xref: /netbsd-src/sys/miscfs/procfs/procfs_subr.c (revision 62a8debe1dc62962e18a1c918def78666141273b)
1 /*	$NetBSD: procfs_subr.c,v 1.95 2009/03/15 17:22:38 cegger 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  *
19  * THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS
20  * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED
21  * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
22  * PURPOSE ARE DISCLAIMED.  IN NO EVENT SHALL THE FOUNDATION OR CONTRIBUTORS
23  * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
24  * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
25  * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
26  * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
27  * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
28  * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
29  * POSSIBILITY OF SUCH DAMAGE.
30  */
31 
32 /*
33  * Copyright (c) 1993
34  *	The Regents of the University of California.  All rights reserved.
35  *
36  * This code is derived from software contributed to Berkeley by
37  * Jan-Simon Pendry.
38  *
39  * Redistribution and use in source and binary forms, with or without
40  * modification, are permitted provided that the following conditions
41  * are met:
42  * 1. Redistributions of source code must retain the above copyright
43  *    notice, this list of conditions and the following disclaimer.
44  * 2. Redistributions in binary form must reproduce the above copyright
45  *    notice, this list of conditions and the following disclaimer in the
46  *    documentation and/or other materials provided with the distribution.
47  * 3. Neither the name of the University nor the names of its contributors
48  *    may be used to endorse or promote products derived from this software
49  *    without specific prior written permission.
50  *
51  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
52  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
53  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
54  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
55  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
56  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
57  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
58  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
59  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
60  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
61  * SUCH DAMAGE.
62  *
63  *	@(#)procfs_subr.c	8.6 (Berkeley) 5/14/95
64  */
65 
66 /*
67  * Copyright (c) 1994 Christopher G. Demetriou.  All rights reserved.
68  * Copyright (c) 1993 Jan-Simon Pendry
69  *
70  * This code is derived from software contributed to Berkeley by
71  * Jan-Simon Pendry.
72  *
73  * Redistribution and use in source and binary forms, with or without
74  * modification, are permitted provided that the following conditions
75  * are met:
76  * 1. Redistributions of source code must retain the above copyright
77  *    notice, this list of conditions and the following disclaimer.
78  * 2. Redistributions in binary form must reproduce the above copyright
79  *    notice, this list of conditions and the following disclaimer in the
80  *    documentation and/or other materials provided with the distribution.
81  * 3. All advertising materials mentioning features or use of this software
82  *    must display the following acknowledgement:
83  *	This product includes software developed by the University of
84  *	California, Berkeley and its contributors.
85  * 4. Neither the name of the University nor the names of its contributors
86  *    may be used to endorse or promote products derived from this software
87  *    without specific prior written permission.
88  *
89  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
90  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
91  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
92  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
93  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
94  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
95  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
96  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
97  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
98  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
99  * SUCH DAMAGE.
100  *
101  *	@(#)procfs_subr.c	8.6 (Berkeley) 5/14/95
102  */
103 
104 #include <sys/cdefs.h>
105 __KERNEL_RCSID(0, "$NetBSD: procfs_subr.c,v 1.95 2009/03/15 17:22:38 cegger Exp $");
106 
107 #include <sys/param.h>
108 #include <sys/systm.h>
109 #include <sys/time.h>
110 #include <sys/kernel.h>
111 #include <sys/proc.h>
112 #include <sys/vnode.h>
113 #include <sys/malloc.h>
114 #include <sys/stat.h>
115 #include <sys/file.h>
116 #include <sys/filedesc.h>
117 #include <sys/kauth.h>
118 
119 #include <miscfs/procfs/procfs.h>
120 
121 void procfs_hashins(struct pfsnode *);
122 void procfs_hashrem(struct pfsnode *);
123 struct vnode *procfs_hashget(pid_t, pfstype, int, struct mount *, int);
124 
125 LIST_HEAD(pfs_hashhead, pfsnode) *pfs_hashtbl;
126 u_long	pfs_ihash;	/* size of hash table - 1 */
127 #define PFSPIDHASH(pid)	((pid) & pfs_ihash)
128 
129 kmutex_t pfs_hashlock;
130 kmutex_t pfs_ihash_lock;
131 
132 #define	ISSET(t, f)	((t) & (f))
133 
134 /*
135  * allocate a pfsnode/vnode pair.  the vnode is
136  * referenced, and locked.
