Merge branch 'master' of ssh://crater.dragonflybsd.org/repository/git/dragonfly
[dragonfly.git] / sys / kern / kern_subr.c
blob87c68f4bbf62fa9a50929f48f0b7e53dfc3f0975
1 /*
2 * Copyright (c) 1982, 1986, 1991, 1993
3 * The Regents of the University of California. All rights reserved.
4 * (c) UNIX System Laboratories, Inc.
5 * All or some portions of this file are derived from material licensed
6 * to the University of California by American Telephone and Telegraph
7 * Co. or Unix System Laboratories, Inc. and are reproduced herein with
8 * the permission of UNIX System Laboratories, Inc.
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. All advertising materials mentioning features or use of this software
19 * must display the following acknowledgement:
20 * This product includes software developed by the University of
21 * California, Berkeley and its contributors.
22 * 4. Neither the name of the University nor the names of its contributors
23 * may be used to endorse or promote products derived from this software
24 * without specific prior written permission.
26 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
27 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
28 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
29 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
30 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
31 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
32 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
33 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
34 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
35 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
36 * SUCH DAMAGE.
38 * @(#)kern_subr.c 8.3 (Berkeley) 1/21/94
39 * $FreeBSD: src/sys/kern/kern_subr.c,v 1.31.2.2 2002/04/21 08:09:37 bde Exp $
40 * $DragonFly: src/sys/kern/kern_subr.c,v 1.27 2007/01/29 20:44:02 tgen Exp $
43 #include "opt_ddb.h"
45 #include <sys/param.h>
46 #include <sys/systm.h>
47 #include <sys/kernel.h>
48 #include <sys/proc.h>
49 #include <sys/malloc.h>
50 #include <sys/lock.h>
51 #include <sys/resourcevar.h>
52 #include <sys/sysctl.h>
53 #include <sys/uio.h>
54 #include <sys/vnode.h>
55 #include <sys/sfbuf.h>
56 #include <sys/thread2.h>
57 #include <machine/limits.h>
59 #include <vm/vm.h>
60 #include <vm/vm_page.h>
61 #include <vm/vm_map.h>
63 SYSCTL_INT(_kern, KERN_IOV_MAX, iov_max, CTLFLAG_RD, NULL, UIO_MAXIOV,
64 "Maximum number of elements in an I/O vector; sysconf(_SC_IOV_MAX)");
67 * UIO_READ: copy the kernelspace cp to the user or kernelspace UIO
68 * UIO_WRITE: copy the user or kernelspace UIO to the kernelspace cp
70 * For userspace UIO's, uio_td must be the current thread.
72 int
73 uiomove(caddr_t cp, int n, struct uio *uio)
75 struct iovec *iov;
76 u_int cnt;
77 int error = 0;
78 int save = 0;
79 int baseticks = ticks;
81 KASSERT(uio->uio_rw == UIO_READ || uio->uio_rw == UIO_WRITE,
82 ("uiomove: mode"));
83 KASSERT(uio->uio_segflg != UIO_USERSPACE || uio->uio_td == curthread,
84 ("uiomove proc"));
86 if (curproc) {
87 save = curproc->p_flag & P_DEADLKTREAT;
88 curproc->p_flag |= P_DEADLKTREAT;
91 while (n > 0 && uio->uio_resid) {
92 iov = uio->uio_iov;
93 cnt = iov->iov_len;
94 if (cnt == 0) {
95 uio->uio_iov++;
96 uio->uio_iovcnt--;
97 continue;
99 if (cnt > n)
100 cnt = n;
102 switch (uio->uio_segflg) {
104 case UIO_USERSPACE:
105 if (ticks - baseticks >= hogticks) {
106 uio_yield();
107 baseticks = ticks;
109 if (uio->uio_rw == UIO_READ)
110 error = copyout(cp, iov->iov_base, cnt);
111 else
112 error = copyin(iov->iov_base, cp, cnt);
113 if (error)
114 break;
115 break;
117 case UIO_SYSSPACE:
118 if (uio->uio_rw == UIO_READ)
119 bcopy((caddr_t)cp, iov->iov_base, cnt);
120 else
121 bcopy(iov->iov_base, (caddr_t)cp, cnt);
122 break;
123 case UIO_NOCOPY:
124 break;
126 iov->iov_base = (char *)iov->iov_base + cnt;
127 iov->iov_len -= cnt;
128 uio->uio_resid -= cnt;
129 uio->uio_offset += cnt;
130 cp += cnt;
131 n -= cnt;
133 if (curproc)
134 curproc->p_flag = (curproc->p_flag & ~P_DEADLKTREAT) | save;
135 return (error);
138 * Wrapper for uiomove() that validates the arguments against a known-good
139 * kernel buffer.  Currently, uiomove accepts a signed (n) argument, which
140 * is almost definitely a bad thing, so we catch that here as well.  We
141 * return a runtime failure, but it might be desirable to generate a runtime
142 * assertion failure instead.
