Home | History | Annotate | Line # | Download | only in procfs
      1 /*	$NetBSD: procfs_vfsops.c,v 1.120 2024/09/14 01:37:42 pgoyette 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 1.120 2024/09/14 01:37:42 pgoyette Exp $");
     80 
     81 #if defined(_KERNEL_OPT)
     82 #include "opt_compat_netbsd.h"
     83 #include "opt_sysv_ipc.h"
     84 #include "opt_mqueue.h"
     85 #endif
     86 
     87 #include <sys/param.h>
     88 #include <sys/atomic.h>
     89 #include <sys/buf.h>
     90 #include <sys/dirent.h>
     91 #include <sys/file.h>
     92 #include <sys/filedesc.h>
     93 #include <sys/fstrans.h>
     94 #include <sys/kauth.h>
     95 #include <sys/kernel.h>
     96 #include <sys/module.h>
     97 #include <sys/mount.h>
     98 #include <sys/proc.h>
     99 #include <sys/signalvar.h>
    100 #include <sys/sysctl.h>
    101 #include <sys/syslog.h>
    102 #include <sys/systm.h>
    103 #include <sys/time.h>
    104 #include <sys/vnode.h>
    105 
    106 #include <miscfs/genfs/genfs.h>
    107 
    108 #include <miscfs/procfs/procfs.h>
    109 
    110 #include <uvm/uvm_extern.h>			/* for PAGE_SIZE */
    111 
    112 MODULE(MODULE_CLASS_VFS, procfs, "ptrace_common"
    113 #if defined(MQUEUE)
    114 				 ",mqueue"
    115 #endif
    116 #if defined(SYSVSHM) || defined(SYSVSEM) || defined(SYSVMSG)
    117 				 ",sysv_ipc"
    118 #endif
    119 );
    120 
    121 VFS_PROTOS(procfs);
    122 
    123 #define PROCFS_HASHSIZE	256
    124 #define PROCFS_EXEC_HOOK ((void *)1)
    125 #define PROCFS_EXIT_HOOK ((void *)2)
    126 
    127 static kauth_listener_t procfs_listener;
    128 static void *procfs_exechook;
    129 static void *procfs_exithook;
    130 LIST_HEAD(hashhead, pfsnode);
    131 static u_long procfs_hashmask;
    132 static struct hashhead *procfs_hashtab;
    133 static kmutex_t procfs_hashlock;
    134 
    135 static struct hashhead *
    136 procfs_hashhead(pid_t pid)
    137 {
    138 
    139 	return &procfs_hashtab[pid & procfs_hashmask];
    140 }
    141 
    142 void
    143 procfs_hashrem(struct pfsnode *pfs)
    144 {
    145 
    146 	mutex_enter(&procfs_hashlock);
    147 	LIST_REMOVE(pfs, pfs_hash);
    148 	mutex_exit(&procfs_hashlock);
    149 }
    150 
    151 /*
    152  * VFS Operations.
    153  *
    154  * mount system call
    155  */
    156 /* ARGSUSED */
    157 int
    158 procfs_mount(
    159     struct mount *mp,
    160     const char *path,
    161     void *data,
    162     size_t *data_len)
    163 {
    164 	struct lwp *l = curlwp;
    165 	struct procfsmount *pmnt;
    166 	struct procfs_args *args = data;
    167 	int error;
    168 
    169 	if (args == NULL)
    170 		return EINVAL;
    171 
    172 	if (UIO_MX & (UIO_MX-1)) {
    173 		log(LOG_ERR, "procfs: invalid directory entry size");
    174 		return (EINVAL);
    175 	}
    176 
    177 	if (mp->mnt_flag & MNT_GETARGS) {
