xref: /netbsd-src/sys/miscfs/procfs/procfs_subr.c (revision 7c3f385475147b6e1c4753f2bee961630e2dfc40)
1 /*	$NetBSD: procfs_subr.c,v 1.86 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
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_subr.c	8.6 (Berkeley) 5/14/95
71  */
72 
73 /*
74  * Copyright (c) 1994 Christopher G. Demetriou.  All rights reserved.
75  * Copyright (c) 1993 Jan-Simon Pendry
76  *
77  * This code is derived from software contributed to Berkeley by
78  * Jan-Simon Pendry.
79  *
80  * Redistribution and use in source and binary forms, with or without
81  * modification, are permitted provided that the following conditions
82  * are met:
83  * 1. Redistributions of source code must retain the above copyright
84  *    notice, this list of conditions and the following disclaimer.
85  * 2. Redistributions in binary form must reproduce the above copyright
86  *    notice, this list of conditions and the following disclaimer in the
87  *    documentation and/or other materials provided with the distribution.
88  * 3. All advertising materials mentioning features or use of this software
89  *    must display the following acknowledgement:
90  *	This product includes software developed by the University of
91  *	California, Berkeley and its contributors.
92  * 4. Neither the name of the University nor the names of its contributors
93  *    may be used to endorse or promote products derived from this software
94  *    without specific prior written permission.
95  *
96  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
97  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
98  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
99  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
100  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
101  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
102  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
103  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
104  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
105  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
106  * SUCH DAMAGE.
107  *
108  *	@(#)procfs_subr.c	8.6 (Berkeley) 5/14/95
109  */
110 
111 #include <sys/cdefs.h>
112 __KERNEL_RCSID(0, "$NetBSD: procfs_subr.c,v 1.86 2008/03/21 21:55:00 ad Exp $");
113 
114 #include <sys/param.h>
115 #include <sys/systm.h>
116 #include <sys/time.h>
117 #include <sys/kernel.h>
118 #include <sys/proc.h>
119 #include <sys/vnode.h>
120 #include <sys/malloc.h>
121 #include <sys/stat.h>
122 #include <sys/file.h>
123 #include <sys/filedesc.h>
124 #include <sys/kauth.h>
125 
126 #include <miscfs/procfs/procfs.h>
127 
128 void procfs_hashins(struct pfsnode *);
129 void procfs_hashrem(struct pfsnode *);
130 struct vnode *procfs_hashget(pid_t, pfstype, int, struct mount *, int);
131 
132 LIST_HEAD(pfs_hashhead, pfsnode) *pfs_hashtbl;
133 u_long	pfs_ihash;	/* size of hash table - 1 */
134 #define PFSPIDHASH(pid)	((pid) & pfs_ihash)
135 
136 kmutex_t pfs_hashlock;
137 kmutex_t pfs_ihash_lock;
138 
139 #define	ISSET(t, f)	((t) & (f))
140 
141 /*
142  * allocate a pfsnode/vnode pair.  the vnode is
143  * referenced, and locked.
144  *
145  * the pid, pfs_type, and mount point uniquely
146  * identify a pfsnode.  the mount point is needed
147  * because someone might mount this filesystem
148  * twice.
149  *
150  * all pfsnodes are maintained on a singly-linked
151  * list.  new nodes are only allocated when they cannot
152  * be found on this list.  entries on the list are
153  * removed when the vfs reclaim entry is called.
154  *
155  * a single lock is kept for the entire list.  this is
156  * needed because the getnewvnode() function can block
157  * waiting for a vnode to become free, in which case there
158  * may be more than one process trying to get the same
159  * vnode.  this lock is only taken if we are going to
160  * call getnewvnode, since the kernel itself is single-threaded.
161  *
162  * if an entry is found on the list, then call vget() to
163  * take a reference.  this is done because there may be
164  * zero references to it and so it needs to removed from
165  * the vnode free list.
