xref: /netbsd-src/sys/miscfs/procfs/procfs_vfsops.c (revision c90f9c8c81c73a682a17b65b85df29fcecbb21a3)
1 /*	$NetBSD: procfs_vfsops.c,v 1.104 2020/04/04 20:49:30 ad Exp $	*/
2 
3 /*
4  * Copyright (c) 1993
5  *	The Regents of the University of California.  All rights reserved.
6  *
7  * This code is derived from software contributed to Berkeley by
8  * Jan-Simon Pendry.
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. Neither the name of the University nor the names of its contributors
19  *    may be used to endorse or promote products derived from this software
20  *    without specific prior written permission.
21  *
22  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
23  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
24  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
25  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
26  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
27  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
28  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
29  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
30  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
31  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
32  * SUCH DAMAGE.
33  *
34  *	@(#)procfs_vfsops.c	8.7 (Berkeley) 5/10/95
35  */
36 
37 /*
38  * Copyright (c) 1993 Jan-Simon Pendry
39  *
40  * This code is derived from software contributed to Berkeley by
41  * Jan-Simon Pendry.
42  *
43  * Redistribution and use in source and binary forms, with or without
44  * modification, are permitted provided that the following conditions
45  * are met:
46  * 1. Redistributions of source code must retain the above copyright
47  *    notice, this list of conditions and the following disclaimer.
48  * 2. Redistributions in binary form must reproduce the above copyright
49  *    notice, this list of conditions and the following disclaimer in the
50  *    documentation and/or other materials provided with the distribution.
51  * 3. All advertising materials mentioning features or use of this software
52  *    must display the following acknowledgement:
53  *	This product includes software developed by the University of
54  *	California, Berkeley and its contributors.
55  * 4. Neither the name of the University nor the names of its contributors
56  *    may be used to endorse or promote products derived from this software
57  *    without specific prior written permission.
58  *
59  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
60  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
61  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
62  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
63  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
64  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
65  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
66  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
67  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
68  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
69  * SUCH DAMAGE.
70  *
71  *	@(#)procfs_vfsops.c	8.7 (Berkeley) 5/10/95
72  */
73 
74 /*
75  * procfs VFS interface
76  */
77 
78 #include <sys/cdefs.h>
79 __KERNEL_RCSID(0, "$NetBSD: procfs_vfsops.c,v 1.104 2020/04/04 20:49:30 ad Exp $");
80 
81 #if defined(_KERNEL_OPT)
82 #include "opt_compat_netbsd.h"
83 #endif
84 
85 #include <sys/param.h>
86 #include <sys/time.h>
87 #include <sys/kernel.h>
88 #include <sys/systm.h>
89 #include <sys/sysctl.h>
90 #include <sys/proc.h>
91 #include <sys/buf.h>
92 #include <sys/syslog.h>
93 #include <sys/mount.h>
94 #include <sys/dirent.h>
95 #include <sys/signalvar.h>
96 #include <sys/vnode.h>
97 #include <sys/file.h>
98 #include <sys/filedesc.h>
99 #include <sys/kauth.h>
100 #include <sys/module.h>
101 
102 #include <miscfs/genfs/genfs.h>
103 
104 #include <miscfs/procfs/procfs.h>
105 
106 #include <uvm/uvm_extern.h>			/* for PAGE_SIZE */
107 
108 MODULE(MODULE_CLASS_VFS, procfs, "ptrace_common");
109 
110 VFS_PROTOS(procfs);
111 
112 static kauth_listener_t procfs_listener;
113 
114 /*
115  * VFS Operations.
