Home | History | Annotate | Line # | Download | only in procfs
procfs_subr.c revision 1.90
      1  1.90        ad /*	$NetBSD: procfs_subr.c,v 1.90 2008/05/05 17:11:17 ad Exp $	*/
      2  1.75        ad 
      3  1.75        ad /*-
      4  1.86        ad  * Copyright (c) 2006, 2007, 2008 The NetBSD Foundation, Inc.
      5  1.75        ad  * All rights reserved.
      6  1.75        ad  *
      7  1.75        ad  * This code is derived from software contributed to The NetBSD Foundation
      8  1.75        ad  * by Andrew Doran.
      9  1.75        ad  *
     10  1.75        ad  * Redistribution and use in source and binary forms, with or without
     11  1.75        ad  * modification, are permitted provided that the following conditions
     12  1.75        ad  * are met:
     13  1.75        ad  * 1. Redistributions of source code must retain the above copyright
     14  1.75        ad  *    notice, this list of conditions and the following disclaimer.
     15  1.75        ad  * 2. Redistributions in binary form must reproduce the above copyright
     16  1.75        ad  *    notice, this list of conditions and the following disclaimer in the
     17  1.75        ad  *    documentation and/or other materials provided with the distribution.
     18  1.75        ad  *
     19  1.75        ad  * THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS
     20  1.75        ad  * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED
     21  1.75        ad  * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
     22  1.75        ad  * PURPOSE ARE DISCLAIMED.  IN NO EVENT SHALL THE FOUNDATION OR CONTRIBUTORS
     23  1.75        ad  * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
     24  1.75        ad  * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
     25  1.75        ad  * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
     26  1.75        ad  * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
     27  1.75        ad  * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
     28  1.75        ad  * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
     29  1.75        ad  * POSSIBILITY OF SUCH DAMAGE.
     30  1.75        ad  */
     31  1.57       agc 
     32  1.57       agc /*
     33  1.57       agc  * Copyright (c) 1993
     34  1.57       agc  *	The Regents of the University of California.  All rights reserved.
     35  1.57       agc  *
     36  1.57       agc  * This code is derived from software contributed to Berkeley by
     37  1.57       agc  * Jan-Simon Pendry.
     38  1.57       agc  *
     39  1.57       agc  * Redistribution and use in source and binary forms, with or without
     40  1.57       agc  * modification, are permitted provided that the following conditions
     41  1.57       agc  * are met:
     42  1.57       agc  * 1. Redistributions of source code must retain the above copyright
     43  1.57       agc  *    notice, this list of conditions and the following disclaimer.
     44  1.57       agc  * 2. Redistributions in binary form must reproduce the above copyright
     45  1.57       agc  *    notice, this list of conditions and the following disclaimer in the
     46  1.57       agc  *    documentation and/or other materials provided with the distribution.
     47  1.57       agc  * 3. Neither the name of the University nor the names of its contributors
     48  1.57       agc  *    may be used to endorse or promote products derived from this software
     49  1.57       agc  *    without specific prior written permission.
     50  1.57       agc  *
     51  1.57       agc  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
     52  1.57       agc  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
     53  1.57       agc  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
     54  1.57       agc  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
     55  1.57       agc  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
     56  1.57       agc  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
     57  1.57       agc  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
     58  1.57       agc  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
     59  1.57       agc  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
     60  1.57       agc  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
     61  1.57       agc  * SUCH DAMAGE.
     62  1.57       agc  *
     63  1.57       agc  *	@(#)procfs_subr.c	8.6 (Berkeley) 5/14/95
     64  1.57       agc  */
     65  1.13       cgd 
     66   1.1        pk /*
     67  1.20   thorpej  * Copyright (c) 1994 Christopher G. Demetriou.  All rights reserved.
     68   1.5       cgd  * Copyright (c) 1993 Jan-Simon Pendry
     69   1.2        pk  *
     70   1.5       cgd  * This code is derived from software contributed to Berkeley by
     71   1.5       cgd  * Jan-Simon Pendry.
     72   1.5       cgd  *
     73   1.2        pk  * Redistribution and use in source and binary forms, with or without
     74   1.2        pk  * modification, are permitted provided that the following conditions
     75   1.2        pk  * are met:
     76   1.2        pk  * 1. Redistributions of source code must retain the above copyright
     77   1.2        pk  *    notice, this list of conditions and the following disclaimer.
     78   1.2        pk  * 2. Redistributions in binary form must reproduce the above copyright
     79   1.2        pk  *    notice, this list of conditions and the following disclaimer in the
     80   1.2        pk  *    documentation and/or other materials provided with the distribution.
     81   1.2        pk  * 3. All advertising materials mentioning features or use of this software
     82   1.2        pk  *    must display the following acknowledgement:
     83   1.5       cgd  *	This product includes software developed by the University of
     84   1.5       cgd  *	California, Berkeley and its contributors.
     85   1.5       cgd  * 4. Neither the name of the University nor the names of its contributors
     86   1.5       cgd  *    may be used to endorse or promote products derived from this software
     87   1.5       cgd  *    without specific prior written permission.