166  */
167 int
168 procfs_allocvp(mp, vpp, pid, pfs_type, fd, p)
169 	struct mount *mp;
170 	struct vnode **vpp;
171 	pid_t pid;
172 	pfstype pfs_type;
173 	int fd;
174 	struct proc *p;
175 {
176 	struct pfsnode *pfs;
177 	struct vnode *vp;
178 	int error;
179 
180  retry:
181 	*vpp = procfs_hashget(pid, pfs_type, fd, mp, LK_EXCLUSIVE);
182 	if (*vpp != NULL)
183 		return (0);
184 
185 	if ((error = getnewvnode(VT_PROCFS, mp, procfs_vnodeop_p, &vp)) != 0) {
186 		*vpp = NULL;
187 		return (error);
188 	}
189 	MALLOC(pfs, void *, sizeof(struct pfsnode), M_TEMP, M_WAITOK);
190 
191 	mutex_enter(&pfs_hashlock);
192 	if ((*vpp = procfs_hashget(pid, pfs_type, fd, mp, 0)) != NULL) {
193 		mutex_exit(&pfs_hashlock);
194 		ungetnewvnode(vp);
195 		FREE(pfs, M_TEMP);
196 		goto retry;
197 	}
198 
199 	vp->v_data = pfs;
200 	pfs->pfs_pid = pid;
201 	pfs->pfs_type = pfs_type;
202 	pfs->pfs_vnode = vp;
203 	pfs->pfs_flags = 0;
204 	pfs->pfs_fileno = PROCFS_FILENO(pid, pfs_type, fd);
205 	pfs->pfs_fd = fd;
206 
207 	switch (pfs_type) {
208 	case PFSroot:	/* /proc = dr-xr-xr-x */
209 		vp->v_vflag |= VV_ROOT;
210 		/*FALLTHROUGH*/
211 	case PFSproc:	/* /proc/N = dr-xr-xr-x */
212 		pfs->pfs_mode = S_IRUSR|S_IXUSR|S_IRGRP|S_IXGRP|S_IROTH|S_IXOTH;
213 		vp->v_type = VDIR;
214 		break;
215 
216 	case PFScurproc:	/* /proc/curproc = lr-xr-xr-x */
217 	case PFSself:	/* /proc/self    = lr-xr-xr-x */
218 	case PFScwd:	/* /proc/N/cwd = lr-xr-xr-x */
219 	case PFSchroot:	/* /proc/N/chroot = lr-xr-xr-x */
220 	case PFSexe:	/* /proc/N/exe = lr-xr-xr-x */
221 		pfs->pfs_mode = S_IRUSR|S_IXUSR|S_IRGRP|S_IXGRP|S_IROTH|S_IXOTH;
222 		vp->v_type = VLNK;
223 		break;
224 
225 	case PFSfd:
226 		if (fd == -1) {	/* /proc/N/fd = dr-xr-xr-x */
227 			pfs->pfs_mode = S_IRUSR|S_IXUSR;
228 			vp->v_type = VDIR;
229 		} else {	/* /proc/N/fd/M = [ps-]rw------- */
230 			file_t *fp;
231 			vnode_t *vxp;
232 
233 			if ((fp = fd_getfile2(p, pfs->pfs_fd)) == NULL) {
234 				error = EBADF;
235 				goto bad;
236 			}
237 
238 			pfs->pfs_mode = S_IRUSR|S_IWUSR;
239 			switch (fp->f_type) {
240 			case DTYPE_VNODE:
241 				vxp = fp->f_data;
242 
243 				/*
244 				 * We make symlinks for directories
245 				 * to avoid cycles.