116  *
117  * mount system call
118  */
119 /* ARGSUSED */
120 int
121 procfs_mount(
122     struct mount *mp,
123     const char *path,
124     void *data,
125     size_t *data_len)
126 {
127 	struct lwp *l = curlwp;
128 	struct procfsmount *pmnt;
129 	struct procfs_args *args = data;
130 	int error;
131 
132 	if (args == NULL)
133 		return EINVAL;
134 
135 	if (UIO_MX & (UIO_MX-1)) {
136 		log(LOG_ERR, "procfs: invalid directory entry size");
137 		return (EINVAL);
138 	}
139 
140 	if (mp->mnt_flag & MNT_GETARGS) {
141 		if (*data_len < sizeof *args)
142 			return EINVAL;
143 
144 		pmnt = VFSTOPROC(mp);
145 		if (pmnt == NULL)
146 			return EIO;
147 		args->version = PROCFS_ARGSVERSION;
148 		args->flags = pmnt->pmnt_flags;
149 		*data_len = sizeof *args;
150 		return 0;
151 	}
152 
153 	if (mp->mnt_flag & MNT_UPDATE)
154 		return (EOPNOTSUPP);
155 
156 	if (*data_len >= sizeof *args && args->version != PROCFS_ARGSVERSION)
157 		return EINVAL;
158 
159 	pmnt = kmem_zalloc(sizeof(struct procfsmount), KM_SLEEP);
160 
161 	mp->mnt_stat.f_namemax = PROCFS_MAXNAMLEN;
162 	mp->mnt_flag |= MNT_LOCAL;
163 	mp->mnt_data = pmnt;
164 	vfs_getnewfsid(mp);
165 
166 	error = set_statvfs_info(path, UIO_USERSPACE, "procfs", UIO_SYSSPACE,
167 	    mp->mnt_op->vfs_name, mp, l);
168 	pmnt->pmnt_exechook = exechook_establish(procfs_revoke_vnodes, mp);
169 	if (*data_len >= sizeof *args)
170 		pmnt->pmnt_flags = args->flags;
171 	else
172 		pmnt->pmnt_flags = 0;
173 
174 	mp->mnt_iflag |= IMNT_MPSAFE | IMNT_SHRLOOKUP;
175 	return error;
176 }
177 
178 /*
179  * unmount system call
180  */
181 int
182 procfs_unmount(struct mount *mp, int mntflags)
183 {
184 	int error;
185 	int flags = 0;
186 
187 	if (mntflags & MNT_FORCE)
188 		flags |= FORCECLOSE;
189 
190 	if ((error = vflush(mp, 0, flags)) != 0)
191 		return (error);
192 
193 	exechook_disestablish(VFSTOPROC(mp)->pmnt_exechook);
194 
195 	kmem_free(mp->mnt_data, sizeof(struct procfsmount));
196 	mp->mnt_data = NULL;
197 
198 	return 0;
199 }
200 
201 int
202 procfs_root(struct mount *mp, int lktype, struct vnode **vpp)
203 {
204 	int error;
205 
206 	error = procfs_allocvp(mp, vpp, 0, PFSroot, -1);
207 	if (error == 0) {
208 		error = vn_lock(*vpp, lktype);
209 		if (error != 0) {
210 			vrele(*vpp);
211 			*vpp = NULL;
212 		}
213 	}
214 
215 	return error;
216 }
217 
218 /* ARGSUSED */
219 int
220 procfs_start(struct mount *mp, int flags)
221 {
222 
223 	return (0);
224 }
225 
226 /*
227  * Get file system statistics.