    178 		if (*data_len < sizeof *args)
    179 			return EINVAL;
    180 
    181 		pmnt = VFSTOPROC(mp);
    182 		if (pmnt == NULL)
    183 			return EIO;
    184 		args->version = PROCFS_ARGSVERSION;
    185 		args->flags = pmnt->pmnt_flags;
    186 		*data_len = sizeof *args;
    187 		return 0;
    188 	}
    189 
    190 	if (mp->mnt_flag & MNT_UPDATE)
    191 		return (EOPNOTSUPP);
    192 
    193 	if (*data_len >= sizeof *args && args->version != PROCFS_ARGSVERSION)
    194 		return EINVAL;
    195 
    196 	pmnt = kmem_zalloc(sizeof(struct procfsmount), KM_SLEEP);
    197 
    198 	mp->mnt_stat.f_namemax = PROCFS_MAXNAMLEN;
    199 	mp->mnt_flag |= MNT_LOCAL;
    200 	mp->mnt_data = pmnt;
    201 	vfs_getnewfsid(mp);
    202 
    203 	error = set_statvfs_info(path, UIO_USERSPACE, "procfs", UIO_SYSSPACE,
    204 	    mp->mnt_op->vfs_name, mp, l);
    205 	if (*data_len >= sizeof *args)
    206 		pmnt->pmnt_flags = args->flags;
    207 	else
    208 		pmnt->pmnt_flags = 0;
    209 
    210 	mp->mnt_iflag |= IMNT_MPSAFE | IMNT_SHRLOOKUP;
    211 	return error;
    212 }
    213 
    214 /*
    215  * unmount system call
    216  */
    217 int
    218 procfs_unmount(struct mount *mp, int mntflags)
    219 {
    220 	int error;
    221 	int flags = 0;
    222 
    223 	if (mntflags & MNT_FORCE)
    224 		flags |= FORCECLOSE;
    225 
    226 	if ((error = vflush(mp, 0, flags)) != 0)
    227 		return (error);
    228 
    229 	kmem_free(mp->mnt_data, sizeof(struct procfsmount));
    230 	mp->mnt_data = NULL;
    231 
    232 	return 0;
    233 }
    234 
    235 int
    236 procfs_root(struct mount *mp, int lktype, struct vnode **vpp)
    237 {
    238 	int error;
    239 
    240 	error = procfs_allocvp(mp, vpp, 0, PFSroot, -1);
    241 	if (error == 0) {
    242 		error = vn_lock(*vpp, lktype);
    243 		if (error != 0) {
    244 			vrele(*vpp);
    245 			*vpp = NULL;
    246 		}
    247 	}
    248 
    249 	return error;
    250 }
    251 
    252 /* ARGSUSED */
    253 int
    254 procfs_start(struct mount *mp, int flags)
    255 {
    256 
    257 	return (0);
    258 }
    259 
    260 /*
    261  * Get file system statistics.
    262  */
    263 int
    264 procfs_statvfs(struct mount *mp, struct statvfs *sbp)
    265 {
    266 
    267 	genfs_statvfs(mp, sbp);
    268 
    269 	sbp->f_bsize = PAGE_SIZE;
    270 	sbp->f_frsize = PAGE_SIZE;
    271 	sbp->f_iosize = PAGE_SIZE;
    272 	sbp->f_blocks = 1;
    273 	sbp->f_files = maxproc;					/* approx */
    274 	sbp->f_ffree = maxproc - atomic_load_relaxed(&nprocs);	/* approx */
    275 	sbp->f_favail = maxproc - atomic_load_relaxed(&nprocs);	/* approx */
    276 
    277 	return (0);
    278 }
    279 
    280 /*ARGSUSED*/
    281 int
    282 procfs_sync(
    283     struct mount *mp,
    284     int waitfor,
    285     kauth_cred_t uc)
    286 {
    287 
    288 	return (0);
    289 }
    290 
