EPERM; 365 break; 366 } 367 break; 368 369 default: 370 break; 371 } 372 373 procfs_proc_unlock(p2); 374 return (error); 375 } 376 377 /* 378 * close the pfsnode (vp) after doing i/o. 379 * (vp) is not locked on entry or exit. 380 * 381 * nothing to do for procfs other than undo 382 * any exclusive open flag (see _open above). 383 */ 384 int 385 procfs_close(v) 386 void *v; 387 { 388 struct vop_close_args /* { 389 struct vnode *a_vp; 390 int a_fflag; 391 kauth_cred_t a_cred; 392 } */ *ap = v; 393 struct pfsnode *pfs = VTOPFS(ap->a_vp); 394 395 switch (pfs->pfs_type) { 396 case PFSmem: 397 if ((ap->a_fflag & FWRITE) && (pfs->pfs_flags & O_EXCL)) 398 pfs->pfs_flags &= ~(FWRITE|O_EXCL); 399 break; 400 401 default: 402 break; 403 } 404 405 return (0); 406 } 407 408 /* 409 * _inactive is called when the pfsnode 410 * is vrele'd and the reference count goes 411 * to zero. (vp) will be on the vnode free 412 * list, so to get it back vget() must be 413 * used. 414 * 415 * (vp) is locked on entry, but must be unlocked on exit. 416 */ 417 int 418 procfs_inactive(v) 419 void *v; 420 { 421 struct vop_inactive_args /* { 422 struct vnode *a_vp; 423 bool *a_recycle; 424 } */ *ap = v; 425 struct vnode *vp = ap->a_vp; 426 struct pfsnode *pfs = VTOPFS(vp); 427 428 mutex_enter(&proclist_lock); 429 *ap->a_recycle = (p_find(pfs->pfs_pid, PFIND_LOCKED) == NULL); 430 mutex_exit(&proclist_lock); 431 432 VOP_UNLOCK(vp, 0); 433 434 return (0); 435 } 436 437 /* 438 * _reclaim is called when getnewvnode() 439 * wants to make use of an entry on the vnode 440 * free list. at this time the filesystem needs 441 * to free any private data and remove the node 442 * from any private lists. 443 */ 444 int 445 procfs_reclaim(v) 446 void *v; 447 { 448 struct vop_reclaim_args /* { 449 struct vnode *a_vp; 450 } */ *ap = v; 451 452 return (procfs_freevp(ap->a_vp)); 453 } 454 455 /* 456 * Return POSIX pathconf information applicable to special devices. 457 */ 458 int 459 procfs_pathconf(v) 460 void *v; 461 { 462 struct vop_pathconf_args /* { 463 struct vnode *a_vp; 464 int a_name; 465 register_t *a_retval; 466 } */ *ap = v; 467 468 switch (ap->a_name) { 469 case _PC_LINK_MAX: 470 *ap->a_retval = LINK_MAX; 471 return (0); 472 case _PC_MAX_CANON: 473 *ap->a_retval = MAX_CANON; 474 return (0); 475 case _PC_MAX_INPUT: 476 *ap->a_retval = MAX_INPUT; 477 return (0); 478 case _PC_PIPE_BUF: 479 *ap->a_retval = PIPE_BUF; 480 return (0); 481 case _PC_CHOWN_RESTRICTED: 482 *ap->a_retval = 1; 483 return (0); 484 case _PC_VDISABLE: 485 *ap->a_retval = _POSIX_VDISABLE; 486 return (0); 487 case _PC_SYNC_IO: 488 *ap->a_retval = 1; 489 return (0); 490 default: 491 return (EINVAL); 492 } 493 /* NOTREACHED */ 494 } 495 496 /* 497 * _print is used for debugging. 498 * just print a readable description 499 * of (vp). 500 */ 501 int 502 procfs_print(v) 503 void *v; 504 { 505 struct vop_print_args /* { 506 struct vnode *a_vp; 507 } */ *ap = v; 508 struct pfsnode *pfs = VTOPFS(ap->a_vp); 509 510 printf("tag VT_PROCFS, type %d, pid %d, mode %x, flags %lx\n", 511 pfs->pfs_type, pfs->pfs_pid, pfs->pfs_mode, pfs->pfs_flags); 512 return 0; 513 } 514 515 int 516 procfs_link(v) 517 void *v; 518 { 519 struct vop_link_args /* { 520 struct vnode *a_dvp; 521 struct vnode *a_vp; 522 struct componentname *a_cnp; 523 } */ *ap = v; 524 525 VOP_ABORTOP(ap->a_dvp, ap->a_cnp); 526 vput(ap->a_dvp); 527 return (EROFS); 528 } 529 530 int 531 procfs_symlink(v) 532 void *v; 533 { 534 struct vop_symlink_args /* { 535 struct vnode *a_dvp; 536 struct vnode **a_vpp; 537 struct componentname *a_cnp; 538 struct vattr *a_vap; 539 char *a_target; 540 } */ *ap = v; 541 542 VOP_ABORTOP(ap->a_dvp, ap->a_cnp); 543 vput(ap->a_dvp); 544 return (EROFS); 545 } 546 547 /* 548 * Works out the path to (and vnode of) the target process's current 549 * working directory or chroot. If the caller is in a chroot and 550 * can't "reach" the target's cwd or root (or some other error 551 * occurs), a "/" is returned for the path and a NULL pointer is 552 * returned for the vnode. 