228  */
229 int
230 procfs_statvfs(struct mount *mp, struct statvfs *sbp)
231 {
232 
233 	genfs_statvfs(mp, sbp);
234 
235 	sbp->f_bsize = PAGE_SIZE;
236 	sbp->f_frsize = PAGE_SIZE;
237 	sbp->f_iosize = PAGE_SIZE;
238 	sbp->f_blocks = 1;
239 	sbp->f_files = maxproc;			/* approx */
240 	sbp->f_ffree = maxproc - nprocs;	/* approx */
241 	sbp->f_favail = maxproc - nprocs;	/* approx */
242 
243 	return (0);
244 }
245 
246 /*ARGSUSED*/
247 int
248 procfs_sync(
249     struct mount *mp,
250     int waitfor,
251     kauth_cred_t uc)
252 {
253 
254 	return (0);
255 }
256 
257 /*ARGSUSED*/
258 int
259 procfs_vget(struct mount *mp, ino_t ino, int lktype,
260     struct vnode **vpp)
261 {
262 	return (EOPNOTSUPP);
263 }
264 
265 int
266 procfs_loadvnode(struct mount *mp, struct vnode *vp,
267     const void *key, size_t key_len, const void **new_key)
268 {
269 	int error;
270 	struct pfskey pfskey;
271 	struct pfsnode *pfs;
272 
273 	KASSERT(key_len == sizeof(pfskey));
274 	memcpy(&pfskey, key, key_len);
275 
276 	pfs = kmem_alloc(sizeof(*pfs), KM_SLEEP);
277 	pfs->pfs_pid = pfskey.pk_pid;
278 	pfs->pfs_type = pfskey.pk_type;
279 	pfs->pfs_fd = pfskey.pk_fd;
280 	pfs->pfs_vnode = vp;
281 	pfs->pfs_flags = 0;
282 	pfs->pfs_fileno =
283 	    PROCFS_FILENO(pfs->pfs_pid, pfs->pfs_type, pfs->pfs_fd);
284 	vp->v_tag = VT_PROCFS;
285 	vp->v_op = procfs_vnodeop_p;
286 	vp->v_data = pfs;
287 
288 	switch (pfs->pfs_type) {
289 	case PFSroot:	/* /proc = dr-xr-xr-x */
290 		vp->v_vflag |= VV_ROOT;
291 		/*FALLTHROUGH*/
292 	case PFSproc:	/* /proc/N = dr-xr-xr-x */
293 		pfs->pfs_mode = S_IRUSR|S_IXUSR|S_IRGRP|S_IXGRP|S_IROTH|S_IXOTH;
294 		vp->v_type = VDIR;
295 		break;
296 
297 	case PFStask:	/* /proc/N/task = dr-xr-xr-x */
298 		if (pfs->pfs_fd == -1) {
299 			pfs->pfs_mode = S_IRUSR|S_IXUSR|S_IRGRP|S_IXGRP|
300 			    S_IROTH|S_IXOTH;
301 			vp->v_type = VDIR;
302 			break;
303 		}
304 		/*FALLTHROUGH*/
305 	case PFScurproc:	/* /proc/curproc = lr-xr-xr-x */
306 	case PFSself:	/* /proc/self    = lr-xr-xr-x */
307 	case PFScwd:	/* /proc/N/cwd = lr-xr-xr-x */
308 	case PFSchroot:	/* /proc/N/chroot = lr-xr-xr-x */
309 	case PFSexe:	/* /proc/N/exe = lr-xr-xr-x */
310 		pfs->pfs_mode = S_IRUSR|S_IXUSR|S_IRGRP|S_IXGRP|S_IROTH|S_IXOTH;
311 		vp->v_type = VLNK;
312 		break;
313 
314 	case PFSfd:
315 		if (pfs->pfs_fd == -1) {	/* /proc/N/fd = dr-x------ */
316 			pfs->pfs_mode = S_IRUSR|S_IXUSR;
317 			vp->v_type = VDIR;
318 		} else {	/* /proc/N/fd/M = [ps-]rw------- */
319 			file_t *fp;
320 			vnode_t *vxp;
321 			struct proc *p;
322 
323 			mutex_enter(proc_lock);
324 			p = proc_find(pfs->pfs_pid);
325 			mutex_exit(proc_lock);
326 			if (p == NULL) {
327 				error = ENOENT;
328 				goto bad;
329 			}
330 			KASSERT(rw_read_held(&p->p_reflock));
331 			if ((fp = fd_getfile2(p, pfs->pfs_fd)) == NULL) {
332 				error = EBADF;
333 				goto bad;
334 			}
335 
336 			pfs->pfs_mode = S_IRUSR|S_IWUSR;
337 			switch (fp->f_type) {
338 			case DTYPE_VNODE:
339 				vxp = fp->f_vnode;
340 
341 				/*
342 				 * We make symlinks for directories
343 				 * to avoid cycles.