246 				 */
247 				if (vxp->v_type == VDIR)
248 					goto symlink;
249 				vp->v_type = vxp->v_type;
250 				break;
251 			case DTYPE_PIPE:
252 				vp->v_type = VFIFO;
253 				break;
254 			case DTYPE_SOCKET:
255 				vp->v_type = VSOCK;
256 				break;
257 			case DTYPE_KQUEUE:
258 			case DTYPE_MISC:
259 			symlink:
260 				pfs->pfs_mode = S_IRUSR|S_IXUSR|S_IRGRP|
261 				    S_IXGRP|S_IROTH|S_IXOTH;
262 				vp->v_type = VLNK;
263 				break;
264 			default:
265 				error = EOPNOTSUPP;
266 				closef(fp);
267 				goto bad;
268 			}
269 			closef(fp);
270 		}
271 		break;
272 
273 	case PFSfile:	/* /proc/N/file = -rw------- */
274 	case PFSmem:	/* /proc/N/mem = -rw------- */
275 	case PFSregs:	/* /proc/N/regs = -rw------- */
276 	case PFSfpregs:	/* /proc/N/fpregs = -rw------- */
277 		pfs->pfs_mode = S_IRUSR|S_IWUSR;
278 		vp->v_type = VREG;
279 		break;
280 
281 	case PFSctl:	/* /proc/N/ctl = --w------ */
282 	case PFSnote:	/* /proc/N/note = --w------ */
283 	case PFSnotepg:	/* /proc/N/notepg = --w------ */
284 		pfs->pfs_mode = S_IWUSR;
285 		vp->v_type = VREG;
286 		break;
287 
288 	case PFSmap:	/* /proc/N/map = -r--r--r-- */
289 	case PFSmaps:	/* /proc/N/maps = -r--r--r-- */
290 	case PFSstatus:	/* /proc/N/status = -r--r--r-- */
291 	case PFSstat:	/* /proc/N/stat = -r--r--r-- */
292 	case PFScmdline:	/* /proc/N/cmdline = -r--r--r-- */
293 	case PFSemul:	/* /proc/N/emul = -r--r--r-- */
294 	case PFSmeminfo:	/* /proc/meminfo = -r--r--r-- */
295 	case PFScpustat:	/* /proc/stat = -r--r--r-- */
296 	case PFSdevices:	/* /proc/devices = -r--r--r-- */
297 	case PFScpuinfo:	/* /proc/cpuinfo = -r--r--r-- */
298 	case PFSuptime:	/* /proc/uptime = -r--r--r-- */
299 	case PFSmounts:	/* /proc/mounts = -r--r--r-- */
300 	case PFSloadavg:	/* /proc/loadavg = -r--r--r-- */
301 	case PFSstatm:	/* /proc/N/statm = -r--r--r-- */
302 		pfs->pfs_mode = S_IRUSR|S_IRGRP|S_IROTH;
303 		vp->v_type = VREG;
304 		break;
305 
306 #ifdef __HAVE_PROCFS_MACHDEP
307 	PROCFS_MACHDEP_NODETYPE_CASES
308 		procfs_machdep_allocvp(vp);
309 		break;
310 #endif
311 
312 	default:
313 		panic("procfs_allocvp");
314 	}
315 
316 	procfs_hashins(pfs);
317 	uvm_vnp_setsize(vp, 0);
318 	mutex_exit(&pfs_hashlock);
319 
320 	*vpp = vp;
321 	return (0);
322 
323  bad:
324 	mutex_exit(&pfs_hashlock);
325 	FREE(pfs, M_TEMP);
326 	vp->v_data = NULL;
327 	ungetnewvnode(vp);
328 	return (error);
329 }
330 
331 int
332 procfs_freevp(vp)
333 	struct vnode *vp;
334 {
335 	struct pfsnode *pfs = VTOPFS(vp);
336 
337 	procfs_hashrem(pfs);
338 
339 	FREE(vp->v_data, M_TEMP);
340 	vp->v_data = 0;
341 	return (0);
342 }
343 
344 int
345 procfs_rw(v)
346 	void *v;
347 {
348 	struct vop_read_args *ap = v;
349 	struct vnode *vp = ap->a_vp;
350 	struct uio *uio = ap->a_uio;
351 	struct lwp *curl;
352 	struct lwp *l;
353 	struct pfsnode *pfs = VTOPFS(vp);
354 	struct proc *p;
355 	int error;
356 
357 	if (uio->uio_offset < 0)
358 		return EINVAL;
359 
360 	if ((error = procfs_proc_lock(pfs->pfs_pid, &p, ESRCH)) != 0)
361 		return error;
362 
363 	curl = curlwp;
364 
365 	/*
366 	 * Do not allow init to be modified while in secure mode; it
367 	 * could be duped into changing the security level.