553 */ 554 static void 555 procfs_dir(pfstype t, struct lwp *caller, struct proc *target, char **bpp, 556 char *path, size_t len) 557 { 558 struct cwdinfo *cwdi; 559 struct vnode *vp, *rvp; 560 char *bp; 561 562 cwdi = caller->l_proc->p_cwdi; 563 rw_enter(&cwdi->cwdi_lock, RW_READER); 564 565 rvp = cwdi->cwdi_rdir; 566 bp = bpp ? *bpp : NULL; 567 568 switch (t) { 569 case PFScwd: 570 vp = target->p_cwdi->cwdi_cdir; 571 break; 572 case PFSchroot: 573 vp = target->p_cwdi->cwdi_rdir; 574 break; 575 case PFSexe: 576 vp = target->p_textvp; 577 break; 578 default: 579 rw_exit(&cwdi->cwdi_lock); 580 return; 581 } 582 583 /* 584 * XXX: this horrible kludge avoids locking panics when 585 * attempting to lookup links that point to within procfs 586 */ 587 if (vp != NULL && vp->v_tag == VT_PROCFS) { 588 if (bpp) { 589 *--bp = '/'; 590 *bpp = bp; 591 } 592 rw_exit(&cwdi->cwdi_lock); 593 return; 594 } 595 596 if (rvp == NULL) 597 rvp = rootvnode; 598 if (vp == NULL || getcwd_common(vp, rvp, bp ? &bp : NULL, path, 599 len / 2, 0, caller) != 0) { 600 vp = NULL; 601 if (bpp) { 602 /* 603 if (t == PFSexe) { 604 snprintf(path, len, "%s/%d/file" 605 mp->mnt_stat.f_mntonname, pfs->pfs_pid); 606 } else */ { 607 bp = *bpp; 608 *--bp = '/'; 609 } 610 } 611 } 612 613 if (bpp) 614 *bpp = bp; 615 616 rw_exit(&cwdi->cwdi_lock); 617 } 618 619 /* 620 * Invent attributes for pfsnode (vp) and store 621 * them in (vap). 622 * Directories lengths are returned as zero since 623 * any real length would require the genuine size 624 * to be computed, and nothing cares anyway. 625 * 626 * this is relatively minimal for procfs. 627 */ 628 int 629 procfs_getattr(v) 630 void *v; 631 { 632 struct vop_getattr_args /* { 633 struct vnode *a_vp; 634 struct vattr *a_vap; 635 kauth_cred_t a_cred; 636 } */ *ap = v; 637 struct pfsnode *pfs = VTOPFS(ap->a_vp); 638 struct vattr *vap = ap->a_vap; 639 struct proc *procp; 640 char *path; 641 int error; 642 643 /* first check the process still exists */ 644 switch (pfs->pfs_type) { 645 case PFSroot: 646 case PFScurproc: 647 case PFSself: 648 procp = NULL; 649 break; 650 651 default: 652 error = procfs_proc_lock(pfs->pfs_pid, &procp, ENOENT); 653 if (error != 0) 654 return (error); 655 break; 656 } 657 658 switch (pfs->pfs_type) { 659 case PFScwd: 660 case PFSchroot: 661 case PFSexe: 662 MALLOC(path, char *, MAXPATHLEN + 4, M_TEMP, 663 M_WAITOK|M_CANFAIL); 664 if (path == NULL && procp != NULL) { 665 procfs_proc_unlock(procp); 666 return (ENOMEM); 667 } 668 break; 669 670 default: 671 path = NULL; 672 break; 673 } 674 675 if (procp != NULL) { 676 mutex_enter(&procp->p_mutex); 677 error = kauth_authorize_process(kauth_cred_get(), 678 KAUTH_PROCESS_CANSEE, procp, 679 KAUTH_ARG(KAUTH_REQ_PROCESS_CANSEE_ENTRY), NULL, NULL); 680 mutex_exit(&procp->p_mutex); 681 if (error != 0) { 682 procfs_proc_unlock(procp); 683 if (path != NULL) 684 free(path, M_TEMP); 685 return (ENOENT); 686 } 687 } 688 689 error = 0; 690 691 /* start by zeroing out the attributes */ 692 VATTR_NULL(vap); 693 694 /* next do all the common fields */ 695 vap->va_type = ap->a_vp->v_type; 696 vap->va_mode = pfs->pfs_mode; 697 vap->va_fileid = pfs->pfs_fileno; 698 vap->va_flags = 0; 699 vap->va_blocksize = PAGE_SIZE; 700 701 /* 702 * Make all times be current TOD. 703 * 704 * It would be possible to get the process start 705 * time from the p_stats structure, but there's 706 * no "file creation" time stamp anyway, and the 707 * p_stats structure is not addressable if u. gets 708 * swapped out for that process. 