145 uiomove_frombuf(void *buf, int buflen, struct uio *uio)
147 unsigned int offset, n;
149 if (uio->uio_offset < 0 || uio->uio_resid < 0 ||
150 (offset = uio->uio_offset) != uio->uio_offset)
151 return (EINVAL);
152 if (buflen <= 0 || offset >= buflen)
153 return (0);
154 if ((n = buflen - offset) > INT_MAX)
155 return (EINVAL);
156 return (uiomove((char *)buf + offset, n, uio));
161 uiomoveco(caddr_t cp, int n, struct uio *uio, struct vm_object *obj)
163 struct iovec *iov;
164 u_int cnt;
165 int error;
166 int baseticks = ticks;
168 KASSERT(uio->uio_rw == UIO_READ || uio->uio_rw == UIO_WRITE,
169 ("uiomoveco: mode"));
170 KASSERT(uio->uio_segflg != UIO_USERSPACE || uio->uio_td == curthread,
171 ("uiomoveco proc"));
173 while (n > 0 && uio->uio_resid) {
174 iov = uio->uio_iov;
175 cnt = iov->iov_len;
176 if (cnt == 0) {
177 uio->uio_iov++;
178 uio->uio_iovcnt--;
179 continue;
181 if (cnt > n)
182 cnt = n;
184 switch (uio->uio_segflg) {
186 case UIO_USERSPACE:
187 if (ticks - baseticks >= hogticks) {
188 uio_yield();
189 baseticks = ticks;
191 if (uio->uio_rw == UIO_READ) {
192 error = copyout(cp, iov->iov_base, cnt);
193 } else {
194 error = copyin(iov->iov_base, cp, cnt);
196 if (error)
197 return (error);
198 break;
200 case UIO_SYSSPACE:
201 if (uio->uio_rw == UIO_READ)
202 bcopy((caddr_t)cp, iov->iov_base, cnt);
203 else
204 bcopy(iov->iov_base, (caddr_t)cp, cnt);
205 break;
206 case UIO_NOCOPY:
207 break;
209 iov->iov_base = (char *)iov->iov_base + cnt;
210 iov->iov_len -= cnt;
211 uio->uio_resid -= cnt;
212 uio->uio_offset += cnt;
213 cp += cnt;
214 n -= cnt;
216 return (0);
220 * Give next character to user as result of read.
223 ureadc(int c, struct uio *uio)
225 struct iovec *iov;
226 char *iov_base;
228 again:
229 if (uio->uio_iovcnt == 0 || uio->uio_resid == 0)
230 panic("ureadc");
231 iov = uio->uio_iov;
232 if (iov->iov_len == 0) {
233 uio->uio_iovcnt--;
234 uio->uio_iov++;
235 goto again;
237 switch (uio->uio_segflg) {
239 case UIO_USERSPACE:
240 if (subyte(iov->iov_base, c) < 0)
241 return (EFAULT);
242 break;
244 case UIO_SYSSPACE:
245 iov_base = iov->iov_base;
246 *iov_base = c;
247 iov->iov_base = iov_base;
248 break;
250 case UIO_NOCOPY:
251 break;
253 iov->iov_base = (char *)iov->iov_base + 1;
254 iov->iov_len--;
255 uio->uio_resid--;
256 uio->uio_offset++;
257 return (0);
261 * General routine to allocate a hash table. Make the hash table size a
262 * power of 2 greater or equal to the number of elements requested, and
263 * store the masking value in *hashmask.