    291 /*ARGSUSED*/
    292 int
    293 procfs_vget(struct mount *mp, ino_t ino, int lktype,
    294     struct vnode **vpp)
    295 {
    296 	return (EOPNOTSUPP);
    297 }
    298 
    299 int
    300 procfs_loadvnode(struct mount *mp, struct vnode *vp,
    301     const void *key, size_t key_len, const void **new_key)
    302 {
    303 	int error;
    304 	struct pfskey pfskey;
    305 	struct pfsnode *pfs;
    306 
    307 	KASSERT(key_len == sizeof(pfskey));
    308 	memcpy(&pfskey, key, key_len);
    309 
    310 	pfs = kmem_alloc(sizeof(*pfs), KM_SLEEP);
    311 	pfs->pfs_pid = pfskey.pk_pid;
    312 	pfs->pfs_type = pfskey.pk_type;
    313 	pfs->pfs_fd = pfskey.pk_fd;
    314 	pfs->pfs_vnode = vp;
    315 	pfs->pfs_mount = mp;
    316 	pfs->pfs_flags = 0;
    317 	pfs->pfs_fileno =
    318 	    PROCFS_FILENO(pfs->pfs_pid, pfs->pfs_type, pfs->pfs_fd);
    319 	vp->v_tag = VT_PROCFS;
    320 	vp->v_op = procfs_vnodeop_p;
    321 	vp->v_data = pfs;
    322 
    323 	switch (pfs->pfs_type) {
    324 	case PFSroot:	/* /proc = dr-xr-xr-x */
    325 		vp->v_vflag |= VV_ROOT;
    326 		/*FALLTHROUGH*/
    327 	case PFSproc:	/* /proc/N = dr-xr-xr-x */
    328 		pfs->pfs_mode = S_IRUSR|S_IXUSR|S_IRGRP|S_IXGRP|S_IROTH|S_IXOTH;
    329 		vp->v_type = VDIR;
    330 		break;
    331 
    332 	case PFStask:	/* /proc/N/task = dr-xr-xr-x */
    333 		if (pfs->pfs_fd == -1) {
    334 			pfs->pfs_mode = S_IRUSR|S_IXUSR|S_IRGRP|S_IXGRP|
    335 			    S_IROTH|S_IXOTH;
    336 			vp->v_type = VDIR;
    337 			break;
    338 		}
    339 		/*FALLTHROUGH*/
    340 	case PFScurproc:	/* /proc/curproc = lr-xr-xr-x */
    341 	case PFSself:	/* /proc/self    = lr-xr-xr-x */
    342 	case PFScwd:	/* /proc/N/cwd = lr-xr-xr-x */
    343 	case PFSchroot:	/* /proc/N/chroot = lr-xr-xr-x */
    344 	case PFSexe:	/* /proc/N/exe = lr-xr-xr-x */
    345 		pfs->pfs_mode = S_IRUSR|S_IXUSR|S_IRGRP|S_IXGRP|S_IROTH|S_IXOTH;
    346 		vp->v_type = VLNK;
    347 		break;
    348 
    349 	case PFSfd:
    350 		if (pfs->pfs_fd == -1) {	/* /proc/N/fd = dr-x------ */
    351 			pfs->pfs_mode = S_IRUSR|S_IXUSR;
    352 			vp->v_type = VDIR;
    353 		} else {	/* /proc/N/fd/M = [ps-]rw------- */
    354 			file_t *fp;
    355 			vnode_t *vxp;
    356 			struct proc *p;
    357 
    358 			mutex_enter(&proc_lock);
    359 			p = procfs_proc_find(mp, pfs->pfs_pid);
    360 			mutex_exit(&proc_lock);
    361 			if (p == NULL) {
    362 				error = ENOENT;
    363 				goto bad;
    364 			}
    365 			KASSERT(rw_read_held(&p->p_reflock));
    366 			if ((fp = fd_getfile2(p, pfs->pfs_fd)) == NULL) {
    367 				error = EBADF;
    368 				goto bad;
    369 			}
    370 
    371 			pfs->pfs_mode = S_IRUSR|S_IWUSR;
    372 			switch (fp->f_type) {
    373 			case DTYPE_VNODE:
    374 				vxp = fp->f_vnode;
    375 
    376 				/*
    377 				 * We make symlinks for directories
    378 				 * to avoid cycles.
