procfs_subr.c revision 1.42 1 /* $NetBSD: procfs_subr.c,v 1.42 2003/01/03 13:21:18 christos 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/cdefs.h>
44 __KERNEL_RCSID(0, "$NetBSD: procfs_subr.c,v 1.42 2003/01/03 13:21:18 christos Exp $");
45
46 #include <sys/param.h>
47 #include <sys/systm.h>
48 #include <sys/time.h>
49 #include <sys/kernel.h>
50 #include <sys/proc.h>
51 #include <sys/vnode.h>
52 #include <sys/malloc.h>
53 #include <sys/stat.h>
54 #include <sys/file.h>
55 #include <sys/filedesc.h>
56
57 #include <miscfs/procfs/procfs.h>
58
59 void procfs_hashins __P((struct pfsnode *));
60 void procfs_hashrem __P((struct pfsnode *));
61 struct vnode *procfs_hashget __P((pid_t, pfstype, int, struct mount *));
62
63 LIST_HEAD(pfs_hashhead, pfsnode) *pfs_hashtbl;
64 u_long pfs_ihash; /* size of hash table - 1 */
65 #define PFSPIDHASH(pid) ((pid) & pfs_ihash)
66
67 struct lock pfs_hashlock;
68 struct simplelock pfs_hash_slock;
69
70 #define ISSET(t, f) ((t) & (f))
71
72 /*
73 * allocate a pfsnode/vnode pair. the vnode is
74 * referenced, and locked.
75 *
76 * the pid, pfs_type, and mount point uniquely
77 * identify a pfsnode. the mount point is needed
78 * because someone might mount this filesystem
79 * twice.
80 *
81 * all pfsnodes are maintained on a singly-linked
82 * list. new nodes are only allocated when they cannot
83 * be found on this list. entries on the list are
84 * removed when the vfs reclaim entry is called.
85 *
86 * a single lock is kept for the entire list. this is
87 * needed because the getnewvnode() function can block
88 * waiting for a vnode to become free, in which case there
89 * may be more than one process trying to get the same
90 * vnode. this lock is only taken if we are going to
91 * call getnewvnode, since the kernel itself is single-threaded.
92 *
93 * if an entry is found on the list, then call vget() to
94 * take a reference. this is done because there may be
95 * zero references to it and so it needs to removed from
96 * the vnode free list.
97 */
98 int
99 procfs_allocvp(mp, vpp, pid, pfs_type, fd)
100 struct mount *mp;
101 struct vnode **vpp;
102 pid_t pid;
103 pfstype pfs_type;
104 int fd;
105 {
106 struct pfsnode *pfs;
107 struct vnode *vp;
108 int error;
109
110 do {
111 if ((*vpp = procfs_hashget(pid, pfs_type, fd, mp)) != NULL)
112 return (0);
113 } while (lockmgr(&pfs_hashlock, LK_EXCLUSIVE|LK_SLEEPFAIL, 0));
114
115 if ((error = getnewvnode(VT_PROCFS, mp, procfs_vnodeop_p, vpp)) != 0) {
116 *vpp = NULL;
117 lockmgr(&pfs_hashlock, LK_RELEASE, NULL);
118 return (error);
119 }
120 vp = *vpp;
121
122 MALLOC(pfs, void *, sizeof(struct pfsnode), M_TEMP, M_WAITOK);
123 vp->v_data = pfs;
124
125 pfs->pfs_pid = pid;
126 pfs->pfs_type = pfs_type;
127 pfs->pfs_vnode = vp;
128 pfs->pfs_flags = 0;
129 pfs->pfs_fileno = PROCFS_FILENO(pid, pfs_type, fd);
130 pfs->pfs_fd = fd;
131
132 switch (pfs_type) {
133 case Proot: /* /proc = dr-xr-xr-x */
134 pfs->pfs_mode = S_IRUSR|S_IXUSR|S_IRGRP|S_IXGRP|S_IROTH|S_IXOTH;
135 vp->v_type = VDIR;
136 vp->v_flag = VROOT;
