xref: /netbsd-src/sys/miscfs/procfs/procfs_subr.c (revision 4b71a66d0f279143147d63ebfcfd8a59499a3684)
1 /*	$NetBSD: procfs_subr.c,v 1.90 2008/05/05 17:11:17 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  *
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 1.90 2008/05/05 17:11:17 ad 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(mp, vpp, pid, pfs_type, fd, p)
162 	struct mount *mp;
163 	struct vnode **vpp;
164 	pid_t pid;
165 	pfstype pfs_type;
166 	int fd;
167 	struct proc *p;
168 {
169 	struct pfsnode *pfs;
170 	struct vnode *vp;
171 	int error;
172 
173  retry:
174 	*vpp = procfs_hashget(pid, pfs_type, fd, mp, LK_EXCLUSIVE);
175 	if (*vpp != NULL)
176 		return (0);
177 
178 	if ((error = getnewvnode(VT_PROCFS, mp, procfs_vnodeop_p, &vp)) != 0) {
179 		*vpp = NULL;
180 		return (error);
181 	}
182 	MALLOC(pfs, void *, sizeof(struct pfsnode), M_TEMP, M_WAITOK);
183 
184 	mutex_enter(&pfs_hashlock);
185 	if ((*vpp = procfs_hashget(pid, pfs_type, fd, mp, 0)) != NULL) {
186 		mutex_exit(&pfs_hashlock);
187 		ungetnewvnode(vp);
188 		FREE(pfs, M_TEMP);
189 		goto retry;
190 	}
191 
192 	vp->v_data = pfs;
193 	pfs->pfs_pid = pid;
194 	pfs->pfs_type = pfs_type;
195 	pfs->pfs_vnode = vp;
196 	pfs->pfs_flags = 0;
197 	pfs->pfs_fileno = PROCFS_FILENO(pid, pfs_type, fd);
198 	pfs->pfs_fd = fd;
199 
200 	switch (pfs_type) {
201 	case PFSroot:	/* /proc = dr-xr-xr-x */
202 		vp->v_vflag |= VV_ROOT;
203 		/*FALLTHROUGH*/
204 	case PFSproc:	/* /proc/N = dr-xr-xr-x */
205 		pfs->pfs_mode = S_IRUSR|S_IXUSR|S_IRGRP|S_IXGRP|S_IROTH|S_IXOTH;
206 		vp->v_type = VDIR;
207 		break;
208 
209 	case PFScurproc:	/* /proc/curproc = lr-xr-xr-x */
210 	case PFSself:	/* /proc/self    = lr-xr-xr-x */
211 	case PFScwd:	/* /proc/N/cwd = lr-xr-xr-x */
212 	case PFSchroot:	/* /proc/N/chroot = lr-xr-xr-x */
213 	case PFSexe:	/* /proc/N/exe = lr-xr-xr-x */
214 		pfs->pfs_mode = S_IRUSR|S_IXUSR|S_IRGRP|S_IXGRP|S_IROTH|S_IXOTH;
215 		vp->v_type = VLNK;
216 		break;
217 
218 	case PFSfd:
219 		if (fd == -1) {	/* /proc/N/fd = dr-xr-xr-x */
220 			pfs->pfs_mode = S_IRUSR|S_IXUSR;
221 			vp->v_type = VDIR;
222 		} else {	/* /proc/N/fd/M = [ps-]rw------- */
223 			file_t *fp;
224 			vnode_t *vxp;
225 
226 			if ((fp = fd_getfile2(p, pfs->pfs_fd)) == NULL) {
227 				error = EBADF;
228 				goto bad;
229 			}
230 
231 			pfs->pfs_mode = S_IRUSR|S_IWUSR;
232 			switch (fp->f_type) {
233 			case DTYPE_VNODE:
234 				vxp = fp->f_data;
235 
236 				/*
237 				 * We make symlinks for directories
238 				 * to avoid cycles.