344 				 */
345 				if (vxp->v_type == VDIR)
346 					goto symlink;
347 				vp->v_type = vxp->v_type;
348 				break;
349 			case DTYPE_PIPE:
350 				vp->v_type = VFIFO;
351 				break;
352 			case DTYPE_SOCKET:
353 				vp->v_type = VSOCK;
354 				break;
355 			case DTYPE_KQUEUE:
356 			case DTYPE_MISC:
357 			case DTYPE_SEM:
358 			symlink:
359 				pfs->pfs_mode = S_IRUSR|S_IXUSR|S_IRGRP|
360 				    S_IXGRP|S_IROTH|S_IXOTH;
361 				vp->v_type = VLNK;
362 				break;
363 			default:
364 				error = EOPNOTSUPP;
365 				closef(fp);
366 				goto bad;
367 			}
368 			closef(fp);
369 		}
370 		break;
371 
372 	case PFSfile:	/* /proc/N/file = -rw------- */
373 	case PFSmem:	/* /proc/N/mem = -rw------- */
374 	case PFSregs:	/* /proc/N/regs = -rw------- */
375 	case PFSfpregs:	/* /proc/N/fpregs = -rw------- */
376 		pfs->pfs_mode = S_IRUSR|S_IWUSR;
377 		vp->v_type = VREG;
378 		break;
379 
380 	case PFSnote:	/* /proc/N/note = --w------ */
381 	case PFSnotepg:	/* /proc/N/notepg = --w------ */
382 		pfs->pfs_mode = S_IWUSR;
383 		vp->v_type = VREG;
384 		break;
385 
386 	case PFSmap:		/* /proc/N/map = -r-------- */
387 	case PFSmaps:		/* /proc/N/maps = -r-------- */
388 	case PFSauxv:		/* /proc/N/auxv = -r-------- */
389 		pfs->pfs_mode = S_IRUSR;
390 		vp->v_type = VREG;
391 		break;
392 
393 	case PFSstatus:		/* /proc/N/status = -r--r--r-- */
394 	case PFSstat:		/* /proc/N/stat = -r--r--r-- */
395 	case PFScmdline:	/* /proc/N/cmdline = -r--r--r-- */
396 	case PFSenviron:	/* /proc/N/environ = -r--r--r-- */
397 	case PFSemul:		/* /proc/N/emul = -r--r--r-- */
398 	case PFSmeminfo:	/* /proc/meminfo = -r--r--r-- */
399 	case PFScpustat:	/* /proc/stat = -r--r--r-- */
400 	case PFSdevices:	/* /proc/devices = -r--r--r-- */
401 	case PFScpuinfo:	/* /proc/cpuinfo = -r--r--r-- */
402 	case PFSuptime:		/* /proc/uptime = -r--r--r-- */
403 	case PFSmounts:		/* /proc/mounts = -r--r--r-- */
404 	case PFSloadavg:	/* /proc/loadavg = -r--r--r-- */
405 	case PFSstatm:		/* /proc/N/statm = -r--r--r-- */
406 	case PFSversion:	/* /proc/version = -r--r--r-- */
407 	case PFSlimit:		/* /proc/limit = -r--r--r-- */
408 		pfs->pfs_mode = S_IRUSR|S_IRGRP|S_IROTH;
409 		vp->v_type = VREG;
410 		break;
411 
412 #ifdef __HAVE_PROCFS_MACHDEP
413 	PROCFS_MACHDEP_NODETYPE_CASES
414 		procfs_machdep_allocvp(vp);
415 		break;
416 #endif
417 
418 	default:
419 		panic("procfs_allocvp");
420 	}
421 
422 	uvm_vnp_setsize(vp, 0);
423 	*new_key = &pfs->pfs_key;
424 
425 	return 0;
426 
427 bad:
428 	vp->v_tag =VT_NON;
429 	vp->v_type = VNON;
430 	vp->v_op = NULL;
431 	vp->v_data = NULL;
432 	kmem_free(pfs, sizeof(*pfs));
433 	return error;
434 }
435 
436 void
437 procfs_init(void)
438 {
439 
440 }
441 
442 void
443 procfs_reinit(void)
444 {
445 
446 }
447 
448 void
449 procfs_done(void)
450 {
451 
452 }
453 
454 extern const struct vnodeopv_desc procfs_vnodeop_opv_desc;