137 break;
138
139 case Pcurproc: /* /proc/curproc = lr-xr-xr-x */
140 case Pself: /* /proc/self = lr-xr-xr-x */
141 pfs->pfs_mode = S_IRUSR|S_IXUSR|S_IRGRP|S_IXGRP|S_IROTH|S_IXOTH;
142 vp->v_type = VLNK;
143 break;
144
145 case Pproc: /* /proc/N = dr-xr-xr-x */
146 case Pfd:
147 if (fd == -1) { /* /proc/N/fd = dr-xr-xr-x */
148 pfs->pfs_mode =
149 S_IRUSR|S_IXUSR|S_IRGRP|S_IXGRP|S_IROTH|S_IXOTH;
150 vp->v_type = VDIR;
151 } else { /* /proc/N/fd/M = [ps-]rw------- */
152 struct file *fp;
153 struct vnode *vxp;
154 if ((error = procfs_getfp(pfs, &fp)) != 0)
155 return error;
156 pfs->pfs_mode = S_IRUSR|S_IWUSR;
157 switch (fp->f_type) {
158 case DTYPE_VNODE:
159 vxp = (struct vnode *)fp->f_data;
160 vp->v_type = vxp->v_type;
161 break;
162 case DTYPE_PIPE:
163 vp->v_type = VFIFO;
164 break;
165 case DTYPE_SOCKET:
166 vp->v_type = VSOCK;
167 break;
168 case DTYPE_KQUEUE:
169 case DTYPE_MISC:
170 return EOPNOTSUPP;
171 default:
172 panic("unknown file type %d\n", fp->f_type);
173 }
174 }
175 break;
176
177 case Pfile: /* /proc/N/file = -rw------- */
178 case Pmem: /* /proc/N/mem = -rw------- */
179 case Pregs: /* /proc/N/regs = -rw------- */
180 case Pfpregs: /* /proc/N/fpregs = -rw------- */
181 pfs->pfs_mode = S_IRUSR|S_IWUSR;
182 vp->v_type = VREG;
183 break;
184
185 case Pctl: /* /proc/N/ctl = --w------ */
186 case Pnote: /* /proc/N/note = --w------ */
187 case Pnotepg: /* /proc/N/notepg = --w------ */
188 pfs->pfs_mode = S_IWUSR;
189 vp->v_type = VREG;
190 break;
191
192 case Pmap: /* /proc/N/map = -r--r--r-- */
193 case Pmaps: /* /proc/N/maps = -r--r--r-- */
194 case Pstatus: /* /proc/N/status = -r--r--r-- */
195 case Pcmdline: /* /proc/N/cmdline = -r--r--r-- */
196 case Pmeminfo: /* /proc/meminfo = -r--r--r-- */
197 case Pcpuinfo: /* /proc/cpuinfo = -r--r--r-- */
198 pfs->pfs_mode = S_IRUSR|S_IRGRP|S_IROTH;
199 vp->v_type = VREG;
200 break;
201
202 #ifdef __HAVE_PROCFS_MACHDEP
203 PROCFS_MACHDEP_NODETYPE_CASES
204 procfs_machdep_allocvp(vp);
205 break;
206 #endif
207
208 default:
209 panic("procfs_allocvp");
210 }
211
212 procfs_hashins(pfs);
213 uvm_vnp_setsize(vp, 0);
214 lockmgr(&pfs_hashlock, LK_RELEASE, NULL);
215
216 return (error);
217 }
218
219 int
220 procfs_freevp(vp)
221 struct vnode *vp;
222 {
223 struct pfsnode *pfs = VTOPFS(vp);
224
225 procfs_hashrem(pfs);
226
227 FREE(vp->v_data, M_TEMP);
228 vp->v_data = 0;
229 return (0);
230 }
231
232 int
233 procfs_rw(v)
234 void *v;
235 {
236 struct vop_read_args *ap = v;
237 struct vnode *vp = ap->a_vp;
238 struct uio *uio = ap->a_uio;
239 struct proc *curp = uio->uio_procp;
240 struct pfsnode *pfs = VTOPFS(vp);
241 struct proc *p;
242
243 p = PFIND(pfs->pfs_pid);
244 if (p == 0)
245 return (EINVAL);
246
247 switch (pfs->pfs_type) {
248 case Pregs:
249 case Pfpregs:
250 case Pmem:
251 #if defined(__HAVE_PROCFS_MACHDEP) && defined(PROCFS_MACHDEP_PROTECT_CASES)
252 PROCFS_MACHDEP_PROTECT_CASES
253 #endif
254 /*
255 * Do not allow init to be modified while in secure mode; it
256 * could be duped into changing the security level.