     88   1.5       cgd  *
     89   1.5       cgd  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
     90   1.5       cgd  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
     91   1.5       cgd  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
     92   1.5       cgd  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
     93   1.5       cgd  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
     94   1.5       cgd  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
     95   1.5       cgd  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
     96   1.5       cgd  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
     97   1.5       cgd  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
     98   1.5       cgd  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
     99   1.5       cgd  * SUCH DAMAGE.
    100   1.2        pk  *
    101  1.23      fvdl  *	@(#)procfs_subr.c	8.6 (Berkeley) 5/14/95
    102   1.1        pk  */
    103  1.39     lukem 
    104  1.39     lukem #include <sys/cdefs.h>
    105  1.90        ad __KERNEL_RCSID(0, "$NetBSD: procfs_subr.c,v 1.90 2008/05/05 17:11:17 ad Exp $");
    106   1.5       cgd 
    107   1.4   mycroft #include <sys/param.h>
    108   1.4   mycroft #include <sys/systm.h>
    109   1.4   mycroft #include <sys/time.h>
    110   1.4   mycroft #include <sys/kernel.h>
    111   1.4   mycroft #include <sys/proc.h>
    112   1.4   mycroft #include <sys/vnode.h>
    113  1.11   mycroft #include <sys/malloc.h>
    114  1.18   mycroft #include <sys/stat.h>
    115  1.42  christos #include <sys/file.h>
    116  1.42  christos #include <sys/filedesc.h>
    117  1.73      elad #include <sys/kauth.h>
    118  1.18   mycroft 
    119   1.5       cgd #include <miscfs/procfs/procfs.h>
    120   1.1        pk 
    121  1.65   xtraeme void procfs_hashins(struct pfsnode *);
    122  1.65   xtraeme void procfs_hashrem(struct pfsnode *);
    123  1.82        ad struct vnode *procfs_hashget(pid_t, pfstype, int, struct mount *, int);
    124  1.29      fvdl 
    125  1.29      fvdl LIST_HEAD(pfs_hashhead, pfsnode) *pfs_hashtbl;
    126  1.30      fvdl u_long	pfs_ihash;	/* size of hash table - 1 */
    127  1.38       chs #define PFSPIDHASH(pid)	((pid) & pfs_ihash)
    128  1.29      fvdl 
    129  1.76        ad kmutex_t pfs_hashlock;
    130  1.76        ad kmutex_t pfs_ihash_lock;
    131   1.1        pk 
    132  1.20   thorpej #define	ISSET(t, f)	((t) & (f))
    133  1.20   thorpej 
    134   1.1        pk /*
    135   1.5       cgd  * allocate a pfsnode/vnode pair.  the vnode is
    136  1.27  wrstuden  * referenced, and locked.
    137   1.5       cgd  *
    138   1.5       cgd  * the pid, pfs_type, and mount point uniquely
    139   1.5       cgd  * identify a pfsnode.  the mount point is needed
    140   1.5       cgd  * because someone might mount this filesystem
    141   1.5       cgd  * twice.
    142   1.5       cgd  *
    143   1.5       cgd  * all pfsnodes are maintained on a singly-linked
    144   1.5       cgd  * list.  new nodes are only allocated when they cannot
    145   1.5       cgd  * be found on this list.  entries on the list are
    146   1.5       cgd  * removed when the vfs reclaim entry is called.
    147   1.5       cgd  *
    148   1.5       cgd  * a single lock is kept for the entire list.  this is
    149   1.5       cgd  * needed because the getnewvnode() function can block
    150   1.5       cgd  * waiting for a vnode to become free, in which case there
    151   1.5       cgd  * may be more than one process trying to get the same
    152   1.5       cgd  * vnode.  this lock is only taken if we are going to
    153   1.5       cgd  * call getnewvnode, since the kernel itself is single-threaded.
    154   1.5       cgd  *
    155   1.5       cgd  * if an entry is found on the list, then call vget() to
    156   1.5       cgd  * take a reference.  this is done because there may be
    157   1.5       cgd  * zero references to it and so it needs to removed from
    158   1.5       cgd  * the vnode free list.