709 */ 710 getnanotime(&vap->va_ctime); 711 vap->va_atime = vap->va_mtime = vap->va_ctime; 712 if (procp) 713 TIMEVAL_TO_TIMESPEC(&procp->p_stats->p_start, 714 &vap->va_birthtime); 715 else 716 getnanotime(&vap->va_birthtime); 717 718 switch (pfs->pfs_type) { 719 case PFSmem: 720 case PFSregs: 721 case PFSfpregs: 722 #if defined(__HAVE_PROCFS_MACHDEP) && defined(PROCFS_MACHDEP_PROTECT_CASES) 723 PROCFS_MACHDEP_PROTECT_CASES 724 #endif 725 /* 726 * If the process has exercised some setuid or setgid 727 * privilege, then rip away read/write permission so 728 * that only root can gain access. 729 */ 730 if (procp->p_flag & PK_SUGID) 731 vap->va_mode &= ~(S_IRUSR|S_IWUSR); 732 /* FALLTHROUGH */ 733 case PFSctl: 734 case PFSstatus: 735 case PFSstat: 736 case PFSnote: 737 case PFSnotepg: 738 case PFSmap: 739 case PFSmaps: 740 case PFScmdline: 741 case PFSemul: 742 case PFSstatm: 743 vap->va_nlink = 1; 744 vap->va_uid = kauth_cred_geteuid(procp->p_cred); 745 vap->va_gid = kauth_cred_getegid(procp->p_cred); 746 break; 747 case PFSmeminfo: 748 case PFSdevices: 749 case PFScpuinfo: 750 case PFSuptime: 751 case PFSmounts: 752 case PFScpustat: 753 case PFSloadavg: 754 vap->va_nlink = 1; 755 vap->va_uid = vap->va_gid = 0; 756 break; 757 758 default: 759 break; 760 } 761 762 /* 763 * now do the object specific fields 764 * 765 * The size could be set from struct reg, but it's hardly 766 * worth the trouble, and it puts some (potentially) machine 767 * dependent data into this machine-independent code. If it 768 * becomes important then this function should break out into 769 * a per-file stat function in the corresponding .c file. 770 */ 771 772 switch (pfs->pfs_type) { 773 case PFSroot: 774 /* 775 * Set nlink to 1 to tell fts(3) we don't actually know. 776 */ 777 vap->va_nlink = 1; 778 vap->va_uid = 0; 779 vap->va_gid = 0; 780 vap->va_bytes = vap->va_size = DEV_BSIZE; 781 break; 782 783 case PFSself: 784 case PFScurproc: { 785 char bf[16]; /* should be enough */ 786 vap->va_nlink = 1; 787 vap->va_uid = 0; 788 vap->va_gid = 0; 789 vap->va_bytes = vap->va_size = 790 snprintf(bf, sizeof(bf), "%ld", (long)curproc->p_pid); 791 break; 792 } 793 794 case PFSfd: 795 if (pfs->pfs_fd != -1) { 796 file_t *fp; 797 798 fp = fd_getfile2(procp, pfs->pfs_fd); 799 if (fp == NULL) { 800 error = EBADF; 801 break; 802 } 803 vap->va_nlink = 1; 804 vap->va_uid = kauth_cred_geteuid(fp->f_cred); 805 vap->va_gid = kauth_cred_getegid(fp->f_cred); 806 switch (fp->f_type) { 807 case DTYPE_VNODE: 808 vap->va_bytes = vap->va_size = 809 ((struct vnode *)fp->f_data)->v_size; 810 break; 811 default: 812 vap->va_bytes = vap->va_size = 0; 813 break; 814 } 815 closef(fp); 816 break; 817 } 818 /*FALLTHROUGH*/ 819 case PFSproc: 820 vap->va_nlink = 2; 821 vap->va_uid = kauth_cred_geteuid(procp->p_cred); 822 vap->va_gid = kauth_cred_getegid(procp->p_cred); 823 vap->va_bytes = vap->va_size = DEV_BSIZE; 824 break; 825 826 case PFSfile: 827 error = EOPNOTSUPP; 828 break; 829 830 case PFSmem: 831 vap->va_bytes = vap->va_size = 832 ctob(procp->p_vmspace->vm_tsize + 833 procp->p_vmspace->vm_dsize + 834 procp->p_vmspace->vm_ssize); 835 break; 836 837 #if defined(PT_GETREGS) || defined(PT_SETREGS) 838 case PFSregs: 839 vap->va_bytes = vap->va_size = sizeof(struct reg); 840 break; 841 #endif 842 843 #if defined(PT_GETFPREGS) || defined(PT_SETFPREGS) 844 case PFSfpregs: 845 vap->va_bytes = vap->va_size = sizeof(struct fpreg); 846 break; 847 #endif 848 849 case PFSctl: 850 case PFSstatus: 851 case PFSstat: 852 case PFSnote: 853 case PFSnotepg: 854 case PFScmdline: 855 case PFSmeminfo: 856 case PFSdevices: 857 case PFScpuinfo: 858 case PFSuptime: 859 case PFSmounts: 860 case PFScpustat: 861 case PFSloadavg: 862 case PFSstatm: 863 vap->va_bytes = vap->va_size = 0; 864 break; 865 case PFSmap: 866 case PFSmaps: 867 /* 868 * Advise a larger blocksize for the map files, so that 869 * they may be read in one pass. 870 */ 871 vap->va_blocksize = 4 * PAGE_SIZE; 872 vap->va_bytes = vap->va_size = 0; 873 break; 874 875 case PFScwd: 876 case