137  *
138  * the pid, pfs_type, and mount point uniquely
139  * identify a pfsnode.  the mount point is needed
140  * because someone might mount this filesystem
141  * twice.
142  *
143  * all pfsnodes are maintained on a singly-linked
144  * list.  new nodes are only allocated when they cannot
145  * be found on this list.  entries on the list are
146  * removed when the vfs reclaim entry is called.
147  *
148  * a single lock is kept for the entire list.  this is
149  * needed because the getnewvnode() function can block
150  * waiting for a vnode to become free, in which case there
151  * may be more than one process trying to get the same
152  * vnode.  this lock is only taken if we are going to
153  * call getnewvnode, since the kernel itself is single-threaded.
154  *
155  * if an entry is found on the list, then call vget() to
156  * take a reference.  this is done because there may be
157  * zero references to it and so it needs to removed from
158  * the vnode free list.
159  */
160 int
161 procfs_allocvp(struct mount *mp, struct vnode **vpp, pid_t pid, pfstype pfs_type, int fd, struct proc *p)
162 {
163 	struct pfsnode *pfs;
164 	struct vnode *vp;
165 	int error;
166 
167  retry:
168 	*vpp = procfs_hashget(pid, pfs_type, fd, mp, LK_EXCLUSIVE);
169 	if (*vpp != NULL)
170 		return (0);
171 
172 	if ((error = getnewvnode(VT_PROCFS, mp, procfs_vnodeop_p, &vp)) != 0) {
173 		*vpp = NULL;
174 		return (error);
175 	}
176 	pfs = malloc(sizeof(struct pfsnode), M_TEMP, M_WAITOK);
177 
178 	mutex_enter(&pfs_hashlock);
179 	if ((*vpp = procfs_hashget(pid, pfs_type, fd, mp, 0)) != NULL) {
180 		mutex_exit(&pfs_hashlock);
181 		ungetnewvnode(vp);
182 		free(pfs, M_TEMP);
183 		goto retry;
184 	}
185 
186 	vp->v_data = pfs;
187 	pfs->pfs_pid = pid;
188 	pfs->pfs_type = pfs_type;
189 	pfs->pfs_vnode = vp;
190 	pfs->pfs_flags = 0;
191 	pfs->pfs_fileno = PROCFS_FILENO(pid, pfs_type, fd);
192 	pfs->pfs_fd = fd;
193 
194 	switch (pfs_type) {
195 	case PFSroot:	/* /proc = dr-xr-xr-x */
196 		vp->v_vflag |= VV_ROOT;
197 		/*FALLTHROUGH*/
198 	case PFSproc:	/* /proc/N = dr-xr-xr-x */
199 		pfs->pfs_mode = S_IRUSR|S_IXUSR|S_IRGRP|S_IXGRP|S_IROTH|S_IXOTH;
200 		vp->v_type = VDIR;
201 		break;
202 
203 	case PFScurproc:	/* /proc/curproc = lr-xr-xr-x */
204 	case PFSself:	/* /proc/self    = lr-xr-xr-x */
205 	case PFScwd:	/* /proc/N/cwd = lr-xr-xr-x */
206 	case PFSchroot:	/* /proc/N/chroot = lr-xr-xr-x */
207 	case PFSexe:	/* /proc/N/exe = lr-xr-xr-x */
208 		pfs->pfs_mode = S_IRUSR|S_IXUSR|S_IRGRP|S_IXGRP|S_IROTH|S_IXOTH;
209 		vp->v_type = VLNK;
210 		break;
211 
212 	case PFSfd:
213 		if (fd == -1) {	/* /proc/N/fd = dr-xr-xr-x */
214 			pfs->pfs_mode = S_IRUSR|S_IXUSR;
215 			vp->v_type = VDIR;
216 		} else {	/* /proc/N/fd/M = [ps-]rw------- */
217 			file_t *fp;
218 			vnode_t *vxp;
219 
220 			if ((fp = fd_getfile2(p, pfs->pfs_fd)) == NULL) {
221 				error = EBADF;
222 				goto bad;
223 			}
224 
225 			pfs->pfs_mode = S_IRUSR|S_IWUSR;
226 			switch (fp->f_type) {
227 			case DTYPE_VNODE:
228 				vxp = fp->f_data;
229 
230 				/*
231 				 * We make symlinks for directories
232 				 * to avoid cycles.