257 */
258 if (uio->uio_rw == UIO_WRITE &&
259 p == initproc && securelevel > -1)
260 return (EPERM);
261 break;
262
263 default:
264 break;
265 }
266
267 switch (pfs->pfs_type) {
268 case Pnote:
269 case Pnotepg:
270 return (procfs_donote(curp, p, pfs, uio));
271
272 case Pregs:
273 return (procfs_doregs(curp, p, pfs, uio));
274
275 case Pfpregs:
276 return (procfs_dofpregs(curp, p, pfs, uio));
277
278 case Pctl:
279 return (procfs_doctl(curp, p, pfs, uio));
280
281 case Pstatus:
282 return (procfs_dostatus(curp, p, pfs, uio));
283
284 case Pmap:
285 return (procfs_domap(curp, p, pfs, uio, 0));
286
287 case Pmaps:
288 return (procfs_domap(curp, p, pfs, uio, 1));
289
290 case Pmem:
291 return (procfs_domem(curp, p, pfs, uio));
292
293 case Pcmdline:
294 return (procfs_docmdline(curp, p, pfs, uio));
295
296 case Pmeminfo:
297 return (procfs_domeminfo(curp, p, pfs, uio));
298
299 case Pcpuinfo:
300 return (procfs_docpuinfo(curp, p, pfs, uio));
301
302 case Pfd:
303 return (procfs_dofd(curp, p, pfs, uio));
304
305 #ifdef __HAVE_PROCFS_MACHDEP
306 PROCFS_MACHDEP_NODETYPE_CASES
307 return (procfs_machdep_rw(curp, p, pfs, uio));
308 #endif
309
310 default:
311 return (EOPNOTSUPP);
312 }
313 }
314
315 /*
316 * Get a string from userland into (buf). Strip a trailing
317 * nl character (to allow easy access from the shell).
318 * The buffer should be *buflenp + 1 chars long. vfs_getuserstr
319 * will automatically add a nul char at the end.
320 *
321 * Returns 0 on success or the following errors
322 *
323 * EINVAL: file offset is non-zero.
324 * EMSGSIZE: message is longer than kernel buffer
325 * EFAULT: user i/o buffer is not addressable
326 */
327 int
328 vfs_getuserstr(uio, buf, buflenp)
329 struct uio *uio;
330 char *buf;
331 int *buflenp;
332 {
333 int xlen;
334 int error;
335
336 if (uio->uio_offset != 0)
337 return (EINVAL);
338
339 xlen = *buflenp;
340
341 /* must be able to read the whole string in one go */
342 if (xlen < uio->uio_resid)
343 return (EMSGSIZE);
344 xlen = uio->uio_resid;
345
346 if ((error = uiomove(buf, xlen, uio)) != 0)
347 return (error);
348
349 /* allow multiple writes without seeks */
350 uio->uio_offset = 0;
351
352 /* cleanup string and remove trailing newline */
353 buf[xlen] = '\0';
354 xlen = strlen(buf);
355 if (xlen > 0 && buf[xlen-1] == '\n')
356 buf[--xlen] = '\0';
357 *buflenp = xlen;
358
359 return (0);
360 }
361
362 const vfs_namemap_t *
363 vfs_findname(nm, buf, buflen)
364 const vfs_namemap_t *nm;
365 const char *buf;
366 int buflen;
367 {
368
369 for (; nm->nm_name; nm++)
370 if (memcmp(buf, nm->nm_name, buflen+1) == 0)
371 return (nm);
372
373 return (0);
374 }
375
376 /*
377 * Initialize pfsnode hash table.
378 */
379 void
380 procfs_hashinit()
381 {
382 lockinit(&pfs_hashlock, PINOD, "pfs_hashlock", 0, 0);
383 pfs_hashtbl = hashinit(desiredvnodes / 4, HASH_LIST, M_UFSMNT,
384 M_WAITOK, &pfs_ihash);
385 simple_lock_init(&pfs_hash_slock);
386 }
387
388 void
389 procfs_hashreinit()
390 {
391 struct pfsnode *pp;
392 struct pfs_hashhead *oldhash, *hash;
393 u_long i, oldmask, mask, val;
394
395 hash = hashinit(desiredvnodes / 4, HASH_LIST, M_UFSMNT, M_WAITOK,
396 &mask);
397
398 simple_lock(&pfs_hash_slock);
399 oldhash = pfs_hashtbl;
400 oldmask = pfs_ihash;
401 pfs_hashtbl = hash;
402 pfs_ihash = mask;
403 for (i = 0; i <= oldmask; i++) {
404 while ((pp = LIST_FIRST(&oldhash[i])) != NULL) {
405 LIST_REMOVE(pp, pfs_hash);
406 val = PFSPIDHASH(pp->pfs_pid);
407 LIST_INSERT_HEAD(&hash[val], pp, pfs_hash);
408 }
409 }
410 simple_unlock(&pfs_hash_slock);
411 hashdone(oldhash, M_UFSMNT);
412 }
413
414 /*
415 * Free pfsnode hash table.