    159   1.1        pk  */
    160  1.11   mycroft int
    161  1.75        ad procfs_allocvp(mp, vpp, pid, pfs_type, fd, p)
    162   1.5       cgd 	struct mount *mp;
    163   1.5       cgd 	struct vnode **vpp;
    164  1.42  christos 	pid_t pid;
    165   1.5       cgd 	pfstype pfs_type;
    166  1.42  christos 	int fd;
    167  1.75        ad 	struct proc *p;
    168   1.1        pk {
    169  1.12   mycroft 	struct pfsnode *pfs;
    170  1.12   mycroft 	struct vnode *vp;
    171   1.5       cgd 	int error;
    172   1.5       cgd 
    173  1.82        ad  retry:
    174  1.82        ad 	*vpp = procfs_hashget(pid, pfs_type, fd, mp, LK_EXCLUSIVE);
    175  1.82        ad 	if (*vpp != NULL)
    176  1.76        ad 		return (0);
    177   1.1        pk 
    178  1.44  jdolecek 	if ((error = getnewvnode(VT_PROCFS, mp, procfs_vnodeop_p, &vp)) != 0) {
    179  1.29      fvdl 		*vpp = NULL;
    180  1.29      fvdl 		return (error);
    181   1.5       cgd 	}
    182  1.76        ad 	MALLOC(pfs, void *, sizeof(struct pfsnode), M_TEMP, M_WAITOK);
    183  1.76        ad 
    184  1.76        ad 	mutex_enter(&pfs_hashlock);
    185  1.82        ad 	if ((*vpp = procfs_hashget(pid, pfs_type, fd, mp, 0)) != NULL) {
    186  1.76        ad 		mutex_exit(&pfs_hashlock);
    187  1.76        ad 		ungetnewvnode(vp);
    188  1.76        ad 		FREE(pfs, M_TEMP);
    189  1.82        ad 		goto retry;
    190  1.76        ad 	}
    191   1.5       cgd 
    192  1.11   mycroft 	vp->v_data = pfs;
    193  1.42  christos 	pfs->pfs_pid = pid;
    194   1.5       cgd 	pfs->pfs_type = pfs_type;
    195  1.11   mycroft 	pfs->pfs_vnode = vp;
    196   1.5       cgd 	pfs->pfs_flags = 0;
    197  1.42  christos 	pfs->pfs_fileno = PROCFS_FILENO(pid, pfs_type, fd);
    198  1.42  christos 	pfs->pfs_fd = fd;
    199   1.5       cgd 
    200   1.5       cgd 	switch (pfs_type) {
    201  1.58     darcy 	case PFSroot:	/* /proc = dr-xr-xr-x */
    202  1.82        ad 		vp->v_vflag |= VV_ROOT;
    203  1.74  christos 		/*FALLTHROUGH*/
    204  1.74  christos 	case PFSproc:	/* /proc/N = dr-xr-xr-x */
    205  1.17   mycroft 		pfs->pfs_mode = S_IRUSR|S_IXUSR|S_IRGRP|S_IXGRP|S_IROTH|S_IXOTH;
    206  1.11   mycroft 		vp->v_type = VDIR;
    207  1.11   mycroft 		break;
    208  1.11   mycroft 
    209  1.58     darcy 	case PFScurproc:	/* /proc/curproc = lr-xr-xr-x */
    210  1.58     darcy 	case PFSself:	/* /proc/self    = lr-xr-xr-x */
    211  1.66    atatat 	case PFScwd:	/* /proc/N/cwd = lr-xr-xr-x */
    212  1.66    atatat 	case PFSchroot:	/* /proc/N/chroot = lr-xr-xr-x */
    213  1.70  christos 	case PFSexe:	/* /proc/N/exe = lr-xr-xr-x */
    214  1.22   mycroft 		pfs->pfs_mode = S_IRUSR|S_IXUSR|S_IRGRP|S_IXGRP|S_IROTH|S_IXOTH;
    215  1.11   mycroft 		vp->v_type = VLNK;
    216   1.5       cgd 		break;
    217   1.5       cgd 
    218  1.58     darcy 	case PFSfd:
    219  1.42  christos 		if (fd == -1) {	/* /proc/N/fd = dr-xr-xr-x */
    220  1.52  christos 			pfs->pfs_mode = S_IRUSR|S_IXUSR;
    221  1.42  christos 			vp->v_type = VDIR;
    222  1.42  christos 		} else {	/* /proc/N/fd/M = [ps-]rw------- */
    223  1.86        ad 			file_t *fp;
    224  1.86        ad 			vnode_t *vxp;
    225  1.44  jdolecek 
    226  1.86        ad 			if ((fp = fd_getfile2(p, pfs->pfs_fd)) == NULL) {
    227  1.75        ad 				error = EBADF;
    228  1.44  jdolecek 				goto bad;
    229  1.75        ad 			}
    230  1.44  jdolecek 
    231  1.42  christos 			pfs->pfs_mode = S_IRUSR|S_IWUSR;
    232  1.42  christos 			switch (fp->f_type) {
    233  1.42  christos 			case DTYPE_VNODE:
    234  1.86        ad 				vxp = fp->f_data;
    235  1.50  jdolecek 
    236  1.51  christos 				/*
    237  1.63     perry 				 * We make symlinks for directories
    238  1.51  christos 				 * to avoid cycles.