    379 				 */
    380 				if (vxp->v_type == VDIR ||
    381 				    procfs_proc_is_linux_compat())
    382 					goto symlink;
    383 				vp->v_type = vxp->v_type;
    384 				break;
    385 			case DTYPE_PIPE:
    386 				vp->v_type = VFIFO;
    387 				break;
    388 			case DTYPE_SOCKET:
    389 				vp->v_type = VSOCK;
    390 				break;
    391 			case DTYPE_KQUEUE:
    392 			case DTYPE_MISC:
    393 			case DTYPE_SEM:
    394 			symlink:
    395 				pfs->pfs_mode = S_IRUSR|S_IXUSR|S_IRGRP|
    396 				    S_IXGRP|S_IROTH|S_IXOTH;
    397 				vp->v_type = VLNK;
    398 				break;
    399 			default:
    400 				error = EOPNOTSUPP;
    401 				closef(fp);
    402 				goto bad;
    403 			}
    404 			closef(fp);
    405 		}
    406 		break;
    407 
    408 	case PFSfile:	/* /proc/N/file = -rw------- */
    409 	case PFSmem:	/* /proc/N/mem = -rw------- */
    410 	case PFSregs:	/* /proc/N/regs = -rw------- */
    411 	case PFSfpregs:	/* /proc/N/fpregs = -rw------- */
    412 		pfs->pfs_mode = S_IRUSR|S_IWUSR;
    413 		vp->v_type = VREG;
    414 		break;
    415 
    416 	case PFSnote:	/* /proc/N/note = --w------ */
    417 	case PFSnotepg:	/* /proc/N/notepg = --w------ */
    418 		pfs->pfs_mode = S_IWUSR;
    419 		vp->v_type = VREG;
    420 		break;
    421 
    422 	case PFSmap:		/* /proc/N/map = -r-------- */
    423 	case PFSmaps:		/* /proc/N/maps = -r-------- */
    424 	case PFSauxv:		/* /proc/N/auxv = -r-------- */
    425 	case PFSenviron:	/* /proc/N/environ = -r-------- */
    426 		pfs->pfs_mode = S_IRUSR;
    427 		vp->v_type = VREG;
    428 		break;
    429 
    430 	case PFSstatus:		/* /proc/N/status = -r--r--r-- */
    431 	case PFSstat:		/* /proc/N/stat = -r--r--r-- */
    432 	case PFScmdline:	/* /proc/N/cmdline = -r--r--r-- */
    433 	case PFSemul:		/* /proc/N/emul = -r--r--r-- */
    434 	case PFSmeminfo:	/* /proc/meminfo = -r--r--r-- */
    435 	case PFScpustat:	/* /proc/stat = -r--r--r-- */
    436 	case PFSdevices:	/* /proc/devices = -r--r--r-- */
    437 	case PFScpuinfo:	/* /proc/cpuinfo = -r--r--r-- */
    438 	case PFSuptime:		/* /proc/uptime = -r--r--r-- */
    439 	case PFSmounts:		/* /proc/mounts = -r--r--r-- */
    440 	case PFSloadavg:	/* /proc/loadavg = -r--r--r-- */
    441 	case PFSstatm:		/* /proc/N/statm = -r--r--r-- */
    442 	case PFSversion:	/* /proc/version = -r--r--r-- */
    443 	case PFSlimit:		/* /proc/N/limit = -r--r--r-- */
    444 	case PFSlimits:		/* /proc/N/limits = -r--r--r-- */
    445 		pfs->pfs_mode = S_IRUSR|S_IRGRP|S_IROTH;
    446 		vp->v_type = VREG;
    447 		break;
    448 
    449 	case PFSsys:	/* /proc/sys = dr-xr-xr-x */
    450 	case PFSsysfs:	/* /proc/sys/fs = dr-xr-xr-x */
    451 	case PFSmqueue:	/* /proc/sys/fs/mqueue = dr-xr-xr-x */
    452 	case PFSsysvipc:/* /proc/sysvipc = dr-xr-xr-x */
    453 		if (pfs->pfs_fd == -1) {
    454 			pfs->pfs_mode = S_IRUSR|S_IXUSR|S_IRGRP|S_IXGRP|
    455 			    S_IROTH|S_IXOTH;
    456 			vp->v_type = VDIR;
    457 			break;
    458 		}
    459 		/*FALLTHROUGH*/
    460 	case PFSmq_msg_def:	/* /proc/sys/fs/mqueue/msg_default = -r--r--r-- */
    461 	case PFSmq_msg_max:	/* /proc/sys/fs/mqueue/msg_max = -r--r--r-- */
    462 	case PFSmq_siz_def:	/* /proc/sys/fs/mqueue/msgsize_default = -r--r--r-- */