239 				 */
240 				if (vxp->v_type == VDIR)
241 					goto symlink;
242 				vp->v_type = vxp->v_type;
243 				break;
244 			case DTYPE_PIPE:
245 				vp->v_type = VFIFO;
246 				break;
247 			case DTYPE_SOCKET:
248 				vp->v_type = VSOCK;
249 				break;
250 			case DTYPE_KQUEUE:
251 			case DTYPE_MISC:
252 			symlink:
253 				pfs->pfs_mode = S_IRUSR|S_IXUSR|S_IRGRP|
254 				    S_IXGRP|S_IROTH|S_IXOTH;
255 				vp->v_type = VLNK;
256 				break;
257 			default:
258 				error = EOPNOTSUPP;
259 				closef(fp);
260 				goto bad;
261 			}
262 			closef(fp);
263 		}
264 		break;
265 
266 	case PFSfile:	/* /proc/N/file = -rw------- */
267 	case PFSmem:	/* /proc/N/mem = -rw------- */
268 	case PFSregs:	/* /proc/N/regs = -rw------- */
269 	case PFSfpregs:	/* /proc/N/fpregs = -rw------- */
270 		pfs->pfs_mode = S_IRUSR|S_IWUSR;
271 		vp->v_type = VREG;
272 		break;
273 
274 	case PFSctl:	/* /proc/N/ctl = --w------ */
275 	case PFSnote:	/* /proc/N/note = --w------ */
276 	case PFSnotepg:	/* /proc/N/notepg = --w------ */
277 		pfs->pfs_mode = S_IWUSR;
278 		vp->v_type = VREG;
279 		break;
280 
281 	case PFSmap:	/* /proc/N/map = -r--r--r-- */
282 	case PFSmaps:	/* /proc/N/maps = -r--r--r-- */
283 	case PFSstatus:	/* /proc/N/status = -r--r--r-- */
284 	case PFSstat:	/* /proc/N/stat = -r--r--r-- */
285 	case PFScmdline:	/* /proc/N/cmdline = -r--r--r-- */
286 	case PFSemul:	/* /proc/N/emul = -r--r--r-- */
287 	case PFSmeminfo:	/* /proc/meminfo = -r--r--r-- */
288 	case PFScpustat:	/* /proc/stat = -r--r--r-- */
289 	case PFSdevices:	/* /proc/devices = -r--r--r-- */
290 	case PFScpuinfo:	/* /proc/cpuinfo = -r--r--r-- */
291 	case PFSuptime:	/* /proc/uptime = -r--r--r-- */
292 	case PFSmounts:	/* /proc/mounts = -r--r--r-- */
293 	case PFSloadavg:	/* /proc/loadavg = -r--r--r-- */
294 	case PFSstatm:	/* /proc/N/statm = -r--r--r-- */
295 		pfs->pfs_mode = S_IRUSR|S_IRGRP|S_IROTH;
296 		vp->v_type = VREG;
297 		break;
298 
299 #ifdef __HAVE_PROCFS_MACHDEP
300 	PROCFS_MACHDEP_NODETYPE_CASES
301 		procfs_machdep_allocvp(vp);
302 		break;
303 #endif
304 
305 	default:
306 		panic("procfs_allocvp");
307 	}
308 
309 	procfs_hashins(pfs);
310 	uvm_vnp_setsize(vp, 0);
311 	mutex_exit(&pfs_hashlock);
312 
313 	*vpp = vp;
314 	return (0);
315 
316  bad:
317 	mutex_exit(&pfs_hashlock);
318 	FREE(pfs, M_TEMP);
319 	vp->v_data = NULL;
320 	ungetnewvnode(vp);
321 	return (error);
322 }
323 
324 int
325 procfs_freevp(vp)
326 	struct vnode *vp;
327 {
328 	struct pfsnode *pfs = VTOPFS(vp);
329 
330 	procfs_hashrem(pfs);
331 
332 	FREE(vp->v_data, M_TEMP);
333 	vp->v_data = 0;
334 	return (0);
335 }
336 
337 int
338 procfs_rw(v)
339 	void *v;
340 {
341 	struct vop_read_args *ap = v;
342 	struct vnode *vp = ap->a_vp;
343 	struct uio *uio = ap->a_uio;
344 	struct lwp *curl;
345 	struct lwp *l;
346 	struct pfsnode *pfs = VTOPFS(vp);
347 	struct proc *p;
348 	int error;
349 
350 	if (uio->uio_offset < 0)
351 		return EINVAL;
352 
353 	if ((error = procfs_proc_lock(pfs->pfs_pid, &p, ESRCH)) != 0)
354 		return error;
355 
356 	curl = curlwp;
357 
358 	/*
359 	 * Do not allow init to be modified while in secure mode; it
360 	 * could be duped into changing the security level.
