procfs_subr.c revision 1.36.2.1 1 /* $NetBSD: procfs_subr.c,v 1.36.2.1 2001/03/05 22:49:51 nathanw Exp $ */
2
3 /*
4 * Copyright (c) 1994 Christopher G. Demetriou. All rights reserved.
5 * Copyright (c) 1993 Jan-Simon Pendry
6 * Copyright (c) 1993
7 * The Regents of the University of California. All rights reserved.
8 *
9 * This code is derived from software contributed to Berkeley by
10 * Jan-Simon Pendry.
11 *
12 * Redistribution and use in source and binary forms, with or without
13 * modification, are permitted provided that the following conditions
14 * are met:
15 * 1. Redistributions of source code must retain the above copyright
16 * notice, this list of conditions and the following disclaimer.
17 * 2. Redistributions in binary form must reproduce the above copyright
18 * notice, this list of conditions and the following disclaimer in the
19 * documentation and/or other materials provided with the distribution.
20 * 3. All advertising materials mentioning features or use of this software
21 * must display the following acknowledgement:
22 * This product includes software developed by the University of
23 * California, Berkeley and its contributors.
24 * 4. Neither the name of the University nor the names of its contributors
25 * may be used to endorse or promote products derived from this software
26 * without specific prior written permission.
27 *
28 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
29 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
30 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
31 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
32 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
33 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
34 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
35 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
36 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
37 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
38 * SUCH DAMAGE.
39 *
40 * @(#)procfs_subr.c 8.6 (Berkeley) 5/14/95
41 */
42
43 #include <sys/param.h>
44 #include <sys/systm.h>
45 #include <sys/time.h>
46 #include <sys/kernel.h>
47 #include <sys/proc.h>
48 #include <sys/vnode.h>
49 #include <sys/malloc.h>
50 #include <sys/stat.h>
51
52 #include <miscfs/procfs/procfs.h>
53
54 void procfs_hashins __P((struct pfsnode *));
55 void procfs_hashrem __P((struct pfsnode *));
56 struct vnode *procfs_hashget __P((pid_t, pfstype, struct mount *));
57
58 LIST_HEAD(pfs_hashhead, pfsnode) *pfs_hashtbl;
59 u_long pfs_ihash; /* size of hash table - 1 */
60 #define PFSPIDHASH(pid) (&pfs_hashtbl[(pid) & pfs_ihash])
61
62 struct lock pfs_hashlock;
63 struct simplelock pfs_hash_slock;
64
65 #define ISSET(t, f) ((t) & (f))
66
67 /*
68 * allocate a pfsnode/vnode pair. the vnode is
69 * referenced, and locked.
70 *
71 * the pid, pfs_type, and mount point uniquely
72 * identify a pfsnode. the mount point is needed
73 * because someone might mount this filesystem
74 * twice.
75 *
76 * all pfsnodes are maintained on a singly-linked
77 * list. new nodes are only allocated when they cannot
78 * be found on this list. entries on the list are
79 * removed when the vfs reclaim entry is called.
80 *
81 * a single lock is kept for the entire list. this is
82 * needed because the getnewvnode() function can block
83 * waiting for a vnode to become free, in which case there
84 * may be more than one process trying to get the same
85 * vnode. this lock is only taken if we are going to
86 * call getnewvnode, since the kernel itself is single-threaded.
87 *
88 * if an entry is found on the list, then call vget() to
89 * take a reference. this is done because there may be
90 * zero references to it and so it needs to removed from
91 * the vnode free list.
92 */
93 int
94 procfs_allocvp(mp, vpp, pid, pfs_type)
95 struct mount *mp;
96 struct vnode **vpp;
97 long pid;
98 pfstype pfs_type;
99 {
100 struct pfsnode *pfs;
101 struct vnode *vp;
102 int error;
103
104 do {
105 if ((*vpp = procfs_hashget(pid, pfs_type, mp)) != NULL)
106 return (0);
107 } while (lockmgr(&pfs_hashlock, LK_EXCLUSIVE|LK_SLEEPFAIL, 0));
108
109 if ((error = getnewvnode(VT_PROCFS, mp, procfs_vnodeop_p, vpp)) != 0) {
110 *vpp = NULL;