233 				 */
234 				if (vxp->v_type == VDIR)
235 					goto symlink;
236 				vp->v_type = vxp->v_type;
237 				break;
238 			case DTYPE_PIPE:
239 				vp->v_type = VFIFO;
240 				break;
241 			case DTYPE_SOCKET:
242 				vp->v_type = VSOCK;
243 				break;
244 			case DTYPE_KQUEUE:
245 			case DTYPE_MISC:
246 			symlink:
247 				pfs->pfs_mode = S_IRUSR|S_IXUSR|S_IRGRP|
248 				    S_IXGRP|S_IROTH|S_IXOTH;
249 				vp->v_type = VLNK;
250 				break;
251 			default:
252 				error = EOPNOTSUPP;
253 				closef(fp);
254 				goto bad;
255 			}
256 			closef(fp);
257 		}
258 		break;
259 
260 	case PFSfile:	/* /proc/N/file = -rw------- */
261 	case PFSmem:	/* /proc/N/mem = -rw------- */
262 	case PFSregs:	/* /proc/N/regs = -rw------- */
263 	case PFSfpregs:	/* /proc/N/fpregs = -rw------- */
264 		pfs->pfs_mode = S_IRUSR|S_IWUSR;
265 		vp->v_type = VREG;
266 		break;
267 
268 	case PFSctl:	/* /proc/N/ctl = --w------ */
269 	case PFSnote:	/* /proc/N/note = --w------ */
270 	case PFSnotepg:	/* /proc/N/notepg = --w------ */
271 		pfs->pfs_mode = S_IWUSR;
272 		vp->v_type = VREG;
273 		break;
274 
275 	case PFSmap:	/* /proc/N/map = -r--r--r-- */
276 	case PFSmaps:	/* /proc/N/maps = -r--r--r-- */
277 	case PFSstatus:	/* /proc/N/status = -r--r--r-- */
278 	case PFSstat:	/* /proc/N/stat = -r--r--r-- */
279 	case PFScmdline:	/* /proc/N/cmdline = -r--r--r-- */
280 	case PFSemul:	/* /proc/N/emul = -r--r--r-- */
281 	case PFSmeminfo:	/* /proc/meminfo = -r--r--r-- */
282 	case PFScpustat:	/* /proc/stat = -r--r--r-- */
283 	case PFSdevices:	/* /proc/devices = -r--r--r-- */
284 	case PFScpuinfo:	/* /proc/cpuinfo = -r--r--r-- */
285 	case PFSuptime:	/* /proc/uptime = -r--r--r-- */
286 	case PFSmounts:	/* /proc/mounts = -r--r--r-- */
287 	case PFSloadavg:	/* /proc/loadavg = -r--r--r-- */
288 	case PFSstatm:	/* /proc/N/statm = -r--r--r-- */
289 		pfs->pfs_mode = S_IRUSR|S_IRGRP|S_IROTH;
290 		vp->v_type = VREG;
291 		break;
292 
293 #ifdef __HAVE_PROCFS_MACHDEP
294 	PROCFS_MACHDEP_NODETYPE_CASES
295 		procfs_machdep_allocvp(vp);
296 		break;
297 #endif
298 
299 	default:
300 		panic("procfs_allocvp");
301 	}
302 
303 	procfs_hashins(pfs);
304 	uvm_vnp_setsize(vp, 0);
305 	mutex_exit(&pfs_hashlock);
306 
307 	*vpp = vp;
308 	return (0);
309 
310  bad:
311 	mutex_exit(&pfs_hashlock);
312 	free(pfs, M_TEMP);
313 	vp->v_data = NULL;
314 	ungetnewvnode(vp);
315 	return (error);
316 }
317 
318 int
319 procfs_freevp(struct vnode *vp)
320 {
321 	struct pfsnode *pfs = VTOPFS(vp);
322 
323 	procfs_hashrem(pfs);
324 
325 	free(vp->v_data, M_TEMP);
326 	vp->v_data = NULL;
327 	return (0);
328 }
329 
330 int
331 procfs_rw(void *v)
332 {
333 	struct vop_read_args *ap = v;
334 	struct vnode *vp = ap->a_vp;
335 	struct uio *uio = ap->a_uio;
336 	struct lwp *curl;
337 	struct lwp *l;
338 	struct pfsnode *pfs = VTOPFS(vp);
339 	struct proc *p;
340 	int error;
341 
342 	if (uio->uio_offset < 0)
343 		return EINVAL;
344 
345 	if ((error = procfs_proc_lock(pfs->pfs_pid, &p, ESRCH)) != 0)
346 		return error;
347 
348 	curl = curlwp;
349 
350 	/*
351 	 * Do not allow init to be modified while in secure mode; it
352 	 * could be duped into changing the security level.