    463 	case PFSmq_siz_max:	/* /proc/sys/fs/mqueue/msgsize_max = -r--r--r-- */
    464 	case PFSmq_qmax:	/* /proc/sys/fs/mqueue/queues_max = -r--r--r-- */
    465 	case PFSsysvipc_msg:	/* /proc/sysvipc/msg = -r--r--r-- */
    466 	case PFSsysvipc_sem:	/* /proc/sysvipc/sem = -r--r--r-- */
    467 	case PFSsysvipc_shm:	/* /proc/sysvipc/shm = -r--r--r-- */
    468 		pfs->pfs_mode = S_IRUSR|S_IRGRP|S_IROTH;
    469 		vp->v_type = VREG;
    470 		break;
    471 
    472 #ifdef __HAVE_PROCFS_MACHDEP
    473 	PROCFS_MACHDEP_NODETYPE_CASES
    474 		procfs_machdep_allocvp(vp);
    475 		break;
    476 #endif
    477 
    478 	default:
    479 		panic("procfs_allocvp");
    480 	}
    481 
    482 	mutex_enter(&procfs_hashlock);
    483 	LIST_INSERT_HEAD(procfs_hashhead(pfs->pfs_pid), pfs, pfs_hash);
    484 	mutex_exit(&procfs_hashlock);
    485 
    486 	uvm_vnp_setsize(vp, 0);
    487 	*new_key = &pfs->pfs_key;
    488 
    489 	return 0;
    490 
    491 bad:
    492 	vp->v_tag =VT_NON;
    493 	vp->v_type = VNON;
    494 	vp->v_op = NULL;
    495 	vp->v_data = NULL;
    496 	kmem_free(pfs, sizeof(*pfs));
    497 	return error;
    498 }
    499 
    500 void
    501 procfs_init(void)
    502 {
    503 
    504 }
    505 
    506 void
    507 procfs_reinit(void)
    508 {
    509 
    510 }
    511 
    512 void
    513 procfs_done(void)
    514 {
    515 
    516 }
    517 
    518 extern const struct vnodeopv_desc procfs_vnodeop_opv_desc;
    519 
    520 const struct vnodeopv_desc * const procfs_vnodeopv_descs[] = {
    521 	&procfs_vnodeop_opv_desc,
    522 	NULL,
    523 };
    524 
    525 struct vfsops procfs_vfsops = {
    526 	.vfs_name = MOUNT_PROCFS,
    527 	.vfs_min_mount_data = sizeof (struct procfs_args),
    528 	.vfs_mount = procfs_mount,
    529 	.vfs_start = procfs_start,
    530 	.vfs_unmount = procfs_unmount,
    531 	.vfs_root = procfs_root,
    532 	.vfs_quotactl = (void *)eopnotsupp,
    533 	.vfs_statvfs = procfs_statvfs,
    534 	.vfs_sync = procfs_sync,
    535 	.vfs_vget = procfs_vget,
    536 	.vfs_loadvnode = procfs_loadvnode,
    537 	.vfs_fhtovp = (void *)eopnotsupp,
    538 	.vfs_vptofh = (void *)eopnotsupp,
    539 	.vfs_init = procfs_init,
    540 	.vfs_reinit = procfs_reinit,
    541 	.vfs_done = procfs_done,
    542 	.vfs_snapshot = (void *)eopnotsupp,
    543 	.vfs_extattrctl = vfs_stdextattrctl,
    544 	.vfs_suspendctl = genfs_suspendctl,
    545 	.vfs_renamelock_enter = genfs_renamelock_enter,
    546 	.vfs_renamelock_exit = genfs_renamelock_exit,
    547 	.vfs_fsync = (void *)eopnotsupp,
    548 	.vfs_opv_descs = procfs_vnodeopv_descs
    549 };
    550 
    551 static void
    552 procfs_exechook_cb(struct proc *p, void *arg)
    553 {
    554 	struct hashhead *head;
    555 	struct pfsnode *pfs;
    556 	struct mount *mp;
    557 	struct pfskey key;
    558 	struct vnode *vp;
    559 	int error;
    560 
    561 	if (arg == PROCFS_EXEC_HOOK && !(p->p_flag & PK_SUGID))
    562 		return;
    563 
    564 	head = procfs_hashhead(p->p_pid);
    565 
    566 again:
    567 	mutex_enter(&procfs_hashlock);
    568 	LIST_FOREACH(pfs, head, pfs_hash) {
    569 		if (pfs->pfs_pid != p->p_pid)
    570 			continue;
    571 		mp = pfs->pfs_mount;