368 	 */
369 #define	M2K(m)	((m) == UIO_READ ? KAUTH_REQ_PROCESS_PROCFS_READ : \
370 		 KAUTH_REQ_PROCESS_PROCFS_WRITE)
371 	mutex_enter(&p->p_mutex);
372 	error = kauth_authorize_process(curl->l_cred, KAUTH_PROCESS_PROCFS,
373 	    p, pfs, KAUTH_ARG(M2K(uio->uio_rw)), NULL);
374 	mutex_exit(&p->p_mutex);
375 	if (error) {
376 		procfs_proc_unlock(p);
377 		return (error);
378 	}
379 #undef	M2K
380 
381 	mutex_enter(&p->p_smutex);
382 	l = proc_representative_lwp(p, NULL, 1);
383 	lwp_addref(l);
384 	mutex_exit(&p->p_smutex);
385 
386 	switch (pfs->pfs_type) {
387 	case PFSnote:
388 	case PFSnotepg:
389 		error = procfs_donote(curl, p, pfs, uio);
390 		break;
391 
392 	case PFSregs:
393 		error = procfs_doregs(curl, l, pfs, uio);
394 		break;
395 
396 	case PFSfpregs:
397 		error = procfs_dofpregs(curl, l, pfs, uio);
398 		break;
399 
400 	case PFSctl:
401 		error = procfs_doctl(curl, l, pfs, uio);
402 		break;
403 
404 	case PFSstatus:
405 		error = procfs_dostatus(curl, l, pfs, uio);
406 		break;
407 
408 	case PFSstat:
409 		error = procfs_do_pid_stat(curl, l, pfs, uio);
410 		break;
411 
412 	case PFSmap:
413 		error = procfs_domap(curl, p, pfs, uio, 0);
414 		break;
415 
416 	case PFSmaps:
417 		error = procfs_domap(curl, p, pfs, uio, 1);
418 		break;
419 
420 	case PFSmem:
421 		error = procfs_domem(curl, l, pfs, uio);
422 		break;
423 
424 	case PFScmdline:
425 		error = procfs_docmdline(curl, p, pfs, uio);
426 		break;
427 
428 	case PFSmeminfo:
429 		error = procfs_domeminfo(curl, p, pfs, uio);
430 		break;
431 
432 	case PFSdevices:
433 		error = procfs_dodevices(curl, p, pfs, uio);
434 		break;
435 
436 	case PFScpuinfo:
437 		error = procfs_docpuinfo(curl, p, pfs, uio);
438 		break;
439 
440 	case PFScpustat:
441 		error = procfs_docpustat(curl, p, pfs, uio);
442 		break;
443 
444 	case PFSloadavg:
445 		error = procfs_doloadavg(curl, p, pfs, uio);
446 		break;
447 
448 	case PFSstatm:
449 		error = procfs_do_pid_statm(curl, l, pfs, uio);
450 		break;
451 
452 	case PFSfd:
453 		error = procfs_dofd(curl, p, pfs, uio);
454 		break;
455 
456 	case PFSuptime:
457 		error = procfs_douptime(curl, p, pfs, uio);
458 		break;
459 
460 	case PFSmounts:
461 		error = procfs_domounts(curl, p, pfs, uio);
462 		break;
463 
464 	case PFSemul:
465 		error = procfs_doemul(curl, p, pfs, uio);
466 		break;
467 
468 #ifdef __HAVE_PROCFS_MACHDEP
469 	PROCFS_MACHDEP_NODETYPE_CASES
470 		error = procfs_machdep_rw(curl, l, pfs, uio);
471 		break;
472 #endif
473 
474 	default:
475 		error = EOPNOTSUPP;
476 		break;
477 	}
478 
479 	/*
480 	 * Release the references that we acquired earlier.
481 	 */
482 	lwp_delref(l);
483 	procfs_proc_unlock(p);
484 
485 	return (error);
486 }
487 
488 /*
489  * Get a string from userland into (bf).  Strip a trailing
490  * nl character (to allow easy access from the shell).
491  * The buffer should be *buflenp + 1 chars long.  vfs_getuserstr
492  * will automatically add a nul char at the end.
493  *
494  * Returns 0 on success or the following errors
495  *
496  * EINVAL:    file offset is non-zero.