361 	 */
362 #define	M2K(m)	((m) == UIO_READ ? KAUTH_REQ_PROCESS_PROCFS_READ : \
363 		 KAUTH_REQ_PROCESS_PROCFS_WRITE)
364 	mutex_enter(p->p_lock);
365 	error = kauth_authorize_process(curl->l_cred, KAUTH_PROCESS_PROCFS,
366 	    p, pfs, KAUTH_ARG(M2K(uio->uio_rw)), NULL);
367 	mutex_exit(p->p_lock);
368 	if (error) {
369 		procfs_proc_unlock(p);
370 		return (error);
371 	}
372 #undef	M2K
373 
374 	mutex_enter(p->p_lock);
375 	l = proc_representative_lwp(p, NULL, 1);
376 	lwp_addref(l);
377 	mutex_exit(p->p_lock);
378 
379 	switch (pfs->pfs_type) {
380 	case PFSnote:
381 	case PFSnotepg:
382 		error = procfs_donote(curl, p, pfs, uio);
383 		break;
384 
385 	case PFSregs:
386 		error = procfs_doregs(curl, l, pfs, uio);
387 		break;
388 
389 	case PFSfpregs:
390 		error = procfs_dofpregs(curl, l, pfs, uio);
391 		break;
392 
393 	case PFSctl:
394 		error = procfs_doctl(curl, l, pfs, uio);
395 		break;
396 
397 	case PFSstatus:
398 		error = procfs_dostatus(curl, l, pfs, uio);
399 		break;
400 
401 	case PFSstat:
402 		error = procfs_do_pid_stat(curl, l, pfs, uio);
403 		break;
404 
405 	case PFSmap:
406 		error = procfs_domap(curl, p, pfs, uio, 0);
407 		break;
408 
409 	case PFSmaps:
410 		error = procfs_domap(curl, p, pfs, uio, 1);
411 		break;
412 
413 	case PFSmem:
414 		error = procfs_domem(curl, l, pfs, uio);
415 		break;
416 
417 	case PFScmdline:
418 		error = procfs_docmdline(curl, p, pfs, uio);
419 		break;
420 
421 	case PFSmeminfo:
422 		error = procfs_domeminfo(curl, p, pfs, uio);
423 		break;
424 
425 	case PFSdevices:
426 		error = procfs_dodevices(curl, p, pfs, uio);
427 		break;
428 
429 	case PFScpuinfo:
430 		error = procfs_docpuinfo(curl, p, pfs, uio);
431 		break;
432 
433 	case PFScpustat:
434 		error = procfs_docpustat(curl, p, pfs, uio);
435 		break;
436 
437 	case PFSloadavg:
438 		error = procfs_doloadavg(curl, p, pfs, uio);
439 		break;
440 
441 	case PFSstatm:
442 		error = procfs_do_pid_statm(curl, l, pfs, uio);
443 		break;
444 
445 	case PFSfd:
446 		error = procfs_dofd(curl, p, pfs, uio);
447 		break;
448 
449 	case PFSuptime:
450 		error = procfs_douptime(curl, p, pfs, uio);
451 		break;
452 
453 	case PFSmounts:
454 		error = procfs_domounts(curl, p, pfs, uio);
455 		break;
456 
457 	case PFSemul:
458 		error = procfs_doemul(curl, p, pfs, uio);
459 		break;
460 
461 #ifdef __HAVE_PROCFS_MACHDEP
462 	PROCFS_MACHDEP_NODETYPE_CASES
463 		error = procfs_machdep_rw(curl, l, pfs, uio);
464 		break;
465 #endif
466 
467 	default:
468 		error = EOPNOTSUPP;
469 		break;
470 	}
471 
472 	/*
473 	 * Release the references that we acquired earlier.
474 	 */
475 	lwp_delref(l);
476 	procfs_proc_unlock(p);
477 
478 	return (error);
479 }
480 
481 /*
482  * Get a string from userland into (bf).  Strip a trailing
483  * nl character (to allow easy access from the shell).