    572 		key = pfs->pfs_key;
    573 		vfs_ref(mp);
    574 		mutex_exit(&procfs_hashlock);
    575 
    576 		error = vcache_get(mp, &key, sizeof(key), &vp);
    577 		vfs_rele(mp);
    578 		if (error != 0)
    579 			goto again;
    580 		if (vrecycle(vp))
    581 			goto again;
    582 		do {
    583 			error = vfs_suspend(mp, 0);
    584 		} while (error == EINTR || error == ERESTART);
    585 		vgone(vp);
    586 		if (error == 0)
    587 			vfs_resume(mp);
    588 		goto again;
    589 	}
    590 	mutex_exit(&procfs_hashlock);
    591 }
    592 
    593 static int
    594 procfs_listener_cb(kauth_cred_t cred, kauth_action_t action, void *cookie,
    595     void *arg0, void *arg1, void *arg2, void *arg3)
    596 {
    597 	struct proc *p;
    598 	struct pfsnode *pfs;
    599 	int result;
    600 
    601 	result = KAUTH_RESULT_DEFER;
    602 	p = arg0;
    603 	pfs = arg1;
    604 
    605 	if (action != KAUTH_PROCESS_PROCFS)
    606 		return result;
    607 
    608 	switch (pfs->pfs_type) {
    609 	case PFSregs:
    610 	case PFSfpregs:
    611 	case PFSmem:
    612 		if (kauth_cred_getuid(cred) != kauth_cred_getuid(p->p_cred) ||
    613 		    ISSET(p->p_flag, PK_SUGID))
    614 			break;
    615 
    616 		/*FALLTHROUGH*/
    617 	default:
    618 		result = KAUTH_RESULT_ALLOW;
    619 		break;
    620 	}
    621 
    622 	return result;
    623 }
    624 
    625 SYSCTL_SETUP(procfs_sysctl_setup, "procfs sysctl")
    626 {
    627 
    628 	sysctl_createv(clog, 0, NULL, NULL,
    629 		       CTLFLAG_PERMANENT,
    630 		       CTLTYPE_NODE, "procfs",
    631 		       SYSCTL_DESCR("Process file system"),
    632 		       NULL, 0, NULL, 0,
    633 		       CTL_VFS, 12, CTL_EOL);
    634 	/*
    635 	 * XXX the "12" above could be dynamic, thereby eliminating
    636 	 * one more instance of the "number to vfs" mapping problem,
    637 	 * but "12" is the order as taken from sys/mount.h
    638 	 */
    639 }
    640 
    641 static int
    642 procfs_modcmd(modcmd_t cmd, void *arg)
    643 {
    644 	int error;
    645 
    646 	switch (cmd) {
    647 	case MODULE_CMD_INIT:
    648 		error = vfs_attach(&procfs_vfsops);
    649 		if (error != 0)
    650 			break;
    651 
    652 		procfs_listener = kauth_listen_scope(KAUTH_SCOPE_PROCESS,
    653 		    procfs_listener_cb, NULL);
    654 
    655 		procfs_exechook = exechook_establish(procfs_exechook_cb,
    656 		    PROCFS_EXEC_HOOK);
    657 		procfs_exithook = exithook_establish(procfs_exechook_cb,
    658 		    PROCFS_EXIT_HOOK);
    659 
    660 		mutex_init(&procfs_hashlock, MUTEX_DEFAULT, IPL_NONE);
    661 		procfs_hashtab = hashinit(PROCFS_HASHSIZE, HASH_LIST, true,
    662 		    &procfs_hashmask);
    663 
    664 		break;
    665 	case MODULE_CMD_FINI:
    666 		error = vfs_detach(&procfs_vfsops);
    667 		if (error != 0)
    668 			break;
    669 		kauth_unlisten_scope(procfs_listener);
    670 		exechook_disestablish(procfs_exechook);
    671 		exithook_disestablish(procfs_exithook);
    672 		mutex_destroy(&procfs_hashlock);
    673 		hashdone(procfs_hashtab, HASH_LIST, procfs_hashmask);
    674 		break;
    675 	default:
    676 		error = ENOTTY;
    677 		break;
    678 	}
    679 
    680 	return (error);
    681 }
    682