416 */
417 void
418 procfs_hashdone()
419 {
420 hashdone(pfs_hashtbl, M_UFSMNT);
421 }
422
423 struct vnode *
424 procfs_hashget(pid, type, fd, mp)
425 pid_t pid;
426 pfstype type;
427 int fd;
428 struct mount *mp;
429 {
430 struct pfs_hashhead *ppp;
431 struct pfsnode *pp;
432 struct vnode *vp;
433
434 loop:
435 simple_lock(&pfs_hash_slock);
436 ppp = &pfs_hashtbl[PFSPIDHASH(pid)];
437 LIST_FOREACH(pp, ppp, pfs_hash) {
438 vp = PFSTOV(pp);
439 if (pid == pp->pfs_pid && pp->pfs_type == type &&
440 pp->pfs_fd == fd && vp->v_mount == mp) {
441 simple_lock(&vp->v_interlock);
442 simple_unlock(&pfs_hash_slock);
443 if (vget(vp, LK_EXCLUSIVE | LK_INTERLOCK))
444 goto loop;
445 return (vp);
446 }
447 }
448 simple_unlock(&pfs_hash_slock);
449 return (NULL);
450 }
451
452 /*
453 * Insert the pfsnode into the hash table and lock it.
454 */
455 void
456 procfs_hashins(pp)
457 struct pfsnode *pp;
458 {
459 struct pfs_hashhead *ppp;
460
461 /* lock the pfsnode, then put it on the appropriate hash list */
462 lockmgr(&pp->pfs_vnode->v_lock, LK_EXCLUSIVE, (struct simplelock *)0);
463
464 simple_lock(&pfs_hash_slock);
465 ppp = &pfs_hashtbl[PFSPIDHASH(pp->pfs_pid)];
466 LIST_INSERT_HEAD(ppp, pp, pfs_hash);
467 simple_unlock(&pfs_hash_slock);
468 }
469
470 /*
471 * Remove the pfsnode from the hash table.
472 */
473 void
474 procfs_hashrem(pp)
475 struct pfsnode *pp;
476 {
477 simple_lock(&pfs_hash_slock);
478 LIST_REMOVE(pp, pfs_hash);
479 simple_unlock(&pfs_hash_slock);
480 }
481
482 void
483 procfs_revoke_vnodes(p, arg)
484 struct proc *p;
485 void *arg;
486 {
487 struct pfsnode *pfs, *pnext;
488 struct vnode *vp;
489 struct mount *mp = (struct mount *)arg;
490 struct pfs_hashhead *ppp;
491
492 if (!(p->p_flag & P_SUGID))
493 return;
494
495 ppp = &pfs_hashtbl[PFSPIDHASH(p->p_pid)];
496 for (pfs = LIST_FIRST(ppp); pfs; pfs = pnext) {
497 vp = PFSTOV(pfs);
498 pnext = LIST_NEXT(pfs, pfs_hash);
499 if (vp->v_usecount > 0 && pfs->pfs_pid == p->p_pid &&
500 vp->v_mount == mp)
501 VOP_REVOKE(vp, REVOKEALL);
502 }
503 }
504
505 int
506 procfs_getfp(pfs, fp)
507 struct pfsnode *pfs;
508 struct file **fp;
509 {
510 struct proc *p = PFIND(pfs->pfs_pid);
511
512 if (p == NULL)
513 return ESRCH;
514
515 if (pfs->pfs_fd == -1)
516 return EINVAL;
517
518 if ((*fp = p->p_fd->fd_ofiles[pfs->pfs_fd]) == NULL)
519 return EBADF;
520 return 0;
521 }
522