353 	 */
354 #define	M2K(m)	((m) == UIO_READ ? KAUTH_REQ_PROCESS_PROCFS_READ : \
355 		 KAUTH_REQ_PROCESS_PROCFS_WRITE)
356 	mutex_enter(p->p_lock);
357 	error = kauth_authorize_process(curl->l_cred, KAUTH_PROCESS_PROCFS,
358 	    p, pfs, KAUTH_ARG(M2K(uio->uio_rw)), NULL);
359 	mutex_exit(p->p_lock);
360 	if (error) {
361 		procfs_proc_unlock(p);
362 		return (error);
363 	}
364 #undef	M2K
365 
366 	mutex_enter(p->p_lock);
367 	LIST_FOREACH(l, &p->p_lwps, l_sibling) {
368 		if (l->l_stat != LSZOMB)
369 			break;
370 	}
371 	/* Process is exiting if no-LWPS or all LWPs are LSZOMB */
372 	if (l == NULL) {
373 		mutex_exit(p->p_lock);
374 		procfs_proc_unlock(p);
375 		return ESRCH;
376 	}
377 
378 	lwp_addref(l);
379 	mutex_exit(p->p_lock);
380 
381 	switch (pfs->pfs_type) {
382 	case PFSnote:
383 	case PFSnotepg:
384 		error = procfs_donote(curl, p, pfs, uio);
385 		break;
386 
387 	case PFSregs:
388 		error = procfs_doregs(curl, l, pfs, uio);
389 		break;
390 
391 	case PFSfpregs:
392 		error = procfs_dofpregs(curl, l, pfs, uio);
393 		break;
394 
395 	case PFSctl:
396 		error = procfs_doctl(curl, l, pfs, uio);
397 		break;
398 
399 	case PFSstatus:
400 		error = procfs_dostatus(curl, l, pfs, uio);
401 		break;
402 
403 	case PFSstat:
404 		error = procfs_do_pid_stat(curl, l, pfs, uio);
405 		break;
406 
407 	case PFSmap:
408 		error = procfs_domap(curl, p, pfs, uio, 0);
409 		break;
410 
411 	case PFSmaps:
412 		error = procfs_domap(curl, p, pfs, uio, 1);
413 		break;
414 
415 	case PFSmem:
416 		error = procfs_domem(curl, l, pfs, uio);
417 		break;
418 
419 	case PFScmdline:
420 		error = procfs_docmdline(curl, p, pfs, uio);
421 		break;
422 
423 	case PFSmeminfo:
424 		error = procfs_domeminfo(curl, p, pfs, uio);
425 		break;
426 
427 	case PFSdevices:
428 		error = procfs_dodevices(curl, p, pfs, uio);
429 		break;
430 
431 	case PFScpuinfo:
432 		error = procfs_docpuinfo(curl, p, pfs, uio);
433 		break;
434 
435 	case PFScpustat:
436 		error = procfs_docpustat(curl, p, pfs, uio);
437 		break;
438 
439 	case PFSloadavg:
440 		error = procfs_doloadavg(curl, p, pfs, uio);
441 		break;
442 
443 	case PFSstatm:
444 		error = procfs_do_pid_statm(curl, l, pfs, uio);
445 		break;
446 
447 	case PFSfd:
448 		error = procfs_dofd(curl, p, pfs, uio);
449 		break;
450 
451 	case PFSuptime:
452 		error = procfs_douptime(curl, p, pfs, uio);
453 		break;
454 
455 	case PFSmounts:
456 		error = procfs_domounts(curl, p, pfs, uio);
457 		break;
458 
459 	case PFSemul:
460 		error = procfs_doemul(curl, p, pfs, uio);
461 		break;
462 
463 #ifdef __HAVE_PROCFS_MACHDEP
464 	PROCFS_MACHDEP_NODETYPE_CASES
465 		error = procfs_machdep_rw(curl, l, pfs, uio);
466 		break;
467 #endif
468 
469 	default:
470 		error = EOPNOTSUPP;
471 		break;
472 	}
473 
474 	/*
475 	 * Release the references that we acquired earlier.