PFSchroot: 877 case PFSexe: { 878 char *bp; 879 880 vap->va_nlink = 1; 881 vap->va_uid = 0; 882 vap->va_gid = 0; 883 bp = path + MAXPATHLEN; 884 *--bp = '\0'; 885 procfs_dir(pfs->pfs_type, curlwp, procp, &bp, path, 886 MAXPATHLEN); 887 vap->va_bytes = vap->va_size = strlen(bp); 888 break; 889 } 890 891 case PFSemul: 892 vap->va_bytes = vap->va_size = strlen(procp->p_emul->e_name); 893 break; 894 895 #ifdef __HAVE_PROCFS_MACHDEP 896 PROCFS_MACHDEP_NODETYPE_CASES 897 error = procfs_machdep_getattr(ap->a_vp, vap, procp); 898 break; 899 #endif 900 901 default: 902 panic("procfs_getattr"); 903 } 904 905 if (procp != NULL) 906 procfs_proc_unlock(procp); 907 if (path != NULL) 908 free(path, M_TEMP); 909 910 return (error); 911 } 912 913 /*ARGSUSED*/ 914 int 915 procfs_setattr(void *v) 916 { 917 /* 918 * just fake out attribute setting 919 * it's not good to generate an error 920 * return, otherwise things like creat() 921 * will fail when they try to set the 922 * file length to 0. worse, this means 923 * that echo $note > /proc/$pid/note will fail. 924 */ 925 926 return (0); 927 } 928 929 /* 930 * implement access checking. 931 * 932 * actually, the check for super-user is slightly 933 * broken since it will allow read access to write-only 934 * objects. this doesn't cause any particular trouble 935 * but does mean that the i/o entry points need to check 936 * that the operation really does make sense. 937 */ 938 int 939 procfs_access(v) 940 void *v; 941 { 942 struct vop_access_args /* { 943 struct vnode *a_vp; 944 int a_mode; 945 kauth_cred_t a_cred; 946 } */ *ap = v; 947 struct vattr va; 948 int error; 949 950 if ((error = VOP_GETATTR(ap->a_vp, &va, ap->a_cred)) != 0) 951 return (error); 952 953 return (vaccess(va.va_type, va.va_mode, 954 va.va_uid, va.va_gid, ap->a_mode, ap->a_cred)); 955 } 956 957 /* 958 * lookup. this is incredibly complicated in the 959 * general case, however for most pseudo-filesystems 960 * very little needs to be done. 961 * 962 * Locking isn't hard here, just poorly documented. 963 * 964 * If we're looking up ".", just vref the parent & return it. 965 * 966 * If we're looking up "..", unlock the parent, and lock "..". If everything 967 * went ok, and we're on the last component and the caller requested the 968 * parent locked, try to re-lock the parent. We do this to prevent lock 969 * races. 970 * 971 * For anything else, get the needed node. Then unlock the parent if not 972 * the last component or not LOCKPARENT (i.e. if we wouldn't re-lock the 973 * parent in the .. case). 974 * 975 * We try to exit with the parent locked in error cases. 976 */ 977 int 978 procfs_lookup(v) 979 void *v; 980 { 981 struct vop_lookup_args /* { 982 struct vnode * a_dvp; 983 struct vnode ** a_vpp; 984 struct componentname * a_cnp; 985 } */ *ap = v; 986 struct componentname *cnp = ap->a_cnp; 987 struct vnode **vpp = ap->a_vpp; 988 struct vnode *dvp = ap->a_dvp; 989 const char *pname = cnp->cn_nameptr; 990 const struct proc_target *pt = NULL; 991 struct vnode *fvp; 992 pid_t pid, vnpid; 993 struct pfsnode *pfs; 994 struct proc *p = NULL; 995 int i, error; 996 pfstype type; 997 998 *vpp = NULL; 999 1000 if (cnp->cn_nameiop == DELETE || cnp->cn_nameiop == RENAME) 1001 return (EROFS); 1002 1003 if (cnp->cn_namelen == 1 && *pname == '.') { 1004 *vpp = dvp; 1005 VREF(dvp); 1006 return (0); 1007 } 1008 1009 pfs = VTOPFS(dvp); 1010 switch (pfs->pfs_type) { 1011 case PFSroot: 1012 /* 1013 * Shouldn't get here with .. in the root node. 1014 */ 1015 if (cnp->cn_flags & ISDOTDOT) 1016 return (EIO); 1017 1018 for (i = 0; i < nproc_root_targets; i++) { 1019 pt = &proc_root_targets[i]; 1020 /* 1021 * check for node match. proc is always NULL here, 1022 * so call pt_valid with constant NULL lwp. 1023 */ 1024 if (cnp->cn_namelen == pt->pt_namlen && 1025 memcmp(pt->pt_name, pname, cnp->cn_namelen) == 0 && 1026 (pt->pt_valid == NULL || 1027 (*pt->pt_valid)(NULL, dvp->v_mount))) 1028 break; 1029 } 1030 1031 if (i != nproc_root_targets) { 1032 error = procfs_allocvp(dvp->v_mount, vpp, 0, 1033 pt->pt_pfstype, -1, NULL); 1034 return (error); 1035 } 1036 1037 if (CNEQ(cnp, "curproc", 7)) { 1038 pid = curproc->p_pid; 1039 vnpid = 0; 1040 type = PFScurproc; 1041 } else if (CNEQ(cnp, "self", 4)) { 1042 pid = curproc->p_pid; 1043 vnpid = 0; 1044 type = PFSself; 1045 } else { 1046 pid = (pid_t)atoi(pname, cnp->cn_namelen); 1047 vnpid = pid; 1048 type = PFSproc; 1049 } 1050 1051 if (procfs_proc_lock(pid, &p, ESRCH) != 0) 1052 break; 1053 error = procfs_allocvp(dvp->v_mount, vpp, vnpid, type, -1, p); 1054 procfs_proc_unlock(p); 1055 return (error); 1056 1057 case PFSproc: 1058 /* 1059 * do the .. dance. We unlock the directory, and then 1060 * get the root dir. That will automatically return .. 1061 * locked. Then if the caller wanted dvp locked, we 1062 * re-lock. 1063 */ 1064 if (cnp->cn_flags & ISDOTDOT) { 1065 VOP_UNLOCK(dvp, 0); 1066 error = procfs_root(dvp->v_mount, vpp); 1067 vn_lock(dvp, LK_EXCLUSIVE | LK_RETRY); 1068 return (error); 1069 } 1070 1071 if (procfs_proc_lock(pfs->pfs_pid, &p, ESRCH) != 0) 1072 break; 1073 1074 for (pt = proc_targets, i = 0; i < nproc_targets; pt++, i++) { 1075 struct lwp *plwp; 1076 int found; 1077 1078 mutex_enter(&p->p_smutex); 1079 plwp = proc_representative_lwp(p, NULL, 1); 1080 lwp_addref(plwp); 1081 mutex_exit(&p->p_smutex); 1082 found = cnp->cn_namelen == pt->pt_namlen && 1083 memcmp(pt->pt_name, pname, cnp->cn_namelen) == 0 && 1084 (pt->pt_valid == NULL 1085 || (*pt->pt_valid)(plwp, dvp->v_mount)); 1086 lwp_delref(plwp); 1087 if (found) 1088 break; 1089 } 1090 if (i == nproc_targets) { 1091 procfs_proc_unlock(p); 1092 break; 1093 } 1094 if (pt->pt_pfstype == PFSfile) { 1095 fvp = p->p_textvp; 1096 /* We already checked that it exists. */ 1097 VREF(fvp); 1098 procfs_proc_unlock(p); 1099 vn_lock(fvp, LK_EXCLUSIVE | LK_RETRY); 1100 *vpp = fvp; 1101 return (0); 1102 } 1103 1104 error = procfs_allocvp(dvp->v_mount, vpp, pfs->pfs_pid, 1105 pt->pt_pfstype, -1, p); 1106 procfs_proc_unlock(p); 1107 return (error); 1108 1109 case PFSfd: { 1110 int fd; 1111 file_t *fp; 1112 1113 if ((error = procfs_proc_lock(pfs->pfs_pid, &p, ENOENT)) != 0) 1114 return error; 1115 1116 /* 1117 * do the .. dance. We unlock the directory, and then 1118 * get the proc dir. That will automatically return .. 1119 * locked. Then re-lock the directory. 1120 */ 1121 if (cnp->cn_flags & ISDOTDOT) { 1122 VOP_UNLOCK(dvp, 0); 1123 error = procfs_allocvp(dvp->v_mount, vpp, pfs->pfs_pid, 1124 PFSproc, -1, p); 1125 procfs_proc_unlock(p); 1126 vn_lock(dvp, LK_EXCLUSIVE | LK_RETRY); 1127 return (error); 1128 } 1129 fd = atoi(pname, cnp->cn_namelen); 1130 1131 fp = fd_getfile2(p, fd); 1132 if (fp == NULL) { 1133 procfs_proc_unlock(p); 1134 return ENOENT; 1135 } 1136 fvp = fp->f_data; 1137 1138 /* Don't show directories */ 1139 if (fp->f_type == DTYPE_VNODE && fvp->v_type != VDIR) { 1140 VREF(fvp); 1141 closef(fp); 1142 procfs_proc_unlock(p); 1143 vn_lock(fvp, LK_EXCLUSIVE | LK_RETRY | 1144 (p == curproc ? LK_CANRECURSE : 0)); 1145 *vpp = fvp; 1146 return 0; 1147 } 1148 1149 closef(fp); 1150 error = procfs_allocvp(dvp->v_mount, vpp, pfs->pfs_pid, 1151 PFSfd, fd, p); 1152 procfs_proc_unlock(p); 1153 return error; 1154 } 1155 default: 1156 return (ENOTDIR); 1157 } 1158 1159 return (cnp->cn_nameiop == LOOKUP ? ENOENT : EROFS); 1160 } 1161 1162 int 1163 procfs_validfile(struct lwp *l, struct mount *mp) 1164 { 1165 return l != NULL && l->l_proc != NULL && l->l_proc->p_textvp != NULL; 1166 } 1167 1168 static int 1169 procfs_validfile_linux(l, mp) 1170 struct lwp *l; 1171 struct mount *mp; 1172 { 1173 int flags; 1174 1175 flags = VFSTOPROC(mp)->pmnt_flags; 1176 return (flags & PROCFSMNT_LINUXCOMPAT) && 1177 (l == NULL || l->l_proc == NULL || procfs_validfile(l, mp)); 1178 } 1179 1180 struct procfs_root_readdir_ctx { 1181 struct uio *uiop; 1182 off_t *cookies; 1183 int ncookies; 1184 off_t off; 1185 off_t startoff; 1186 int error; 1187 }; 1188 1189 static int 1190 procfs_root_readdir_callback(struct proc *p, void *arg) 1191 { 1192 struct procfs_root_readdir_ctx *ctxp = arg; 1193 struct dirent d; 1194 struct uio *uiop; 1195 int error; 1196 1197 uiop = ctxp->uiop; 1198 if (uiop->uio_resid < UIO_MX) 1199 return -1; /* no space */ 1200 1201 if (ctxp->off < ctxp->startoff) { 1202 ctxp->off++; 1203 return 0; 1204 } 1205 1206 if (kauth_authorize_process(kauth_cred_get(), 1207 KAUTH_PROCESS_CANSEE, p, 1208 KAUTH_ARG(KAUTH_REQ_PROCESS_CANSEE_ENTRY), NULL, NULL) != 0) 1209 return 0; 1210 1211 memset(&d, 0, UIO_MX); 1212 d.d_reclen = UIO_MX; 1213 d.d_fileno = PROCFS_FILENO(p->p_pid, PFSproc, -1); 1214 d.d_namlen = snprintf(d.d_name, 1215 UIO_MX - offsetof(struct dirent, d_name), "%ld", (long)p->p_pid); 1216 d.d_type = DT_DIR; 1217 1218 mutex_exit(&proclist_lock); 1219 error = uiomove(&d, UIO_MX, uiop); 1220 mutex_enter(&proclist_lock); 1221 if (error) { 1222 ctxp->error = error; 1223 return -1; 1224 } 1225 1226 ctxp->ncookies++; 1227 if (ctxp->cookies) 1228 *(ctxp->cookies)++ = ctxp->off + 1; 1229 ctxp->off++; 1230 1231 return 0; 1232 } 1233 1234 /* 1235 * readdir returns directory entries from pfsnode (vp). 1236 * 1237 * the strategy here with procfs is to generate a single 1238 * directory entry at a time (struct dirent) and then 1239 * copy that out to userland using uiomove. a more efficent 1240 * though more complex implementation, would try to minimize 1241 * the number of calls to uiomove(). for procfs, this is 1242 * hardly worth the added code complexity. 1243 * 1244 * this should just be done through read() 1245 */ 1246 int 1247 procfs_readdir(v) 1248 void *v; 1249 { 1250 struct vop_readdir_args /* { 1251 struct vnode *a_vp; 1252 struct uio *a_uio; 1253 kauth_cred_t a_cred; 1254 int *a_eofflag; 1255 off_t **a_cookies; 1256 int *a_ncookies; 1257 } */ *ap = v; 1258 struct uio *uio = ap->a_uio; 1259 struct dirent d; 1260 struct pfsnode *pfs; 1261 off_t i; 1262 int error; 1263 off_t *cookies = NULL; 1264 int ncookies; 1265 struct vnode *vp; 1266 const struct proc_target *pt; 1267 struct procfs_root_readdir_ctx ctx; 1268 struct lwp *l; 1269 int nfd; 1270 1271 vp = ap->a_vp; 1272 pfs = VTOPFS(vp); 1273 1274 if (uio->uio_resid < UIO_MX) 1275 return (EINVAL); 1276 if (uio->uio_offset < 0) 1277 return (EINVAL); 1278 1279 error = 0; 1280 i = uio->uio_offset; 1281 memset(&d, 0, UIO_MX); 1282 d.d_reclen = UIO_MX; 1283 ncookies = uio->uio_resid / UIO_MX; 1284 1285 switch (pfs->pfs_type) { 1286 /* 1287 * this is for the process-specific sub-directories. 1288 * all that is needed to is copy out all the entries 1289 * from the procent[] table (top of this file). 1290 */ 1291 case PFSproc: { 1292 struct proc *p; 1293 1294 if (i >= nproc_targets) 1295 return 0; 1296 1297 if (procfs_proc_lock(pfs->pfs_pid, &p, ESRCH) != 0) 1298 break; 1299 1300 if (ap->a_ncookies) { 1301 ncookies = min(ncookies, (nproc_targets - i)); 1302 cookies = malloc(ncookies * sizeof (off_t), 1303 M_TEMP, M_WAITOK); 1304 *ap->a_cookies = cookies; 1305 } 1306 1307 for (pt = &proc_targets[i]; 1308 uio->uio_resid >= UIO_MX && i < nproc_targets; pt++, i++) { 1309 if (pt->pt_valid) { 1310 /* XXX LWP can disappear */ 1311 mutex_enter(&p->p_smutex); 1312 l = proc_representative_lwp(p, NULL, 1); 1313 mutex_exit(&p->p_smutex); 1314 if ((*pt->pt_valid)(l, vp->v_mount) == 0) 1315 continue; 1316 } 1317 1318 d.d_fileno = PROCFS_FILENO(pfs->pfs_pid, 1319 pt->pt_pfstype, -1); 1320 d.d_namlen = pt->pt_namlen; 1321 memcpy(d.d_name, pt->pt_name, pt->pt_namlen + 1); 1322 d.d_type = pt->pt_type; 1323 1324 if ((error = uiomove(&d, UIO_MX, uio)) != 0) 1325 break; 1326 if (cookies) 1327 *cookies++ = i + 1; 