xref: /netbsd-src/sys/miscfs/procfs/procfs_subr.c (revision cac8e449158efc7261bebc8657cbb0125a2cfdde)
1 /*	$NetBSD: procfs_subr.c,v 1.91 2008/07/02 19:49:58 rmind 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 1.91 2008/07/02 19:49:58 rmind 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 = LIST_FIRST(&p->p_lwps);
376 	KASSERT(l != NULL);
377 	lwp_addref(l);
378 	mutex_exit(p->p_lock);
379 
380 	switch (pfs->pfs_type) {
381 	case PFSnote:
382 	case PFSnotepg:
383 		error = procfs_donote(curl, p, pfs, uio);
384 		break;
385 
386 	case PFSregs:
387 		error = procfs_doregs(curl, l, pfs, uio);
388 		break;
389 
390 	case PFSfpregs:
391 		error = procfs_dofpregs(curl, l, pfs, uio);
392 		break;
393 
394 	case PFSctl:
395 		error = procfs_doctl(curl, l, pfs, uio);
396 		break;
397 
398 	case PFSstatus:
399 		error = procfs_dostatus(curl, l, pfs, uio);
400 		break;
401 
402 	case PFSstat:
403 		error = procfs_do_pid_stat(curl, l, pfs, uio);
404 		break;
405 
406 	case PFSmap:
407 		error = procfs_domap(curl, p, pfs, uio, 0);
408 		break;
409 
410 	case PFSmaps:
411 		error = procfs_domap(curl, p, pfs, uio, 1);
412 		break;
413 
414 	case PFSmem:
415 		error = procfs_domem(curl, l, pfs, uio);
416 		break;
417 
418 	case PFScmdline:
419 		error = procfs_docmdline(curl, p, pfs, uio);
420 		break;
421 
422 	case PFSmeminfo:
423 		error = procfs_domeminfo(curl, p, pfs, uio);
424 		break;
425 
426 	case PFSdevices:
427 		error = procfs_dodevices(curl, p, pfs, uio);
428 		break;
429 
430 	case PFScpuinfo:
431 		error = procfs_docpuinfo(curl, p, pfs, uio);
432 		break;
433 
434 	case PFScpustat:
435 		error = procfs_docpustat(curl, p, pfs, uio);
436 		break;
437 
438 	case PFSloadavg:
439 		error = procfs_doloadavg(curl, p, pfs, uio);
440 		break;
441 
442 	case PFSstatm:
443 		error = procfs_do_pid_statm(curl, l, pfs, uio);
444 		break;
445 
446 	case PFSfd:
447 		error = procfs_dofd(curl, p, pfs, uio);
448 		break;
449 
450 	case PFSuptime:
451 		error = procfs_douptime(curl, p, pfs, uio);
452 		break;
453 
454 	case PFSmounts:
455 		error = procfs_domounts(curl, p, pfs, uio);
456 		break;
457 
458 	case PFSemul:
459 		error = procfs_doemul(curl, p, pfs, uio);
460 		break;
461 
462 #ifdef __HAVE_PROCFS_MACHDEP
463 	PROCFS_MACHDEP_NODETYPE_CASES
464 		error = procfs_machdep_rw(curl, l, pfs, uio);
465 		break;
466 #endif
467 
468 	default:
469 		error = EOPNOTSUPP;
470 		break;
471 	}
472 
473 	/*
474 	 * Release the references that we acquired earlier.
475 	 */
476 	lwp_delref(l);
477 	procfs_proc_unlock(p);
478 
479 	return (error);
480 }
481 
482 /*
483  * Get a string from userland into (bf).  Strip a trailing
484  * nl character (to allow easy access from the shell).
485  * The buffer should be *buflenp + 1 chars long.  vfs_getuserstr
486  * will automatically add a nul char at the end.
487  *
488  * Returns 0 on success or the following errors
489  *
490  * EINVAL:    file offset is non-zero.