484  * The buffer should be *buflenp + 1 chars long.  vfs_getuserstr
485  * will automatically add a nul char at the end.
486  *
487  * Returns 0 on success or the following errors
488  *
489  * EINVAL:    file offset is non-zero.
490  * EMSGSIZE:  message is longer than kernel buffer
491  * EFAULT:    user i/o buffer is not addressable
492  */
493 int
494 vfs_getuserstr(uio, bf, buflenp)
495 	struct uio *uio;
496 	char *bf;
497 	int *buflenp;
498 {
499 	int xlen;
500 	int error;
501 
502 	if (uio->uio_offset != 0)
503 		return (EINVAL);
504 
505 	xlen = *buflenp;
506 
507 	/* must be able to read the whole string in one go */
508 	if (xlen < uio->uio_resid)
509 		return (EMSGSIZE);
510 	xlen = uio->uio_resid;
511 
512 	if ((error = uiomove(bf, xlen, uio)) != 0)
513 		return (error);
514 
515 	/* allow multiple writes without seeks */
516 	uio->uio_offset = 0;
517 
518 	/* cleanup string and remove trailing newline */
519 	bf[xlen] = '\0';
520 	xlen = strlen(bf);
521 	if (xlen > 0 && bf[xlen-1] == '\n')
522 		bf[--xlen] = '\0';
523 	*buflenp = xlen;
524 
525 	return (0);
526 }
527 
528 const vfs_namemap_t *
529 vfs_findname(nm, bf, buflen)
530 	const vfs_namemap_t *nm;
531 	const char *bf;
532 	int buflen;
533 {
534 
535 	for (; nm->nm_name; nm++)
536 		if (memcmp(bf, nm->nm_name, buflen+1) == 0)
537 			return (nm);
538 
539 	return (0);
540 }
541 
542 /*
543  * Initialize pfsnode hash table.
544  */
545 void
546 procfs_hashinit()
547 {
548 	mutex_init(&pfs_hashlock, MUTEX_DEFAULT, IPL_NONE);
549 	mutex_init(&pfs_ihash_lock, MUTEX_DEFAULT, IPL_NONE);
550 	pfs_hashtbl = hashinit(desiredvnodes / 4, HASH_LIST, true, &pfs_ihash);
551 }
552 
553 void
554 procfs_hashreinit()
555 {
556 	struct pfsnode *pp;
557 	struct pfs_hashhead *oldhash, *hash;
558 	u_long i, oldmask, mask, val;
559 
560 	hash = hashinit(desiredvnodes / 4, HASH_LIST, true, &mask);
561 
562 	mutex_enter(&pfs_ihash_lock);
563 	oldhash = pfs_hashtbl;
564 	oldmask = pfs_ihash;
565 	pfs_hashtbl = hash;
566 	pfs_ihash = mask;
567 	for (i = 0; i <= oldmask; i++) {
568 		while ((pp = LIST_FIRST(&oldhash[i])) != NULL) {
569 			LIST_REMOVE(pp, pfs_hash);
570 			val = PFSPIDHASH(pp->pfs_pid);
571 			LIST_INSERT_HEAD(&hash[val], pp, pfs_hash);
572 		}
573 	}
574 	mutex_exit(&pfs_ihash_lock);
575 	hashdone(oldhash, HASH_LIST, oldmask);
576 }
577 
578 /*
579  * Free pfsnode hash table.
580  */
581 void
582 procfs_hashdone()
583 {
584 	hashdone(pfs_hashtbl, HASH_LIST, pfs_ihash);
585 	mutex_destroy(&pfs_hashlock);
586 	mutex_destroy(&pfs_ihash_lock);
587 }
588 
589 struct vnode *
590 procfs_hashget(pid, type, fd, mp, flags)
591 	pid_t pid;
592 	pfstype type;
593 	int fd;
594 	struct mount *mp;
595 	int flags;
596 {
597 	struct pfs_hashhead *ppp;
598 	struct pfsnode *pp;
599 	struct vnode *vp;
600 
601 loop:
602 	mutex_enter(&pfs_ihash_lock);
603 	ppp = &pfs_hashtbl[PFSPIDHASH(pid)];
604 	LIST_FOREACH(pp, ppp, pfs_hash) {
605 		vp = PFSTOV(pp);
606 		if (pid == pp->pfs_pid && pp->pfs_type == type &&
607 		    pp->pfs_fd == fd && vp->v_mount == mp) {
608 		    	if (flags == 0) {
609 				mutex_exit(&pfs_ihash_lock);
610 			} else {
611 				mutex_enter(&vp->v_interlock);
612 				mutex_exit(&pfs_ihash_lock);
613 				if (vget(vp, flags | LK_INTERLOCK))
614 					goto loop;
615 			}
616 			return (vp);
617 		}
618 	}
619 	mutex_exit(&pfs_ihash_lock);
620 	return (NULL);
621 }
622 
623 /*
624  * Insert the pfsnode into the hash table and lock it.