497  * EMSGSIZE:  message is longer than kernel buffer
498  * EFAULT:    user i/o buffer is not addressable
499  */
500 int
501 vfs_getuserstr(uio, bf, buflenp)
502 	struct uio *uio;
503 	char *bf;
504 	int *buflenp;
505 {
506 	int xlen;
507 	int error;
508 
509 	if (uio->uio_offset != 0)
510 		return (EINVAL);
511 
512 	xlen = *buflenp;
513 
514 	/* must be able to read the whole string in one go */
515 	if (xlen < uio->uio_resid)
516 		return (EMSGSIZE);
517 	xlen = uio->uio_resid;
518 
519 	if ((error = uiomove(bf, xlen, uio)) != 0)
520 		return (error);
521 
522 	/* allow multiple writes without seeks */
523 	uio->uio_offset = 0;
524 
525 	/* cleanup string and remove trailing newline */
526 	bf[xlen] = '\0';
527 	xlen = strlen(bf);
528 	if (xlen > 0 && bf[xlen-1] == '\n')
529 		bf[--xlen] = '\0';
530 	*buflenp = xlen;
531 
532 	return (0);
533 }
534 
535 const vfs_namemap_t *
536 vfs_findname(nm, bf, buflen)
537 	const vfs_namemap_t *nm;
538 	const char *bf;
539 	int buflen;
540 {
541 
542 	for (; nm->nm_name; nm++)
543 		if (memcmp(bf, nm->nm_name, buflen+1) == 0)
544 			return (nm);
545 
546 	return (0);
547 }
548 
549 /*
550  * Initialize pfsnode hash table.
551  */
552 void
553 procfs_hashinit()
554 {
555 	mutex_init(&pfs_hashlock, MUTEX_DEFAULT, IPL_NONE);
556 	mutex_init(&pfs_ihash_lock, MUTEX_DEFAULT, IPL_NONE);
557 	pfs_hashtbl = hashinit(desiredvnodes / 4, HASH_LIST, M_UFSMNT,
558 	    M_WAITOK, &pfs_ihash);
559 }
560 
561 void
562 procfs_hashreinit()
563 {
564 	struct pfsnode *pp;
565 	struct pfs_hashhead *oldhash, *hash;
566 	u_long i, oldmask, mask, val;
567 
568 	hash = hashinit(desiredvnodes / 4, HASH_LIST, M_UFSMNT, M_WAITOK,
569 	    &mask);
570 
571 	mutex_enter(&pfs_ihash_lock);
572 	oldhash = pfs_hashtbl;
573 	oldmask = pfs_ihash;
574 	pfs_hashtbl = hash;
575 	pfs_ihash = mask;
576 	for (i = 0; i <= oldmask; i++) {
577 		while ((pp = LIST_FIRST(&oldhash[i])) != NULL) {
578 			LIST_REMOVE(pp, pfs_hash);
579 			val = PFSPIDHASH(pp->pfs_pid);
580 			LIST_INSERT_HEAD(&hash[val], pp, pfs_hash);
581 		}
582 	}
583 	mutex_exit(&pfs_ihash_lock);
584 	hashdone(oldhash, M_UFSMNT);
585 }
586 
587 /*
588  * Free pfsnode hash table.
589  */
590 void
591 procfs_hashdone()
592 {
593 	hashdone(pfs_hashtbl, M_UFSMNT);
594 	mutex_destroy(&pfs_hashlock);
595 	mutex_destroy(&pfs_ihash_lock);
596 }
597 
598 struct vnode *
599 procfs_hashget(pid, type, fd, mp, flags)
600 	pid_t pid;
601 	pfstype type;
602 	int fd;
603 	struct mount *mp;
604 	int flags;
605 {
606 	struct pfs_hashhead *ppp;
607 	struct pfsnode *pp;
608 	struct vnode *vp;
609 
610 loop:
611 	mutex_enter(&pfs_ihash_lock);
612 	ppp = &pfs_hashtbl[PFSPIDHASH(pid)];
613 	LIST_FOREACH(pp, ppp, pfs_hash) {
614 		vp = PFSTOV(pp);
615 		if (pid == pp->pfs_pid && pp->pfs_type == type &&
616 		    pp->pfs_fd == fd && vp->v_mount == mp) {
617 		    	if (flags == 0) {
618 				mutex_exit(&pfs_ihash_lock);
619 			} else {
620 				mutex_enter(&vp->v_interlock);
621 				mutex_exit(&pfs_ihash_lock);
622 				if (vget(vp, flags | LK_INTERLOCK))
623 					goto loop;
624 			}
625 			return (vp);
626 		}
627 	}
628 	mutex_exit(&pfs_ihash_lock);
629 	return (NULL);
630 }
631 
632 /*
633  * Insert the pfsnode into the hash table and lock it.