1328 } 1329 1330 procfs_proc_unlock(p); 1331 break; 1332 } 1333 case PFSfd: { 1334 struct proc *p; 1335 file_t *fp; 1336 int lim, nc = 0; 1337 1338 if ((error = procfs_proc_lock(pfs->pfs_pid, &p, ESRCH)) != 0) 1339 return error; 1340 1341 /* XXX Should this be by file as well? */ 1342 if (kauth_authorize_process(kauth_cred_get(), 1343 KAUTH_PROCESS_CANSEE, p, 1344 KAUTH_ARG(KAUTH_REQ_PROCESS_CANSEE_OPENFILES), NULL, 1345 NULL) != 0) { 1346 procfs_proc_unlock(p); 1347 return ESRCH; 1348 } 1349 1350 nfd = p->p_fd->fd_nfiles; 1351 1352 lim = min((int)p->p_rlimit[RLIMIT_NOFILE].rlim_cur, maxfiles); 1353 if (i >= lim) { 1354 procfs_proc_unlock(p); 1355 return 0; 1356 } 1357 1358 if (ap->a_ncookies) { 1359 ncookies = min(ncookies, (nfd + 2 - i)); 1360 cookies = malloc(ncookies * sizeof (off_t), 1361 M_TEMP, M_WAITOK); 1362 *ap->a_cookies = cookies; 1363 } 1364 1365 for (; i < 2 && uio->uio_resid >= UIO_MX; i++) { 1366 pt = &proc_targets[i]; 1367 d.d_namlen = pt->pt_namlen; 1368 d.d_fileno = PROCFS_FILENO(pfs->pfs_pid, 1369 pt->pt_pfstype, -1); 1370 (void)memcpy(d.d_name, pt->pt_name, pt->pt_namlen + 1); 1371 d.d_type = pt->pt_type; 1372 if ((error = uiomove(&d, UIO_MX, uio)) != 0) 1373 break; 1374 if (cookies) 1375 *cookies++ = i + 1; 1376 nc++; 1377 } 1378 if (error) { 1379 ncookies = nc; 1380 break; 1381 } 1382 for (; uio->uio_resid >= UIO_MX && i < nfd; i++) { 1383 /* check the descriptor exists */ 1384 if ((fp = fd_getfile2(p, i - 2)) == NULL) 1385 continue; 1386 closef(fp); 1387 1388 d.d_fileno = PROCFS_FILENO(pfs->pfs_pid, PFSfd, i - 2); 1389 d.d_namlen = snprintf(d.d_name, sizeof(d.d_name), 1390 "%lld", (long long)(i - 2)); 1391 d.d_type = VREG; 1392 if ((error = uiomove(&d, UIO_MX, uio)) != 0) 1393 break; 1394 if (cookies) 1395 *cookies++ = i + 1; 1396 nc++; 1397 } 1398 ncookies = nc; 1399 procfs_proc_unlock(p); 1400 break; 1401 } 1402 1403 /* 1404 * this is for the root of the procfs filesystem 1405 * what is needed are special entries for "curproc" 1406 * and "self" followed by an entry for each process 1407 * on allproc. 1408 */ 1409 1410 case PFSroot: { 1411 int nc = 0; 1412 1413 if (ap->a_ncookies) { 1414 /* 1415 * XXX Potentially allocating too much space here, 1416 * but I'm lazy. This loop needs some work. 1417 */ 1418 cookies = malloc(ncookies * sizeof (off_t), 1419 M_TEMP, M_WAITOK); 1420 *ap->a_cookies = cookies; 1421 } 1422 error = 0; 1423 /* 0 ... 3 are static entries. */ 1424 for (; i <= 3 && uio->uio_resid >= UIO_MX; i++) { 1425 switch (i) { 1426 case 0: /* `.' */ 1427 case 1: /* `..' */ 1428 d.d_fileno = PROCFS_FILENO(0, PFSroot, -1); 1429 d.d_namlen = i + 1; 1430 memcpy(d.d_name, "..", d.d_namlen); 1431 d.d_name[i + 1] = '\0'; 1432 d.d_type = DT_DIR; 1433 break; 1434 1435 case 2: 1436 d.d_fileno = PROCFS_FILENO(0, PFScurproc, -1); 1437 d.d_namlen = sizeof("curproc") - 1; 1438 memcpy(d.d_name, "curproc", sizeof("curproc")); 1439 d.d_type = DT_LNK; 1440 break; 1441 1442 case 3: 1443 d.d_fileno = PROCFS_FILENO(0, PFSself, -1); 1444 d.d_namlen = sizeof("self") - 1; 1445 memcpy(d.d_name, "self", sizeof("self")); 1446 d.d_type = DT_LNK; 1447 break; 1448 } 1449 1450 if ((error = uiomove(&d, UIO_MX, uio)) != 0) 1451 break; 1452 nc++; 1453 if (cookies) 1454 *cookies++ = i + 1; 1455 } 1456 /* 4 ... are process entries. */ 1457 ctx.uiop = uio; 1458 ctx.error = 0; 1459 ctx.off = 4; 1460 ctx.startoff = i; 1461 ctx.cookies = cookies; 1462 ctx.ncookies = nc; 1463 proclist_foreach_call(&allproc, 1464 procfs_root_readdir_callback, &ctx); 1465 cookies = ctx.cookies; 1466 nc = ctx.ncookies; 1467 error = ctx.error; 1468 if (error) 1469 break; 1470 1471 /* misc entries. */ 1472 if (i < ctx.off) 1473 i = ctx.off; 1474 if (i >= ctx.off + nproc_root_targets) 1475 break; 1476 for (pt = &proc_root_targets[i - ctx.off]; 1477 uio->uio_resid >= UIO_MX && 1478 pt < &proc_root_targets[nproc_root_targets]; 1479 