625  */
626 void
627 procfs_hashins(pp)
628 	struct pfsnode *pp;
629 {
630 	struct pfs_hashhead *ppp;
631 
632 	/* lock the pfsnode, then put it on the appropriate hash list */
633 	vlockmgr(&pp->pfs_vnode->v_lock, LK_EXCLUSIVE);
634 
635 	mutex_enter(&pfs_ihash_lock);
636 	ppp = &pfs_hashtbl[PFSPIDHASH(pp->pfs_pid)];
637 	LIST_INSERT_HEAD(ppp, pp, pfs_hash);
638 	mutex_exit(&pfs_ihash_lock);
639 }
640 
641 /*
642  * Remove the pfsnode from the hash table.
643  */
644 void
645 procfs_hashrem(pp)
646 	struct pfsnode *pp;
647 {
648 	mutex_enter(&pfs_ihash_lock);
649 	LIST_REMOVE(pp, pfs_hash);
650 	mutex_exit(&pfs_ihash_lock);
651 }
652 
653 void
654 procfs_revoke_vnodes(p, arg)
655 	struct proc *p;
656 	void *arg;
657 {
658 	struct pfsnode *pfs, *pnext;
659 	struct vnode *vp;
660 	struct mount *mp = (struct mount *)arg;
661 	struct pfs_hashhead *ppp;
662 
663 	if (!(p->p_flag & PK_SUGID))
664 		return;
665 
666 	mutex_enter(&pfs_ihash_lock);
667 	ppp = &pfs_hashtbl[PFSPIDHASH(p->p_pid)];
668 	for (pfs = LIST_FIRST(ppp); pfs; pfs = pnext) {
669 		vp = PFSTOV(pfs);
670 		pnext = LIST_NEXT(pfs, pfs_hash);
671 		mutex_enter(&vp->v_interlock);
672 		if (vp->v_usecount > 0 && pfs->pfs_pid == p->p_pid &&
673 		    vp->v_mount == mp) {
674 		    	vp->v_usecount++;
675 		    	mutex_exit(&vp->v_interlock);
676 			mutex_exit(&pfs_ihash_lock);
677 			VOP_REVOKE(vp, REVOKEALL);
678 			vrele(vp);
679 			mutex_enter(&pfs_ihash_lock);
680 		} else {
681 			mutex_exit(&vp->v_interlock);
682 		}
683 	}
684 	mutex_exit(&pfs_ihash_lock);
685 }
686 
687 int
688 procfs_proc_lock(int pid, struct proc **bunghole, int notfound)
689 {
690 	struct proc *tp;
691 	int error = 0;
692 
693 	mutex_enter(proc_lock);
694 
695 	if (pid == 0)
696 		tp = &proc0;
697 	else if ((tp = p_find(pid, PFIND_LOCKED)) == NULL)
698 		error = notfound;
699 	if (tp != NULL && !rw_tryenter(&tp->p_reflock, RW_READER))
700 		error = EBUSY;
701 
702 	mutex_exit(proc_lock);
703 
704 	*bunghole = tp;
705 	return error;
706 }
707 
708 void
709 procfs_proc_unlock(struct proc *p)
710 {
711 
712 	rw_exit(&p->p_reflock);
713 }
714 
715 int
716 procfs_doemul(struct lwp *curl, struct proc *p,
717     struct pfsnode *pfs, struct uio *uio)
718 {
719 	const char *ename = p->p_emul->e_name;
720 	return uiomove_frombuf(__UNCONST(ename), strlen(ename), uio);
721 }
722