    239  1.51  christos 				 */
    240  1.51  christos 				if (vxp->v_type == VDIR)
    241  1.51  christos 					goto symlink;
    242  1.42  christos 				vp->v_type = vxp->v_type;
    243  1.42  christos 				break;
    244  1.42  christos 			case DTYPE_PIPE:
    245  1.42  christos 				vp->v_type = VFIFO;
    246  1.42  christos 				break;
    247  1.42  christos 			case DTYPE_SOCKET:
    248  1.42  christos 				vp->v_type = VSOCK;
    249  1.51  christos 				break;
    250  1.51  christos 			case DTYPE_KQUEUE:
    251  1.51  christos 			case DTYPE_MISC:
    252  1.51  christos 			symlink:
    253  1.51  christos 				pfs->pfs_mode = S_IRUSR|S_IXUSR|S_IRGRP|
    254  1.51  christos 				    S_IXGRP|S_IROTH|S_IXOTH;
    255  1.51  christos 				vp->v_type = VLNK;
    256  1.42  christos 				break;
    257  1.45  jdolecek 			default:
    258  1.44  jdolecek 				error = EOPNOTSUPP;
    259  1.86        ad 				closef(fp);
    260  1.44  jdolecek 				goto bad;
    261  1.42  christos 			}
    262  1.86        ad 			closef(fp);
    263  1.42  christos 		}
    264   1.5       cgd 		break;
    265   1.5       cgd 
    266  1.58     darcy 	case PFSfile:	/* /proc/N/file = -rw------- */
    267  1.58     darcy 	case PFSmem:	/* /proc/N/mem = -rw------- */
    268  1.58     darcy 	case PFSregs:	/* /proc/N/regs = -rw------- */
    269  1.58     darcy 	case PFSfpregs:	/* /proc/N/fpregs = -rw------- */
    270  1.17   mycroft 		pfs->pfs_mode = S_IRUSR|S_IWUSR;
    271   1.9       cgd 		vp->v_type = VREG;
    272   1.9       cgd 		break;
    273   1.9       cgd 
    274  1.58     darcy 	case PFSctl:	/* /proc/N/ctl = --w------ */
    275  1.58     darcy 	case PFSnote:	/* /proc/N/note = --w------ */
    276  1.58     darcy 	case PFSnotepg:	/* /proc/N/notepg = --w------ */
    277  1.17   mycroft 		pfs->pfs_mode = S_IWUSR;
    278   1.6        ws 		vp->v_type = VREG;
    279   1.5       cgd 		break;
    280   1.5       cgd 
    281  1.58     darcy 	case PFSmap:	/* /proc/N/map = -r--r--r-- */
    282  1.58     darcy 	case PFSmaps:	/* /proc/N/maps = -r--r--r-- */
    283  1.58     darcy 	case PFSstatus:	/* /proc/N/status = -r--r--r-- */
    284  1.58     darcy 	case PFSstat:	/* /proc/N/stat = -r--r--r-- */
    285  1.58     darcy 	case PFScmdline:	/* /proc/N/cmdline = -r--r--r-- */
    286  1.71  christos 	case PFSemul:	/* /proc/N/emul = -r--r--r-- */
    287  1.58     darcy 	case PFSmeminfo:	/* /proc/meminfo = -r--r--r-- */
    288  1.80       agc 	case PFScpustat:	/* /proc/stat = -r--r--r-- */
    289  1.69      manu 	case PFSdevices:	/* /proc/devices = -r--r--r-- */
    290  1.58     darcy 	case PFScpuinfo:	/* /proc/cpuinfo = -r--r--r-- */
    291  1.58     darcy 	case PFSuptime:	/* /proc/uptime = -r--r--r-- */
    292  1.62  jdolecek 	case PFSmounts:	/* /proc/mounts = -r--r--r-- */
    293  1.80       agc 	case PFSloadavg:	/* /proc/loadavg = -r--r--r-- */
    294  1.80       agc 	case PFSstatm:	/* /proc/N/statm = -r--r--r-- */
    295  1.17   mycroft 		pfs->pfs_mode = S_IRUSR|S_IRGRP|S_IROTH;
    296   1.6        ws 		vp->v_type = VREG;
    297   1.5       cgd 		break;
    298   1.5       cgd 
    299  1.40   thorpej #ifdef __HAVE_PROCFS_MACHDEP
    300  1.40   thorpej 	PROCFS_MACHDEP_NODETYPE_CASES
    301  1.40   thorpej 		procfs_machdep_allocvp(vp);
    302  1.40   thorpej 		break;
    303  1.40   thorpej #endif
    304  1.40   thorpej 
    305   1.5       cgd 	default:
    306  1.11   mycroft 		panic("procfs_allocvp");
    307   1.5       cgd 	}
    308  1.27  wrstuden 
    309  1.29      fvdl 	procfs_hashins(pfs);
    310  1.34       chs 	uvm_vnp_setsize(vp, 0);
    311  1.76        ad 	mutex_exit(&pfs_hashlock);
    312   1.1        pk 
    313  1.44  jdolecek 	*vpp = vp;
    314  1.44  jdolecek 	return (0);
    315  1.44  jdolecek 
    316  1.44  jdolecek  bad:
    317  1.76        ad 	mutex_exit(&pfs_hashlock);
    318  1.44  jdolecek 	FREE(pfs, M_TEMP);
    319  1.82        ad 	vp->v_data = NULL;
    320  1.44  jdolecek 	ungetnewvnode(vp);
    321   1.5       cgd 	return (error);
    322   1.1        pk }
    323   1.1        pk 
    324  1.11   mycroft int
    325   1.5       cgd procfs_freevp(vp)
    326   1.5       cgd 	struct vnode *vp;
    327   1.1        pk {
    328   1.5       cgd 	struct pfsnode *pfs = VTOPFS(vp);
    329   1.5       cgd 
    330  1.29      fvdl 	procfs_hashrem(pfs);
    331   1.1        pk 
    332  1.11   mycroft 	FREE(vp->v_data, M_TEMP);
    333  1.11   mycroft 	vp->v_data = 0;
    334   1.5       cgd 	return (0);
    335   1.1        pk }
    336   1.1        pk 
    337  1.11   mycroft int
    338  1.15  christos procfs_rw(v)
    339  1.15  christos 	void *v;
    340   1.1        pk {
    341  1.15  christos 	struct vop_read_args *ap = v;
    342  1.11   mycroft 	struct vnode *vp = ap->a_vp;
    343  1.11   mycroft 	struct uio *uio = ap->a_uio;
    344  1.67  christos 	struct lwp *curl;
    345  1.67  christos 	struct lwp *l;
    346  1.56      fvdl 	struct pfsnode *pfs = VTOPFS(vp);
    347   1.5       cgd 	struct proc *p;
    348  1.73      elad 	int error;
    349   1.5       cgd 
    350  1.59  christos 	if (uio->uio_offset < 0)
    351  1.59  christos 		return EINVAL;
    352  1.73      elad 
    353  1.75        ad 	if ((error = procfs_proc_lock(pfs->pfs_pid, &p, ESRCH)) != 0)
    354  1.75        ad 		return error;
    355  1.73      elad 
    356  1.73      elad 	curl = curlwp;
    357  1.73      elad 
    358  1.59  christos 	/*
    359  1.59  christos 	 * Do not allow init to be modified while in secure mode; it
    360  1.59  christos 	 * could be duped into changing the security level.