455 
456 const struct vnodeopv_desc * const procfs_vnodeopv_descs[] = {
457 	&procfs_vnodeop_opv_desc,
458 	NULL,
459 };
460 
461 struct vfsops procfs_vfsops = {
462 	.vfs_name = MOUNT_PROCFS,
463 	.vfs_min_mount_data = sizeof (struct procfs_args),
464 	.vfs_mount = procfs_mount,
465 	.vfs_start = procfs_start,
466 	.vfs_unmount = procfs_unmount,
467 	.vfs_root = procfs_root,
468 	.vfs_quotactl = (void *)eopnotsupp,
469 	.vfs_statvfs = procfs_statvfs,
470 	.vfs_sync = procfs_sync,
471 	.vfs_vget = procfs_vget,
472 	.vfs_loadvnode = procfs_loadvnode,
473 	.vfs_fhtovp = (void *)eopnotsupp,
474 	.vfs_vptofh = (void *)eopnotsupp,
475 	.vfs_init = procfs_init,
476 	.vfs_reinit = procfs_reinit,
477 	.vfs_done = procfs_done,
478 	.vfs_snapshot = (void *)eopnotsupp,
479 	.vfs_extattrctl = vfs_stdextattrctl,
480 	.vfs_suspendctl = genfs_suspendctl,
481 	.vfs_renamelock_enter = genfs_renamelock_enter,
482 	.vfs_renamelock_exit = genfs_renamelock_exit,
483 	.vfs_fsync = (void *)eopnotsupp,
484 	.vfs_opv_descs = procfs_vnodeopv_descs
485 };
486 
487 static int
488 procfs_listener_cb(kauth_cred_t cred, kauth_action_t action, void *cookie,
489     void *arg0, void *arg1, void *arg2, void *arg3)
490 {
491 	struct proc *p;
492 	struct pfsnode *pfs;
493 	int result;
494 
495 	result = KAUTH_RESULT_DEFER;
496 	p = arg0;
497 	pfs = arg1;
498 
499 	if (action != KAUTH_PROCESS_PROCFS)
500 		return result;
501 
502 	switch (pfs->pfs_type) {
503 	case PFSregs:
504 	case PFSfpregs:
505 	case PFSmem:
506 		if (kauth_cred_getuid(cred) != kauth_cred_getuid(p->p_cred) ||
507 		    ISSET(p->p_flag, PK_SUGID))
508 			break;
509 
510 		/*FALLTHROUGH*/
511 	default:
512 		result = KAUTH_RESULT_ALLOW;
513 		break;
514 	}
515 
516 	return result;
517 }
518 
519 SYSCTL_SETUP(procfs_sysctl_setup, "procfs sysctl")
520 {
521 
522 	sysctl_createv(clog, 0, NULL, NULL,
523 		       CTLFLAG_PERMANENT,
524 		       CTLTYPE_NODE, "procfs",
525 		       SYSCTL_DESCR("Process file system"),
526 		       NULL, 0, NULL, 0,
527 		       CTL_VFS, 12, CTL_EOL);
528 	/*
529 	 * XXX the "12" above could be dynamic, thereby eliminating
530 	 * one more instance of the "number to vfs" mapping problem,
531 	 * but "12" is the order as taken from sys/mount.h
532 	 */
533 }
534 
535 static int
536 procfs_modcmd(modcmd_t cmd, void *arg)
537 {
538 	int error;
539 
540 	switch (cmd) {
541 	case MODULE_CMD_INIT:
542 		error = vfs_attach(&procfs_vfsops);
543 		if (error != 0)
544 			break;
545 
546 		procfs_listener = kauth_listen_scope(KAUTH_SCOPE_PROCESS,
547 		    procfs_listener_cb, NULL);
548 
549 		break;
550 	case MODULE_CMD_FINI:
551 		error = vfs_detach(&procfs_vfsops);
552 		if (error != 0)
553 			break;
554 		kauth_unlisten_scope(procfs_listener);
555 		break;
556 	default:
557 		error = ENOTTY;
558 		break;
559 	}
560 
561 	return (error);
562 }
563