491  * EMSGSIZE:  message is longer than kernel buffer
492  * EFAULT:    user i/o buffer is not addressable
493  */
494 int
495 vfs_getuserstr(uio, bf, buflenp)
496 	struct uio *uio;
497 	char *bf;
498 	int *buflenp;
499 {
500 	int xlen;
501 	int error;
502 
503 	if (uio->uio_offset != 0)
504 		return (EINVAL);
505 
506 	xlen = *buflenp;
507 
508 	/* must be able to read the whole string in one go */
509 	if (xlen < uio->uio_resid)
510 		return (EMSGSIZE);
511 	xlen = uio->uio_resid;
512 
513 	if ((error = uiomove(bf, xlen, uio)) != 0)
514 		return (error);
515 
516 	/* allow multiple writes without seeks */
517 	uio->uio_offset = 0;
518 
519 	/* cleanup string and remove trailing newline */
520 	bf[xlen] = '\0';
521 	xlen = strlen(bf);
522 	if (xlen > 0 && bf[xlen-1] == '\n')
523 		bf[--xlen] = '\0';
524 	*buflenp = xlen;
525 
526 	return (0);
527 }
528 
529 const vfs_namemap_t *
530 vfs_findname(nm, bf, buflen)
531 	const vfs_namemap_t *nm;
532 	const char *bf;
533 	int buflen;
534 {
535 
536 	for (; nm->nm_name; nm++)
537 		if (memcmp(bf, nm->nm_name, buflen+1) == 0)
538 			return (nm);
539 
540 	return (0);
541 }
542 
543 /*
544  * Initialize pfsnode hash table.
545  */
546 void
547 procfs_hashinit()
548 {
549 	mutex_init(&pfs_hashlock, MUTEX_DEFAULT, IPL_NONE);
550 	mutex_init(&pfs_ihash_lock, MUTEX_DEFAULT, IPL_NONE);
551 	pfs_hashtbl = hashinit(desiredvnodes / 4, HASH_LIST, true, &pfs_ihash);
552 }
553 
554 void
555 procfs_hashreinit()
556 {
557 	struct pfsnode *pp;
558 	struct pfs_hashhead *oldhash, *hash;
559 	u_long i, oldmask, mask, val;
560 
561 	hash = hashinit(desiredvnodes / 4, HASH_LIST, true, &mask);
562 
563 	mutex_enter(&pfs_ihash_lock);
564 	oldhash = pfs_hashtbl;
565 	oldmask = pfs_ihash;
566 	pfs_hashtbl = hash;
567 	pfs_ihash = mask;
568 	for (i = 0; i <= oldmask; i++) {
569 		while ((pp = LIST_FIRST(&oldhash[i])) != NULL) {
570 			LIST_REMOVE(pp, pfs_hash);
571 			val = PFSPIDHASH(pp->pfs_pid);
572 			LIST_INSERT_HEAD(&hash[val], pp, pfs_hash);
573 		}
574 	}
575 	mutex_exit(&pfs_ihash_lock);
576 	hashdone(oldhash, HASH_LIST, oldmask);
577 }
578 
579 /*
580  * Free pfsnode hash table.
581  */
582 void
583 procfs_hashdone()
584 {
585 	hashdone(pfs_hashtbl, HASH_LIST, pfs_ihash);
586 	mutex_destroy(&pfs_hashlock);
587 	mutex_destroy(&pfs_ihash_lock);
588 }
589 
590 struct vnode *
591 procfs_hashget(pid, type, fd, mp, flags)
592 	pid_t pid;
593 	pfstype type;
594 	int fd;
595 	struct mount *mp;
596 	int flags;
597 {
598 	struct pfs_hashhead *ppp;
599 	struct pfsnode *pp;
600 	struct vnode *vp;
601 
602 loop:
603 	mutex_enter(&pfs_ihash_lock);
604 	ppp = &pfs_hashtbl[PFSPIDHASH(pid)];
605 	LIST_FOREACH(pp, ppp, pfs_hash) {
606 		vp = PFSTOV(pp);
607 		if (pid == pp->pfs_pid && pp->pfs_type == type &&
608 		    pp->pfs_fd == fd && vp->v_mount == mp) {
609 		    	if (flags == 0) {
610 				mutex_exit(&pfs_ihash_lock);
611 			} else {
612 				mutex_enter(&vp->v_interlock);
613 				mutex_exit(&pfs_ihash_lock);
614 				if (vget(vp, flags | LK_INTERLOCK))
615 					goto loop;
616 			}
617 			return (vp);
618 		}
619 	}
620 	mutex_exit(&pfs_ihash_lock);
621 	return (NULL);
622 }
623 
624 /*
625  * Insert the pfsnode into the hash table and lock it.