    361  1.59  christos 	 */
    362  1.84      elad #define	M2K(m)	((m) == UIO_READ ? KAUTH_REQ_PROCESS_PROCFS_READ : \
    363  1.84      elad 		 KAUTH_REQ_PROCESS_PROCFS_WRITE)
    364  1.88        ad 	mutex_enter(p->p_lock);
    365  1.84      elad 	error = kauth_authorize_process(curl->l_cred, KAUTH_PROCESS_PROCFS,
    366  1.73      elad 	    p, pfs, KAUTH_ARG(M2K(uio->uio_rw)), NULL);
    367  1.88        ad 	mutex_exit(p->p_lock);
    368  1.75        ad 	if (error) {
    369  1.75        ad 		procfs_proc_unlock(p);
    370  1.73      elad 		return (error);
    371  1.75        ad 	}
    372  1.73      elad #undef	M2K
    373  1.67  christos 
    374  1.88        ad 	mutex_enter(p->p_lock);
    375  1.75        ad 	l = proc_representative_lwp(p, NULL, 1);
    376  1.75        ad 	lwp_addref(l);
    377  1.88        ad 	mutex_exit(p->p_lock);
    378  1.63     perry 
    379  1.19   mycroft 	switch (pfs->pfs_type) {
    380  1.58     darcy 	case PFSnote:
    381  1.58     darcy 	case PFSnotepg:
    382  1.75        ad 		error = procfs_donote(curl, p, pfs, uio);
    383  1.75        ad 		break;
    384   1.5       cgd 
    385  1.58     darcy 	case PFSregs:
    386  1.75        ad 		error = procfs_doregs(curl, l, pfs, uio);
    387  1.75        ad 		break;
    388   1.9       cgd 
    389  1.58     darcy 	case PFSfpregs:
    390  1.75        ad 		error = procfs_dofpregs(curl, l, pfs, uio);
    391  1.75        ad 		break;
    392   1.5       cgd 
    393  1.58     darcy 	case PFSctl:
    394  1.75        ad 		error = procfs_doctl(curl, l, pfs, uio);
    395  1.75        ad 		break;
    396   1.5       cgd 
    397  1.58     darcy 	case PFSstatus:
    398  1.75        ad 		error = procfs_dostatus(curl, l, pfs, uio);
    399  1.75        ad 		break;
    400  1.53  christos 
    401  1.58     darcy 	case PFSstat:
    402  1.75        ad 		error = procfs_do_pid_stat(curl, l, pfs, uio);
    403  1.75        ad 		break;
    404  1.25   msaitoh 
    405  1.58     darcy 	case PFSmap:
    406  1.75        ad 		error = procfs_domap(curl, p, pfs, uio, 0);
    407  1.75        ad 		break;
    408  1.37      fvdl 
    409  1.58     darcy 	case PFSmaps:
    410  1.75        ad 		error = procfs_domap(curl, p, pfs, uio, 1);
    411  1.75        ad 		break;
    412   1.1        pk 
    413  1.58     darcy 	case PFSmem:
    414  1.75        ad 		error = procfs_domem(curl, l, pfs, uio);
    415  1.75        ad 		break;
    416  1.26  christos 
    417  1.58     darcy 	case PFScmdline:
    418  1.75        ad 		error = procfs_docmdline(curl, p, pfs, uio);
    419  1.75        ad 		break;
    420  1.35      fvdl 
    421  1.58     darcy 	case PFSmeminfo:
    422  1.75        ad 		error = procfs_domeminfo(curl, p, pfs, uio);
    423  1.75        ad 		break;
    424  1.40   thorpej 
    425  1.69      manu 	case PFSdevices:
    426  1.75        ad 		error = procfs_dodevices(curl, p, pfs, uio);
    427  1.75        ad 		break;
    428  1.69      manu 
    429  1.58     darcy 	case PFScpuinfo:
    430  1.75        ad 		error = procfs_docpuinfo(curl, p, pfs, uio);
    431  1.75        ad 		break;
    432  1.40   thorpej 
    433  1.80       agc 	case PFScpustat:
    434  1.80       agc 		error = procfs_docpustat(curl, p, pfs, uio);
    435  1.80       agc 		break;
    436  1.80       agc 
    437  1.80       agc 	case PFSloadavg:
    438  1.80       agc 		error = procfs_doloadavg(curl, p, pfs, uio);
    439  1.80       agc 		break;
    440  1.80       agc 
    441  1.80       agc 	case PFSstatm:
    442  1.80       agc 		error = procfs_do_pid_statm(curl, l, pfs, uio);
    443  1.80       agc 		break;
    444  1.80       agc 
    445  1.58     darcy 	case PFSfd:
    446  1.75        ad 		error = procfs_dofd(curl, p, pfs, uio);
    447  1.75        ad 		break;
    448  1.46       jrf 
    449  1.58     darcy 	case PFSuptime:
    450  1.75        ad 		error = procfs_douptime(curl, p, pfs, uio);