265 void *
266 hashinit(int elements, struct malloc_type *type, u_long *hashmask)
268 long hashsize;
269 LIST_HEAD(generic, generic) *hashtbl;
270 int i;
272 if (elements <= 0)
273 panic("hashinit: bad elements");
274 for (hashsize = 2; hashsize < elements; hashsize <<= 1)
275 continue;
276 hashtbl = kmalloc((u_long)hashsize * sizeof(*hashtbl), type, M_WAITOK);
277 for (i = 0; i < hashsize; i++)
278 LIST_INIT(&hashtbl[i]);
279 *hashmask = hashsize - 1;
280 return (hashtbl);
283 static int primes[] = { 1, 13, 31, 61, 127, 251, 509, 761, 1021, 1531, 2039,
284 2557, 3067, 3583, 4093, 4603, 5119, 5623, 6143, 6653,
285 7159, 7673, 8191, 12281, 16381, 24571, 32749 };
286 #define NPRIMES (sizeof(primes) / sizeof(primes[0]))
289 * General routine to allocate a prime number sized hash table.
291 void *
292 phashinit(int elements, struct malloc_type *type, u_long *nentries)
294 long hashsize;
295 LIST_HEAD(generic, generic) *hashtbl;
296 int i;
298 if (elements <= 0)
299 panic("phashinit: bad elements");
300 for (i = 1, hashsize = primes[1]; hashsize <= elements;) {
301 i++;
302 if (i == NPRIMES)
303 break;
304 hashsize = primes[i];
306 hashsize = primes[i - 1];
307 hashtbl = kmalloc((u_long)hashsize * sizeof(*hashtbl), type, M_WAITOK);
308 for (i = 0; i < hashsize; i++)
309 LIST_INIT(&hashtbl[i]);
310 *nentries = hashsize;
311 return (hashtbl);
315 * Copyin an iovec. If the iovec array fits, use the preallocated small
316 * iovec structure. If it is too big, dynamically allocate an iovec array
317 * of sufficient size.
319 * MPSAFE
322 iovec_copyin(struct iovec *uiov, struct iovec **kiov, struct iovec *siov,
323 size_t iov_cnt, int *iov_len)
325 struct iovec *iovp;
326 int error, i;
328 if (iov_cnt > UIO_MAXIOV)
329 return EMSGSIZE;
330 if (iov_cnt > UIO_SMALLIOV) {
331 MALLOC(*kiov, struct iovec *, sizeof(struct iovec) * iov_cnt,
332 M_IOV, M_WAITOK);
333 } else {
334 *kiov = siov;
336 error = copyin(uiov, *kiov, iov_cnt * sizeof(struct iovec));
337 if (error == 0) {
338 *iov_len = 0;
339 for (i = 0, iovp = *kiov; i < iov_cnt; i++, iovp++) {
341 * Check for both *iov_len overflows and out of
342 * range iovp->iov_len's. We limit to the
343 * capabilities of signed integers.
345 if (*iov_len + (int)iovp->iov_len < *iov_len)
346 error = EINVAL;
347 *iov_len += (int)iovp->iov_len;
350 if (error)
351 iovec_free(kiov, siov);
352 return (error);
357 * Copyright (c) 2004 Alan L. Cox <alc@cs.rice.edu>
358 * Copyright (c) 1982, 1986, 1991, 1993
359 * The Regents of the University of California. All rights reserved.