626  */
627 void
628 procfs_hashins(pp)
629 	struct pfsnode *pp;
630 {
631 	struct pfs_hashhead *ppp;
632 
633 	/* lock the pfsnode, then put it on the appropriate hash list */
634 	vlockmgr(&pp->pfs_vnode->v_lock, LK_EXCLUSIVE);
635 
636 	mutex_enter(&pfs_ihash_lock);
637 	ppp = &pfs_hashtbl[PFSPIDHASH(pp->pfs_pid)];
638 	LIST_INSERT_HEAD(ppp, pp, pfs_hash);
639 	mutex_exit(&pfs_ihash_lock);
640 }
641 
642 /*
643  * Remove the pfsnode from the hash table.
644  */
645 void
646 procfs_hashrem(pp)
647 	struct pfsnode *pp;
648 {
649 	mutex_enter(&pfs_ihash_lock);
650 	LIST_REMOVE(pp, pfs_hash);
651 	mutex_exit(&pfs_ihash_lock);
652 }
653 
654 void
655 procfs_revoke_vnodes(p, arg)
656 	struct proc *p;
657 	void *arg;
658 {
659 	struct pfsnode *pfs, *pnext;
660 	struct vnode *vp;
661 	struct mount *mp = (struct mount *)arg;
662 	struct pfs_hashhead *ppp;
663 
664 	if (!(p->p_flag & PK_SUGID))
665 		return;
666 
667 	mutex_enter(&pfs_ihash_lock);
668 	ppp = &pfs_hashtbl[PFSPIDHASH(p->p_pid)];
669 	for (pfs = LIST_FIRST(ppp); pfs; pfs = pnext) {
670 		vp = PFSTOV(pfs);
671 		pnext = LIST_NEXT(pfs, pfs_hash);
672 		mutex_enter(&vp->v_interlock);
673 		if (vp->v_usecount > 0 && pfs->pfs_pid == p->p_pid &&
674 		    vp->v_mount == mp) {
675 		    	vp->v_usecount++;
676 		    	mutex_exit(&vp->v_interlock);
677 			mutex_exit(&pfs_ihash_lock);
678 			VOP_REVOKE(vp, REVOKEALL);
679 			vrele(vp);
680 			mutex_enter(&pfs_ihash_lock);
681 		} else {
682 			mutex_exit(&vp->v_interlock);
683 		}
684 	}
685 	mutex_exit(&pfs_ihash_lock);
686 }
687 
688 int
689 procfs_proc_lock(int pid, struct proc **bunghole, int notfound)
690 {
691 	struct proc *tp;
692 	int error = 0;
693 
694 	mutex_enter(proc_lock);
695 
696 	if (pid == 0)
697 		tp = &proc0;
698 	else if ((tp = p_find(pid, PFIND_LOCKED)) == NULL)
699 		error = notfound;
700 	if (tp != NULL && !rw_tryenter(&tp->p_reflock, RW_READER))
701 		error = EBUSY;
702 
703 	mutex_exit(proc_lock);
704 
705 	*bunghole = tp;
706 	return error;
707 }
708 
709 void
710 procfs_proc_unlock(struct proc *p)
711 {
712 
713 	rw_exit(&p->p_reflock);
714 }
715 
716 int
717 procfs_doemul(struct lwp *curl, struct proc *p,
718     struct pfsnode *pfs, struct uio *uio)
719 {
720 	const char *ename = p->p_emul->e_name;
721 	return uiomove_frombuf(__UNCONST(ename), strlen(ename), uio);
722 }
723