    451  1.75        ad 		break;
    452  1.42  christos 
    453  1.62  jdolecek 	case PFSmounts:
    454  1.75        ad 		error = procfs_domounts(curl, p, pfs, uio);
    455  1.75        ad 		break;
    456  1.62  jdolecek 
    457  1.71  christos 	case PFSemul:
    458  1.75        ad 		error = procfs_doemul(curl, p, pfs, uio);
    459  1.75        ad 		break;
    460  1.71  christos 
    461  1.40   thorpej #ifdef __HAVE_PROCFS_MACHDEP
    462  1.40   thorpej 	PROCFS_MACHDEP_NODETYPE_CASES
    463  1.75        ad 		error = procfs_machdep_rw(curl, l, pfs, uio);
    464  1.75        ad 		break;
    465  1.40   thorpej #endif
    466   1.1        pk 
    467   1.5       cgd 	default:
    468  1.75        ad 		error = EOPNOTSUPP;
    469  1.75        ad 		break;
    470   1.5       cgd 	}
    471  1.75        ad 
    472  1.75        ad 	/*
    473  1.75        ad 	 * Release the references that we acquired earlier.
    474  1.75        ad 	 */
    475  1.75        ad 	lwp_delref(l);
    476  1.75        ad 	procfs_proc_unlock(p);
    477  1.75        ad 
    478  1.75        ad 	return (error);
    479   1.1        pk }
    480   1.1        pk 
    481   1.5       cgd /*
    482  1.64  christos  * Get a string from userland into (bf).  Strip a trailing
    483   1.5       cgd  * nl character (to allow easy access from the shell).
    484  1.11   mycroft  * The buffer should be *buflenp + 1 chars long.  vfs_getuserstr
    485   1.5       cgd  * will automatically add a nul char at the end.
    486   1.5       cgd  *
    487   1.5       cgd  * Returns 0 on success or the following errors
    488   1.5       cgd  *
    489   1.5       cgd  * EINVAL:    file offset is non-zero.
    490   1.5       cgd  * EMSGSIZE:  message is longer than kernel buffer
    491   1.5       cgd  * EFAULT:    user i/o buffer is not addressable
    492   1.5       cgd  */
    493  1.11   mycroft int
    494  1.64  christos vfs_getuserstr(uio, bf, buflenp)
    495   1.5       cgd 	struct uio *uio;
    496  1.64  christos 	char *bf;
    497   1.5       cgd 	int *buflenp;
    498   1.1        pk {
    499   1.5       cgd 	int xlen;
    500   1.5       cgd 	int error;
    501   1.5       cgd 
    502  1.11   mycroft 	if (uio->uio_offset != 0)
    503  1.11   mycroft 		return (EINVAL);
    504  1.11   mycroft 
    505   1.5       cgd 	xlen = *buflenp;
    506   1.1        pk 
    507   1.5       cgd 	/* must be able to read the whole string in one go */
    508   1.5       cgd 	if (xlen < uio->uio_resid)
    509   1.5       cgd 		return (EMSGSIZE);
    510   1.5       cgd 	xlen = uio->uio_resid;
    511   1.5       cgd 
    512  1.64  christos 	if ((error = uiomove(bf, xlen, uio)) != 0)
    513   1.5       cgd 		return (error);
    514   1.5       cgd 
    515  1.11   mycroft 	/* allow multiple writes without seeks */
    516  1.11   mycroft 	uio->uio_offset = 0;
    517  1.11   mycroft 
    518   1.5       cgd 	/* cleanup string and remove trailing newline */
    519  1.64  christos 	bf[xlen] = '\0';
    520  1.64  christos 	xlen = strlen(bf);
    521  1.64  christos 	if (xlen > 0 && bf[xlen-1] == '\n')
    522  1.64  christos 		bf[--xlen] = '\0';
    523   1.5       cgd 	*buflenp = xlen;
    524   1.1        pk 
    525   1.5       cgd 	return (0);
    526   1.1        pk }
    527   1.1        pk 
    528  1.36  jdolecek const vfs_namemap_t *
    529  1.64  christos vfs_findname(nm, bf, buflen)
    530  1.36  jdolecek 	const vfs_namemap_t *nm;
    531  1.64  christos 	const char *bf;
    532   1.5       cgd 	int buflen;
    533   1.1        pk {
    534  1.11   mycroft 
    535   1.5       cgd 	for (; nm->nm_name; nm++)
    536  1.64  christos 		if (memcmp(bf, nm->nm_name, buflen+1) == 0)
    537   1.5       cgd 			return (nm);
    538   1.5       cgd 
    539   1.5       cgd 	return (0);
    540  1.29      fvdl }
    541  1.29      fvdl 
    542  1.29      fvdl /*
    543  1.29      fvdl  * Initialize pfsnode hash table.