476 	 */
477 	lwp_delref(l);
478 	procfs_proc_unlock(p);
479 
480 	return (error);
481 }
482 
483 /*
484  * Get a string from userland into (bf).  Strip a trailing
485  * nl character (to allow easy access from the shell).
486  * The buffer should be *buflenp + 1 chars long.  vfs_getuserstr
487  * will automatically add a nul char at the end.
488  *
489  * Returns 0 on success or the following errors
490  *
491  * EINVAL:    file offset is non-zero.
492  * EMSGSIZE:  message is longer than kernel buffer
493  * EFAULT:    user i/o buffer is not addressable
494  */
495 int
496 vfs_getuserstr(struct uio *uio, char *bf, int *buflenp)
497 {
498 	int xlen;
499 	int error;
500 
501 	if (uio->uio_offset != 0)
502 		return (EINVAL);
503 
504 	xlen = *buflenp;
505 
506 	/* must be able to read the whole string in one go */
507 	if (xlen < uio->uio_resid)
508 		return (EMSGSIZE);
509 	xlen = uio->uio_resid;
510 
511 	if ((error = uiomove(bf, xlen, uio)) != 0)
512 		return (error);
513 
514 	/* allow multiple writes without seeks */
515 	uio->uio_offset = 0;
516 
517 	/* cleanup string and remove trailing newline */
518 	bf[xlen] = '\0';
519 	xlen = strlen(bf);
520 	if (xlen > 0 && bf[xlen-1] == '\n')
521 		bf[--xlen] = '\0';
522 	*buflenp = xlen;
523 
524 	return (0);
525 }
526 
527 const vfs_namemap_t *
528 vfs_findname(const vfs_namemap_t *nm, const char *bf, int buflen)
529 {
530 
531 	for (; nm->nm_name; nm++)
532 		if (memcmp(bf, nm->nm_name, buflen+1) == 0)
533 			return (nm);
534 
535 	return (0);
536 }
537 
538 /*
539  * Initialize pfsnode hash table.
540  */
541 void
542 procfs_hashinit(void)
543 {
544 	mutex_init(&pfs_hashlock, MUTEX_DEFAULT, IPL_NONE);
545 	mutex_init(&pfs_ihash_lock, MUTEX_DEFAULT, IPL_NONE);
546 	pfs_hashtbl = hashinit(desiredvnodes / 4, HASH_LIST, true, &pfs_ihash);
547 }
548 
549 void
550 procfs_hashreinit(void)
551 {
552 	struct pfsnode *pp;
553 	struct pfs_hashhead *oldhash, *hash;
554 	u_long i, oldmask, mask, val;
555 
556 	hash = hashinit(desiredvnodes / 4, HASH_LIST, true, &mask);
557 
558 	mutex_enter(&pfs_ihash_lock);
559 	oldhash = pfs_hashtbl;
560 	oldmask = pfs_ihash;
561 	pfs_hashtbl = hash;
562 	pfs_ihash = mask;
563 	for (i = 0; i <= oldmask; i++) {
564 		while ((pp = LIST_FIRST(&oldhash[i])) != NULL) {
565 			LIST_REMOVE(pp, pfs_hash);
566 			val = PFSPIDHASH(pp->pfs_pid);
567 			LIST_INSERT_HEAD(&hash[val], pp, pfs_hash);
568 		}
569 	}
570 	mutex_exit(&pfs_ihash_lock);
571 	hashdone(oldhash, HASH_LIST, oldmask);
572 }
573 
574 /*
575  * Free pfsnode hash table.