360 * (c) UNIX System Laboratories, Inc.
361 * All or some portions of this file are derived from material licensed
362 * to the University of California by American Telephone and Telegraph
363 * Co. or Unix System Laboratories, Inc. and are reproduced herein with
364 * the permission of UNIX System Laboratories, Inc.
366 * Redistribution and use in source and binary forms, with or without
367 * modification, are permitted provided that the following conditions
368 * are met:
369 * 1. Redistributions of source code must retain the above copyright
370 * notice, this list of conditions and the following disclaimer.
371 * 2. Redistributions in binary form must reproduce the above copyright
372 * notice, this list of conditions and the following disclaimer in the
373 * documentation and/or other materials provided with the distribution.
374 * 4. Neither the name of the University nor the names of its contributors
375 * may be used to endorse or promote products derived from this software
376 * without specific prior written permission.
378 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
379 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
380 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
381 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
382 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
383 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
384 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
385 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
386 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
387 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
388 * SUCH DAMAGE.
390 * @(#)kern_subr.c 8.3 (Berkeley) 1/21/94
391 * $FreeBSD: src/sys/i386/i386/uio_machdep.c,v 1.1 2004/03/21 20:28:36 alc Exp $
395 * Implement uiomove(9) from physical memory using sf_bufs to reduce
396 * the creation and destruction of ephemeral mappings.
399 uiomove_fromphys(vm_page_t *ma, vm_offset_t offset, int n, struct uio *uio)
401 struct sf_buf *sf;
402 struct thread *td = curthread;
403 struct iovec *iov;
404 void *cp;
405 vm_offset_t page_offset;
406 vm_page_t m;
407 size_t cnt;
408 int error = 0;
409 int save = 0;
411 KASSERT(uio->uio_rw == UIO_READ || uio->uio_rw == UIO_WRITE,
412 ("uiomove_fromphys: mode"));
413 KASSERT(uio->uio_segflg != UIO_USERSPACE || uio->uio_td == curthread,
414 ("uiomove_fromphys proc"));
416 crit_enter();
417 save = td->td_flags & TDF_DEADLKTREAT;
418 td->td_flags |= TDF_DEADLKTREAT;
419 crit_exit();
421 while (n > 0 && uio->uio_resid) {
422 iov = uio->uio_iov;
423 cnt = iov->iov_len;
424 if (cnt == 0) {
425 uio->uio_iov++;
426 uio->uio_iovcnt--;
427 continue;
429 if (cnt > n)
430 cnt = n;
431 page_offset = offset & PAGE_MASK;
432 cnt = min(cnt, PAGE_SIZE - page_offset);
433 m = ma[offset >> PAGE_SHIFT];
434 sf = sf_buf_alloc(m, SFB_CPUPRIVATE);
435 cp = (char *)sf_buf_kva(sf) + page_offset;
436 switch (uio->uio_segflg) {
437 case UIO_USERSPACE:
439 * note: removed uioyield (it was the wrong place to
440 * put it).
442 if (uio->uio_rw == UIO_READ)
443 error = copyout(cp, iov->iov_base, cnt);
444 else
445 error = copyin(iov->iov_base, cp, cnt);
446 if (error) {
447 sf_buf_free(sf);
448 goto out;
450 break;
451 case UIO_SYSSPACE:
452 if (uio->uio_rw == UIO_READ)
453 bcopy(cp, iov->iov_base, cnt);
454 else
455 bcopy(iov->iov_base, cp, cnt);
456 break;
457 case UIO_NOCOPY:
458 break;
460 sf_buf_free(sf);
461 iov->iov_base = (char *)iov->iov_base + cnt;
462 iov->iov_len -= cnt;
463 uio->uio_resid -= cnt;
464 uio->uio_offset += cnt;
465 offset += cnt;
466 n -= cnt;
468 out:
469 if (save == 0) {
470 crit_enter();
471 td->td_flags &= ~TDF_DEADLKTREAT;
472 crit_exit();
474 return (error);