    544  1.29      fvdl  */
    545  1.29      fvdl void
    546  1.29      fvdl procfs_hashinit()
    547  1.29      fvdl {
    548  1.76        ad 	mutex_init(&pfs_hashlock, MUTEX_DEFAULT, IPL_NONE);
    549  1.76        ad 	mutex_init(&pfs_ihash_lock, MUTEX_DEFAULT, IPL_NONE);
    550  1.90        ad 	pfs_hashtbl = hashinit(desiredvnodes / 4, HASH_LIST, true, &pfs_ihash);
    551  1.31  jdolecek }
    552  1.31  jdolecek 
    553  1.38       chs void
    554  1.38       chs procfs_hashreinit()
    555  1.38       chs {
    556  1.38       chs 	struct pfsnode *pp;
    557  1.38       chs 	struct pfs_hashhead *oldhash, *hash;
    558  1.41   thorpej 	u_long i, oldmask, mask, val;
    559  1.38       chs 
    560  1.90        ad 	hash = hashinit(desiredvnodes / 4, HASH_LIST, true, &mask);
    561  1.38       chs 
    562  1.76        ad 	mutex_enter(&pfs_ihash_lock);
    563  1.38       chs 	oldhash = pfs_hashtbl;
    564  1.38       chs 	oldmask = pfs_ihash;
    565  1.38       chs 	pfs_hashtbl = hash;
    566  1.38       chs 	pfs_ihash = mask;
    567  1.38       chs 	for (i = 0; i <= oldmask; i++) {
    568  1.38       chs 		while ((pp = LIST_FIRST(&oldhash[i])) != NULL) {
    569  1.38       chs 			LIST_REMOVE(pp, pfs_hash);
    570  1.38       chs 			val = PFSPIDHASH(pp->pfs_pid);
    571  1.38       chs 			LIST_INSERT_HEAD(&hash[val], pp, pfs_hash);
    572  1.38       chs 		}
    573  1.38       chs 	}
    574  1.76        ad 	mutex_exit(&pfs_ihash_lock);
    575  1.90        ad 	hashdone(oldhash, HASH_LIST, oldmask);
    576  1.38       chs }
    577  1.38       chs 
    578  1.31  jdolecek /*
    579  1.31  jdolecek  * Free pfsnode hash table.
    580  1.31  jdolecek  */
    581  1.31  jdolecek void
    582  1.31  jdolecek procfs_hashdone()
    583  1.31  jdolecek {
    584  1.90        ad 	hashdone(pfs_hashtbl, HASH_LIST, pfs_ihash);
    585  1.78        ad 	mutex_destroy(&pfs_hashlock);
    586  1.78        ad 	mutex_destroy(&pfs_ihash_lock);
    587  1.29      fvdl }
    588  1.29      fvdl 
    589  1.29      fvdl struct vnode *
    590  1.82        ad procfs_hashget(pid, type, fd, mp, flags)
    591  1.29      fvdl 	pid_t pid;
    592  1.29      fvdl 	pfstype type;
    593  1.42  christos 	int fd;
    594  1.29      fvdl 	struct mount *mp;
    595  1.82        ad 	int flags;
    596  1.29      fvdl {
    597  1.38       chs 	struct pfs_hashhead *ppp;
    598  1.29      fvdl 	struct pfsnode *pp;
    599  1.29      fvdl 	struct vnode *vp;
    600  1.29      fvdl 
    601  1.29      fvdl loop:
    602  1.76        ad 	mutex_enter(&pfs_ihash_lock);
    603  1.38       chs 	ppp = &pfs_hashtbl[PFSPIDHASH(pid)];
    604  1.38       chs 	LIST_FOREACH(pp, ppp, pfs_hash) {
    605  1.29      fvdl 		vp = PFSTOV(pp);
    606  1.29      fvdl 		if (pid == pp->pfs_pid && pp->pfs_type == type &&
    607  1.42  christos 		    pp->pfs_fd == fd && vp->v_mount == mp) {
    608  1.82        ad 		    	if (flags == 0) {
    609  1.82        ad 				mutex_exit(&pfs_ihash_lock);
    610  1.82        ad 			} else {
    611  1.83        ad 				mutex_enter(&vp->v_interlock);
    612  1.82        ad 				mutex_exit(&pfs_ihash_lock);
    613  1.82        ad 				if (vget(vp, flags | LK_INTERLOCK))
    614  1.82        ad 					goto loop;
    615  1.82        ad 			}
    616  1.29      fvdl 			return (vp);
    617  1.29      fvdl 		}
    618  1.29      fvdl 	}
    619  1.76        ad 	mutex_exit(&pfs_ihash_lock);
    620  1.29      fvdl 	return (NULL);
    621  1.29      fvdl }
    622  1.29      fvdl 
    623  1.29      fvdl /*
    624  1.29      fvdl  * Insert the pfsnode into the hash table and lock it.