576  */
577 void
578 procfs_hashdone(void)
579 {
580 	hashdone(pfs_hashtbl, HASH_LIST, pfs_ihash);
581 	mutex_destroy(&pfs_hashlock);
582 	mutex_destroy(&pfs_ihash_lock);
583 }
584 
585 struct vnode *
586 procfs_hashget(pid_t pid, pfstype type, int fd, struct mount *mp, int flags)
587 {
588 	struct pfs_hashhead *ppp;
589 	struct pfsnode *pp;
590 	struct vnode *vp;
591 
592 loop:
593 	mutex_enter(&pfs_ihash_lock);
594 	ppp = &pfs_hashtbl[PFSPIDHASH(pid)];
595 	LIST_FOREACH(pp, ppp, pfs_hash) {
596 		vp = PFSTOV(pp);
597 		if (pid == pp->pfs_pid && pp->pfs_type == type &&
598 		    pp->pfs_fd == fd && vp->v_mount == mp) {
599 		    	if (flags == 0) {
600 				mutex_exit(&pfs_ihash_lock);
601 			} else {
602 				mutex_enter(&vp->v_interlock);
603 				mutex_exit(&pfs_ihash_lock);
604 				if (vget(vp, flags | LK_INTERLOCK))
605 					goto loop;
606 			}
607 			return (vp);
608 		}
609 	}
610 	mutex_exit(&pfs_ihash_lock);
611 	return (NULL);
612 }
613 
614 /*
615  * Insert the pfsnode into the hash table and lock it.
616  */
617 void
618 procfs_hashins(struct pfsnode *pp)
619 {
620 	struct pfs_hashhead *ppp;
621 
622 	/* lock the pfsnode, then put it on the appropriate hash list */
623 	vlockmgr(&pp->pfs_vnode->v_lock, LK_EXCLUSIVE);
624 
625 	mutex_enter(&pfs_ihash_lock);
626 	ppp = &pfs_hashtbl[PFSPIDHASH(pp->pfs_pid)];
627 	LIST_INSERT_HEAD(ppp, pp, pfs_hash);
628 	mutex_exit(&pfs_ihash_lock);
629 }
630 
631 /*
632  * Remove the pfsnode from the hash table.
633  */
634 void
635 procfs_hashrem(struct pfsnode *pp)
636 {
637 	mutex_enter(&pfs_ihash_lock);
638 	LIST_REMOVE(pp, pfs_hash);
639 	mutex_exit(&pfs_ihash_lock);
640 }
641 
642 void
643 procfs_revoke_vnodes(struct proc *p, void *arg)
644 {
645 	struct pfsnode *pfs, *pnext;
646 	struct vnode *vp;
647 	struct mount *mp = (struct mount *)arg;
648 	struct pfs_hashhead *ppp;
649 
650 	if (!(p->p_flag & PK_SUGID))
651 		return;
652 
653 	mutex_enter(&pfs_ihash_lock);
654 	ppp = &pfs_hashtbl[PFSPIDHASH(p->p_pid)];
655 	for (pfs = LIST_FIRST(ppp); pfs; pfs = pnext) {
656 		vp = PFSTOV(pfs);
657 		pnext = LIST_NEXT(pfs, pfs_hash);
658 		mutex_enter(&vp->v_interlock);
659 		if (vp->v_usecount > 0 && pfs->pfs_pid == p->p_pid &&
660 		    vp->v_mount == mp) {
661 		    	vp->v_usecount++;
662 		    	mutex_exit(&vp->v_interlock);
663 			mutex_exit(&pfs_ihash_lock);
664 			VOP_REVOKE(vp, REVOKEALL);
665 			vrele(vp);
666 			mutex_enter(&pfs_ihash_lock);
667 		} else {
668 			mutex_exit(&vp->v_interlock);
669 		}
670 	}
671 	mutex_exit(&pfs_ihash_lock);
672 }
673 
674 int
675 procfs_proc_lock(int pid, struct proc **bunghole, int notfound)
676 {
677 	struct proc *tp;
678 	int error = 0;
679 
680 	mutex_enter(proc_lock);
681 
682 	if (pid == 0)
683 		tp = &proc0;
684 	else if ((tp = p_find(pid, PFIND_LOCKED)) == NULL)
685 		error = notfound;
686 	if (tp != NULL && !rw_tryenter(&tp->p_reflock, RW_READER))
687 		error = EBUSY;
688 
689 	mutex_exit(proc_lock);
690 
691 	*bunghole = tp;
692 	return error;
693 }
694 
695 void
696 procfs_proc_unlock(struct proc *p)
697 {
698 
699 	rw_exit(&p->p_reflock);
700 }
701 
702 int
703 procfs_doemul(struct lwp *curl, struct proc *p,
704     struct pfsnode *pfs, struct uio *uio)
705 {
706 	const char *ename = p->p_emul->e_name;
707 	return uiomove_frombuf(__UNCONST(ename), strlen(ename), uio);
708 }
709