pt++, i++) { 1480 if (pt->pt_valid && 1481 (*pt->pt_valid)(NULL, vp->v_mount) == 0) 1482 continue; 1483 d.d_fileno = PROCFS_FILENO(0, pt->pt_pfstype, -1); 1484 d.d_namlen = pt->pt_namlen; 1485 memcpy(d.d_name, pt->pt_name, pt->pt_namlen + 1); 1486 d.d_type = pt->pt_type; 1487 1488 if ((error = uiomove(&d, UIO_MX, uio)) != 0) 1489 break; 1490 nc++; 1491 if (cookies) 1492 *cookies++ = i + 1; 1493 } 1494 1495 ncookies = nc; 1496 break; 1497 } 1498 1499 default: 1500 error = ENOTDIR; 1501 break; 1502 } 1503 1504 if (ap->a_ncookies) { 1505 if (error) { 1506 if (cookies) 1507 free(*ap->a_cookies, M_TEMP); 1508 *ap->a_ncookies = 0; 1509 *ap->a_cookies = NULL; 1510 } else 1511 *ap->a_ncookies = ncookies; 1512 } 1513 uio->uio_offset = i; 1514 return (error); 1515 } 1516 1517 /* 1518 * readlink reads the link of `curproc' and others 1519 */ 1520 int 1521 procfs_readlink(v) 1522 void *v; 1523 { 1524 struct vop_readlink_args *ap = v; 1525 char bf[16]; /* should be enough */ 1526 char *bp = bf; 1527 char *path = NULL; 1528 int len = 0; 1529 int error = 0; 1530 struct pfsnode *pfs = VTOPFS(ap->a_vp); 1531 struct proc *pown; 1532 1533 if (pfs->pfs_fileno == PROCFS_FILENO(0, PFScurproc, -1)) 1534 len = snprintf(bf, sizeof(bf), "%ld", (long)curproc->p_pid); 1535 else if (pfs->pfs_fileno == PROCFS_FILENO(0, PFSself, -1)) 1536 len = snprintf(bf, sizeof(bf), "%s", "curproc"); 1537 else if (pfs->pfs_fileno == PROCFS_FILENO(pfs->pfs_pid, PFScwd, -1) || 1538 pfs->pfs_fileno == PROCFS_FILENO(pfs->pfs_pid, PFSchroot, -1) || 1539 pfs->pfs_fileno == PROCFS_FILENO(pfs->pfs_pid, PFSexe, -1)) { 1540 if ((error = procfs_proc_lock(pfs->pfs_pid, &pown, ESRCH)) != 0) 1541 return error; 1542 MALLOC(path, char *, MAXPATHLEN + 4, M_TEMP, 1543 M_WAITOK|M_CANFAIL); 1544 if (path == NULL) { 1545 procfs_proc_unlock(pown); 1546 return (ENOMEM); 1547 } 1548 bp = path + MAXPATHLEN; 1549 *--bp = '\0'; 1550 procfs_dir(PROCFS_TYPE(pfs->pfs_fileno), curlwp, pown, 1551 &bp, path, MAXPATHLEN); 1552 procfs_proc_unlock(pown); 1553 len = strlen(bp); 1554 } else { 1555 file_t *fp; 1556 struct vnode *vxp, *vp; 1557 1558 if ((error = procfs_proc_lock(pfs->pfs_pid, &pown, ESRCH)) != 0) 1559 return error; 1560 1561 fp = fd_getfile2(pown, pfs->pfs_fd); 1562 if (fp == NULL) { 1563 procfs_proc_unlock(pown); 1564 return EBADF; 1565 } 1566 1567 switch (fp->f_type) { 1568 case DTYPE_VNODE: 1569 vxp = (struct vnode *)fp->f_data; 1570 if (vxp->v_type != VDIR) { 1571 error = EINVAL; 1572 break; 1573 } 1574 if ((path = malloc(MAXPATHLEN, M_TEMP, M_WAITOK)) 1575 == NULL) { 1576 error = ENOMEM; 1577 break; 1578 } 1579 bp = path + MAXPATHLEN; 1580 *--bp = '\0'; 1581 1582 /* 1583 * XXX: kludge to avoid locking against ourselves 1584 * in getcwd() 1585 */ 1586 if (vxp->v_tag == VT_PROCFS) { 1587 *--bp = '/'; 1588 } else { 1589 rw_enter(&curproc->p_cwdi->cwdi_lock, RW_READER); 1590 vp = curproc->p_cwdi->cwdi_rdir; 1591 if (vp == NULL) 1592 vp = rootvnode; 1593 error = getcwd_common(vxp, vp, &bp, path, 1594 MAXPATHLEN / 2, 0, curlwp); 1595 rw_exit(&curproc->p_cwdi->cwdi_lock); 1596 } 1597 if (error) 1598 break; 1599 len = strlen(bp); 1600 break; 1601 1602 case DTYPE_MISC: 1603 len = snprintf(bf, sizeof(bf), "%s", "[misc]"); 1604 break; 1605 1606 case DTYPE_KQUEUE: 1607 len = snprintf(bf, sizeof(bf), "%s", "[kqueue]"); 1608 break; 1609 1610 default: 1611 error = EINVAL; 1612 break; 1613 } 1614 closef(fp); 1615 procfs_proc_unlock(pown); 1616 } 1617 1618 if (error == 0) 1619 error = uiomove(bp, len, ap->a_uio); 1620 if (path) 1621 free(path, M_TEMP); 1622 return error; 1623 } 1624 1625 /* 1626 * convert decimal ascii to int 1627 */ 1628 static int 1629 atoi(b, len) 1630 const char *b; 1631 size_t len; 1632 { 1633 int p = 0; 1634 1635 while (len--) { 1636 char c = *b++; 1637 if (c < '0' || c > '9') 1638 return -1; 1639 p = 10 * p + (c - '0'); 1640 } 1641 1642 return p; 1643 } 1644