111 lockmgr(&pfs_hashlock, LK_RELEASE, NULL);
112 return (error);
113 }
114 vp = *vpp;
115
116 MALLOC(pfs, void *, sizeof(struct pfsnode), M_TEMP, M_WAITOK);
117 vp->v_data = pfs;
118
119 pfs->pfs_pid = (pid_t) pid;
120 pfs->pfs_type = pfs_type;
121 pfs->pfs_vnode = vp;
122 pfs->pfs_flags = 0;
123 pfs->pfs_fileno = PROCFS_FILENO(pid, pfs_type);
124
125 switch (pfs_type) {
126 case Proot: /* /proc = dr-xr-xr-x */
127 pfs->pfs_mode = S_IRUSR|S_IXUSR|S_IRGRP|S_IXGRP|S_IROTH|S_IXOTH;
128 vp->v_type = VDIR;
129 vp->v_flag = VROOT;
130 break;
131
132 case Pcurproc: /* /proc/curproc = lr-xr-xr-x */
133 case Pself: /* /proc/self = lr-xr-xr-x */
134 pfs->pfs_mode = S_IRUSR|S_IXUSR|S_IRGRP|S_IXGRP|S_IROTH|S_IXOTH;
135 vp->v_type = VLNK;
136 break;
137
138 case Pproc: /* /proc/N = dr-xr-xr-x */
139 pfs->pfs_mode = S_IRUSR|S_IXUSR|S_IRGRP|S_IXGRP|S_IROTH|S_IXOTH;
140 vp->v_type = VDIR;
141 break;
142
143 case Pfile: /* /proc/N/file = -rw------- */
144 case Pmem: /* /proc/N/mem = -rw------- */
145 case Pregs: /* /proc/N/regs = -rw------- */
146 case Pfpregs: /* /proc/N/fpregs = -rw------- */
147 pfs->pfs_mode = S_IRUSR|S_IWUSR;
148 vp->v_type = VREG;
149 break;
150
151 case Pctl: /* /proc/N/ctl = --w------ */
152 case Pnote: /* /proc/N/note = --w------ */
153 case Pnotepg: /* /proc/N/notepg = --w------ */
154 pfs->pfs_mode = S_IWUSR;
155 vp->v_type = VREG;
156 break;
157
158 case Pmap: /* /proc/N/map = -r--r--r-- */
159 case Pstatus: /* /proc/N/status = -r--r--r-- */
160 case Pcmdline: /* /proc/N/cmdline = -r--r--r-- */
161 case Pmeminfo: /* /proc/meminfo = -r--r--r-- */
162 case Pcpuinfo: /* /proc/cpuinfo = -r--r--r-- */
163 pfs->pfs_mode = S_IRUSR|S_IRGRP|S_IROTH;
164 vp->v_type = VREG;
165 break;
166
167 default:
168 panic("procfs_allocvp");
169 }
170
171 procfs_hashins(pfs);
172 uvm_vnp_setsize(vp, 0);
173 lockmgr(&pfs_hashlock, LK_RELEASE, NULL);
174
175 return (error);
176 }
177
178 int
179 procfs_freevp(vp)
180 struct vnode *vp;
181 {
182 struct pfsnode *pfs = VTOPFS(vp);
183
184 procfs_hashrem(pfs);
185
186 FREE(vp->v_data, M_TEMP);
187 vp->v_data = 0;
188 return (0);
189 }
190
191 int
192 procfs_rw(v)
193 void *v;
194 {
195 struct vop_read_args *ap = v;
196 struct vnode *vp = ap->a_vp;
197 struct uio *uio = ap->a_uio;
198 struct proc *curp = uio->uio_procp;
199 struct pfsnode *pfs = VTOPFS(vp);
200 struct lwp *l;
201 struct proc *p;
202
203 p = PFIND(pfs->pfs_pid);
204 if (p == 0)
205 return (EINVAL);
206
207 /* XXX NJWLWP
208 * The entire procfs interface needs work to be useful to
209 * a process with multiple LWPs. For the moment, we'll
210 * just kluge this and fail on others.
211 */
212
213 if (p->p_nlwps > 1)
214 switch (pfs->pfs_type) {
215 case Pnote:
216 case Pnotepg:
217 case Pstatus:
218 case Pmap:
219 case Pmem:
220 case Pcmdline:
221 break;
222 case Pregs:
223 case Pfpregs:
224 case Pctl:
225 default:
226 return (EOPNOTSUPP);
227 }
228
229 l = LIST_FIRST(&p->p_lwps);
230
231 switch (pfs->pfs_type) {
232 case Pregs:
233 case Pfpregs:
234 case Pmem:
235 /*
236 * Do not allow init to be modified while in secure mode; it
237 * could be duped into changing the security level.
238 */
239 if (uio->uio_rw == UIO_WRITE &&
240 p == initproc && securelevel > -1)
241 return (EPERM);
242 break;
243
244 default:
245 break;
246 }
247
248 switch (pfs->pfs_type) {
249 case Pnote:
250 case Pnotepg:
251 return (procfs_donote(curp, p, pfs, uio));
252
253 case Pregs:
254 return (procfs_doregs(curp, l, pfs, uio));
255
256 case Pfpregs:
257 return (procfs_dofpregs(curp, l, pfs, uio));
258
259 case Pctl:
260 return (procfs_doctl(curp, l, pfs, uio));
261
262 case Pstatus:
263 return (procfs_dostatus(curp, l, pfs, uio));
264
265 case Pmap:
266 return (procfs_domap(curp, p, pfs, uio));
267
268 case Pmem:
269 return (procfs_domem(curp, l, pfs, uio));
270
271 case Pcmdline:
272 return (procfs_docmdline(curp, p, pfs, uio));
273
274 case Pmeminfo:
275 return (procfs_domeminfo(curp, p, pfs, uio));
276 case Pcpuinfo:
277 return (procfs_docpuinfo(curp, p, pfs, uio));
278
279 default:
280 return (EOPNOTSUPP);
281 }
282 }
283
284 /*
285 * Get a string from userland into (buf). Strip a trailing
286 * nl character (to allow easy access from the shell).