634  */
635 void
636 procfs_hashins(pp)
637 	struct pfsnode *pp;
638 {
639 	struct pfs_hashhead *ppp;
640 
641 	/* lock the pfsnode, then put it on the appropriate hash list */
642 	vlockmgr(&pp->pfs_vnode->v_lock, LK_EXCLUSIVE);
643 
644 	mutex_enter(&pfs_ihash_lock);
645 	ppp = &pfs_hashtbl[PFSPIDHASH(pp->pfs_pid)];
646 	LIST_INSERT_HEAD(ppp, pp, pfs_hash);
647 	mutex_exit(&pfs_ihash_lock);
648 }
649 
650 /*
651  * Remove the pfsnode from the hash table.
652  */
653 void
654 procfs_hashrem(pp)
655 	struct pfsnode *pp;
656 {
657 	mutex_enter(&pfs_ihash_lock);
658 	LIST_REMOVE(pp, pfs_hash);
659 	mutex_exit(&pfs_ihash_lock);
660 }
661 
662 void
663 procfs_revoke_vnodes(p, arg)
664 	struct proc *p;
665 	void *arg;
666 {
667 	struct pfsnode *pfs, *pnext;
668 	struct vnode *vp;
669 	struct mount *mp = (struct mount *)arg;
670 	struct pfs_hashhead *ppp;
671 
672 	if (!(p->p_flag & PK_SUGID))
673 		return;
674 
675 	mutex_enter(&pfs_ihash_lock);
676 	ppp = &pfs_hashtbl[PFSPIDHASH(p->p_pid)];
677 	for (pfs = LIST_FIRST(ppp); pfs; pfs = pnext) {
678 		vp = PFSTOV(pfs);
679 		pnext = LIST_NEXT(pfs, pfs_hash);
680 		mutex_enter(&vp->v_interlock);
681 		if (vp->v_usecount > 0 && pfs->pfs_pid == p->p_pid &&
682 		    vp->v_mount == mp) {
683 		    	vp->v_usecount++;
684 		    	mutex_exit(&vp->v_interlock);
685 			mutex_exit(&pfs_ihash_lock);
686 			VOP_REVOKE(vp, REVOKEALL);
687 			vrele(vp);
688 			mutex_enter(&pfs_ihash_lock);
689 		} else {
690 			mutex_exit(&vp->v_interlock);
691 		}
692 	}
693 	mutex_exit(&pfs_ihash_lock);
694 }
695 
696 int
697 procfs_proc_lock(int pid, struct proc **bunghole, int notfound)
698 {
699 	struct proc *tp;
700 	int error = 0;
701 
702 	mutex_enter(&proclist_lock);
703 
704 	if (pid == 0)
705 		tp = &proc0;
706 	else if ((tp = p_find(pid, PFIND_LOCKED)) == NULL)
707 		error = notfound;
708 	if (tp != NULL && !rw_tryenter(&tp->p_reflock, RW_READER))
709 		error = EBUSY;
710 
711 	mutex_exit(&proclist_lock);
712 
713 	*bunghole = tp;
714 	return error;
715 }
716 
717 void
718 procfs_proc_unlock(struct proc *p)
719 {
720 
721 	rw_exit(&p->p_reflock);
722 }
723 
724 int
725 procfs_doemul(struct lwp *curl, struct proc *p,
726     struct pfsnode *pfs, struct uio *uio)
727 {
728 	const char *ename = p->p_emul->e_name;
729 	return uiomove_frombuf(__UNCONST(ename), strlen(ename), uio);
730 }
731