    625  1.29      fvdl  */
    626  1.29      fvdl void
    627  1.29      fvdl procfs_hashins(pp)
    628  1.29      fvdl 	struct pfsnode *pp;
    629  1.29      fvdl {
    630  1.29      fvdl 	struct pfs_hashhead *ppp;
    631  1.29      fvdl 
    632  1.29      fvdl 	/* lock the pfsnode, then put it on the appropriate hash list */
    633  1.85        ad 	vlockmgr(&pp->pfs_vnode->v_lock, LK_EXCLUSIVE);
    634  1.29      fvdl 
    635  1.76        ad 	mutex_enter(&pfs_ihash_lock);
    636  1.38       chs 	ppp = &pfs_hashtbl[PFSPIDHASH(pp->pfs_pid)];
    637  1.29      fvdl 	LIST_INSERT_HEAD(ppp, pp, pfs_hash);
    638  1.76        ad 	mutex_exit(&pfs_ihash_lock);
    639  1.29      fvdl }
    640  1.29      fvdl 
    641  1.29      fvdl /*
    642  1.29      fvdl  * Remove the pfsnode from the hash table.
    643  1.29      fvdl  */
    644  1.29      fvdl void
    645  1.29      fvdl procfs_hashrem(pp)
    646  1.29      fvdl 	struct pfsnode *pp;
    647  1.29      fvdl {
    648  1.76        ad 	mutex_enter(&pfs_ihash_lock);
    649  1.29      fvdl 	LIST_REMOVE(pp, pfs_hash);
    650  1.76        ad 	mutex_exit(&pfs_ihash_lock);
    651  1.29      fvdl }
    652  1.29      fvdl 
    653  1.29      fvdl void
    654  1.56      fvdl procfs_revoke_vnodes(p, arg)
    655  1.56      fvdl 	struct proc *p;
    656  1.29      fvdl 	void *arg;
    657  1.29      fvdl {
    658  1.29      fvdl 	struct pfsnode *pfs, *pnext;
    659  1.29      fvdl 	struct vnode *vp;
    660  1.29      fvdl 	struct mount *mp = (struct mount *)arg;
    661  1.38       chs 	struct pfs_hashhead *ppp;
    662  1.29      fvdl 
    663  1.77     pavel 	if (!(p->p_flag & PK_SUGID))
    664  1.29      fvdl 		return;
    665  1.29      fvdl 
    666  1.82        ad 	mutex_enter(&pfs_ihash_lock);
    667  1.38       chs 	ppp = &pfs_hashtbl[PFSPIDHASH(p->p_pid)];
    668  1.38       chs 	for (pfs = LIST_FIRST(ppp); pfs; pfs = pnext) {
    669  1.29      fvdl 		vp = PFSTOV(pfs);
    670  1.38       chs 		pnext = LIST_NEXT(pfs, pfs_hash);
    671  1.83        ad 		mutex_enter(&vp->v_interlock);
    672  1.29      fvdl 		if (vp->v_usecount > 0 && pfs->pfs_pid == p->p_pid &&
    673  1.82        ad 		    vp->v_mount == mp) {
    674  1.82        ad 		    	vp->v_usecount++;
    675  1.83        ad 		    	mutex_exit(&vp->v_interlock);
    676  1.82        ad 			mutex_exit(&pfs_ihash_lock);
    677  1.29      fvdl 			VOP_REVOKE(vp, REVOKEALL);
    678  1.82        ad 			vrele(vp);
    679  1.82        ad 			mutex_enter(&pfs_ihash_lock);
    680  1.82        ad 		} else {
    681  1.83        ad 			mutex_exit(&vp->v_interlock);
    682  1.82        ad 		}
    683  1.29      fvdl 	}
    684  1.82        ad 	mutex_exit(&pfs_ihash_lock);
    685  1.42  christos }
    686  1.42  christos 
    687  1.42  christos int
    688  1.75        ad procfs_proc_lock(int pid, struct proc **bunghole, int notfound)
    689  1.42  christos {
    690  1.75        ad 	struct proc *tp;
    691  1.75        ad 	int error = 0;
    692  1.42  christos 
    693  1.87        ad 	mutex_enter(proc_lock);
    694  1.42  christos 
    695  1.75        ad 	if (pid == 0)
    696  1.75        ad 		tp = &proc0;
    697  1.75        ad 	else if ((tp = p_find(pid, PFIND_LOCKED)) == NULL)
    698  1.75        ad 		error = notfound;
    699  1.82        ad 	if (tp != NULL && !rw_tryenter(&tp->p_reflock, RW_READER))
    700  1.82        ad 		error = EBUSY;
    701  1.75        ad 
    702  1.87        ad 	mutex_exit(proc_lock);
    703  1.42  christos 
    704  1.75        ad 	*bunghole = tp;
    705  1.75        ad 	return error;
    706  1.75        ad }
    707  1.49  jdolecek 
    708  1.75        ad void
    709  1.75        ad procfs_proc_unlock(struct proc *p)
    710  1.75        ad {
    711  1.82        ad 
    712  1.82        ad 	rw_exit(&p->p_reflock);
    713   1.1        pk }
    714  1.71  christos 
    715  1.71  christos int
    716  1.72  christos procfs_doemul(struct lwp *curl, struct proc *p,
    717  1.72  christos     struct pfsnode *pfs, struct uio *uio)
    718  1.71  christos {
    719  1.71  christos 	const char *ename = p->p_emul->e_name;
    720  1.71  christos 	return uiomove_frombuf(__UNCONST(ename), strlen(ename), uio);
    721  1.71  christos }
    722