287 * The buffer should be *buflenp + 1 chars long. vfs_getuserstr
288 * will automatically add a nul char at the end.
289 *
290 * Returns 0 on success or the following errors
291 *
292 * EINVAL: file offset is non-zero.
293 * EMSGSIZE: message is longer than kernel buffer
294 * EFAULT: user i/o buffer is not addressable
295 */
296 int
297 vfs_getuserstr(uio, buf, buflenp)
298 struct uio *uio;
299 char *buf;
300 int *buflenp;
301 {
302 int xlen;
303 int error;
304
305 if (uio->uio_offset != 0)
306 return (EINVAL);
307
308 xlen = *buflenp;
309
310 /* must be able to read the whole string in one go */
311 if (xlen < uio->uio_resid)
312 return (EMSGSIZE);
313 xlen = uio->uio_resid;
314
315 if ((error = uiomove(buf, xlen, uio)) != 0)
316 return (error);
317
318 /* allow multiple writes without seeks */
319 uio->uio_offset = 0;
320
321 /* cleanup string and remove trailing newline */
322 buf[xlen] = '\0';
323 xlen = strlen(buf);
324 if (xlen > 0 && buf[xlen-1] == '\n')
325 buf[--xlen] = '\0';
326 *buflenp = xlen;
327
328 return (0);
329 }
330
331 const vfs_namemap_t *
332 vfs_findname(nm, buf, buflen)
333 const vfs_namemap_t *nm;
334 const char *buf;
335 int buflen;
336 {
337
338 for (; nm->nm_name; nm++)
339 if (memcmp(buf, nm->nm_name, buflen+1) == 0)
340 return (nm);
341
342 return (0);
343 }
344
345 /*
346 * Initialize pfsnode hash table.
347 */
348 void
349 procfs_hashinit()
350 {
351 lockinit(&pfs_hashlock, PINOD, "pfs_hashlock", 0, 0);
352 pfs_hashtbl = hashinit(desiredvnodes / 4, HASH_LIST, M_UFSMNT,
353 M_WAITOK, &pfs_ihash);
354 simple_lock_init(&pfs_hash_slock);
355 }
356
357 /*
358 * Free pfsnode hash table.
359 */
360 void
361 procfs_hashdone()
362 {
363 hashdone(pfs_hashtbl, M_UFSMNT);
364 }
365
366 struct vnode *
367 procfs_hashget(pid, type, mp)
368 pid_t pid;
369 pfstype type;
370 struct mount *mp;
371 {
372 struct pfsnode *pp;
373 struct vnode *vp;
374
375 loop:
376 simple_lock(&pfs_hash_slock);
377 for (pp = PFSPIDHASH(pid)->lh_first; pp; pp = pp->pfs_hash.le_next) {
378 vp = PFSTOV(pp);
379 if (pid == pp->pfs_pid && pp->pfs_type == type &&
380 vp->v_mount == mp) {
381 simple_lock(&vp->v_interlock);
382 simple_unlock(&pfs_hash_slock);
383 if (vget(vp, LK_EXCLUSIVE | LK_INTERLOCK))
384 goto loop;
385 return (vp);
386 }
387 }
388 simple_unlock(&pfs_hash_slock);
389 return (NULL);
390 }
391
392 /*
393 * Insert the pfsnode into the hash table and lock it.
394 */
395 void
396 procfs_hashins(pp)
397 struct pfsnode *pp;
398 {
399 struct pfs_hashhead *ppp;
400
401 /* lock the pfsnode, then put it on the appropriate hash list */
402 lockmgr(&pp->pfs_vnode->v_lock, LK_EXCLUSIVE, (struct simplelock *)0);
403
404 simple_lock(&pfs_hash_slock);
405 ppp = PFSPIDHASH(pp->pfs_pid);
406 LIST_INSERT_HEAD(ppp, pp, pfs_hash);
407 simple_unlock(&pfs_hash_slock);
408 }
409
410 /*
411 * Remove the pfsnode from the hash table.
412 */
413 void
414 procfs_hashrem(pp)
415 struct pfsnode *pp;
416 {
417 simple_lock(&pfs_hash_slock);
418 LIST_REMOVE(pp, pfs_hash);
419 simple_unlock(&pfs_hash_slock);
420 }
421
422 void
423 procfs_revoke_vnodes(p, arg)
424 struct proc *p;
425 void *arg;
426 {
427 struct pfsnode *pfs, *pnext;
428 struct vnode *vp;
429 struct mount *mp = (struct mount *)arg;
430
431 if (!(p->p_flag & P_SUGID))
432 return;
433
434 for (pfs = PFSPIDHASH(p->p_pid)->lh_first; pfs; pfs = pnext) {
435 vp = PFSTOV(pfs);
436 pnext = pfs->pfs_hash.le_next;
437 if (vp->v_usecount > 0 && pfs->pfs_pid == p->p_pid &&
438 vp->v_mount == mp)
439 VOP_REVOKE(vp, REVOKEALL);
440 }
441 }
442