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
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
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 $
44 #include <sys/param.h>
45 #include <sys/systm.h>
46 #include <sys/kernel.h>
48 #include <sys/malloc.h>
50 #include <sys/resourcevar.h>
51 #include <sys/sysctl.h>
53 #include <sys/vnode.h>
54 #include <sys/thread2.h>
55 #include <machine/limits.h>
57 #include <cpu/lwbuf.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 * The syscall interface is responsible for limiting the length to
73 * ssize_t for things like read() or write() which return the bytes
74 * read or written as ssize_t. These functions work with unsigned
78 uiomove(caddr_t cp, size_t n, struct uio *uio)
80 thread_t td = curthread;
86 KASSERT(uio->uio_rw == UIO_READ || uio->uio_rw == UIO_WRITE,
88 KASSERT(uio->uio_segflg != UIO_USERSPACE || uio->uio_td == td,
92 save = td->td_flags & TDF_DEADLKTREAT;
93 td->td_flags |= TDF_DEADLKTREAT;
96 while (n > 0 && uio->uio_resid) {
107 switch (uio->uio_segflg) {
111 if (uio->uio_rw == UIO_READ)
112 error = copyout(cp, iov->iov_base, cnt);
114 error = copyin(iov->iov_base, cp, cnt);
120 if (uio->uio_rw == UIO_READ)
121 bcopy((caddr_t)cp, iov->iov_base, cnt);
123 bcopy(iov->iov_base, (caddr_t)cp, cnt);
128 iov->iov_base = (char *)iov->iov_base + cnt;
130 uio->uio_resid -= cnt;
131 uio->uio_offset += cnt;
136 td->td_flags = (td->td_flags & ~TDF_DEADLKTREAT) | save;
142 * This is the same as uiomove() except (cp, n) is within the bounds of
143 * the passed, locked buffer. Under certain circumstances a VM fault
144 * occuring with a locked buffer held can result in a deadlock or an
145 * attempt to recursively lock the buffer.
147 * This procedure deals with these cases.
149 * If the buffer represents a regular file, is B_CACHE, but the last VM page
150 * is not fully valid we fix-up the last VM page. This should handle the
151 * recursive lock issue.
153 * Deadlocks are another issue. We are holding the vp and the bp locked
154 * and could deadlock against a different vp and/or bp if another thread is
155 * trying to access us while we accessing it. The only solution here is
156 * to release the bp and vnode lock and do the uio to/from a system buffer,
157 * then regain the locks and copyback (if applicable). XXX TODO.
160 uiomovebp(struct buf *bp, caddr_t cp, size_t n, struct uio *uio)
165 if (bp->b_vp && bp->b_vp->v_type == VREG &&
166 (bp->b_flags & B_CACHE) &&
167 (count = bp->b_xio.xio_npages) != 0 &&
168 (m = bp->b_xio.xio_pages[count-1])->valid != VM_PAGE_BITS_ALL) {
169 vm_page_zero_invalid(m, TRUE);
171 return (uiomove(cp, n, uio));
175 * Like uiomove() but copies zero-fill. Only allowed for UIO_READ,
176 * for obvious reasons.
179 uiomovez(size_t n, struct uio *uio)
185 KASSERT(uio->uio_rw == UIO_READ, ("uiomovez: mode"));
186 KASSERT(uio->uio_segflg != UIO_USERSPACE || uio->uio_td == curthread,
189 while (n > 0 && uio->uio_resid) {
200 switch (uio->uio_segflg) {
202 error = copyout(ZeroPage, iov->iov_base, cnt);
207 bzero(iov->iov_base, cnt);
212 iov->iov_base = (char *)iov->iov_base + cnt;
214 uio->uio_resid -= cnt;
215 uio->uio_offset += cnt;
222 * Wrapper for uiomove() that validates the arguments against a known-good
223 * kernel buffer. This function automatically indexes the buffer by
224 * uio_offset and handles all range checking.
227 uiomove_frombuf(void *buf, size_t buflen, struct uio *uio)
231 offset = (size_t)uio->uio_offset;
232 if ((off_t)offset != uio->uio_offset)
234 if (buflen == 0 || offset >= buflen)
236 return (uiomove((char *)buf + offset, buflen - offset, uio));
240 * Give next character to user as result of read.
243 ureadc(int c, struct uio *uio)
249 if (uio->uio_iovcnt == 0 || uio->uio_resid == 0)
252 if (iov->iov_len == 0) {
257 switch (uio->uio_segflg) {
260 if (subyte(iov->iov_base, c) < 0)
265 iov_base = iov->iov_base;
267 iov->iov_base = iov_base;
273 iov->iov_base = (char *)iov->iov_base + 1;
281 * General routine to allocate a hash table. Make the hash table size a
282 * power of 2 greater or equal to the number of elements requested, and
283 * store the masking value in *hashmask.
286 hashinit(int elements, struct malloc_type *type, u_long *hashmask)
289 LIST_HEAD(generic, generic) *hashtbl;
293 panic("hashinit: bad elements");
294 for (hashsize = 2; hashsize < elements; hashsize <<= 1)
296 hashtbl = kmalloc((u_long)hashsize * sizeof(*hashtbl), type, M_WAITOK);
297 for (i = 0; i < hashsize; i++)
298 LIST_INIT(&hashtbl[i]);
299 *hashmask = hashsize - 1;
304 * This is a newer version which allocates a hash table of structures.
306 * The returned array will be zero'd. The caller is responsible for
307 * initializing the structures.
310 hashinit_ext(int elements, size_t size, struct malloc_type *type,
317 panic("hashinit: bad elements");
318 for (hashsize = 2; hashsize < elements; hashsize <<= 1)
320 hashtbl = kmalloc((size_t)hashsize * size, type, M_WAITOK | M_ZERO);
321 *hashmask = hashsize - 1;
325 static int primes[] = { 1, 13, 31, 61, 127, 251, 509, 761, 1021, 1531, 2039,
326 2557, 3067, 3583, 4093, 4603, 5119, 5623, 6143, 6653,
327 7159, 7673, 8191, 12281, 16381, 24571, 32749 };
328 #define NPRIMES NELEM(primes)
331 * General routine to allocate a prime number sized hash table.
334 phashinit(int elements, struct malloc_type *type, u_long *nentries)
337 LIST_HEAD(generic, generic) *hashtbl;
341 panic("phashinit: bad elements");
342 for (i = 1, hashsize = primes[1]; hashsize <= elements;) {
346 hashsize = primes[i];
348 hashsize = primes[i - 1];
349 hashtbl = kmalloc((u_long)hashsize * sizeof(*hashtbl), type, M_WAITOK);
350 for (i = 0; i < hashsize; i++)
351 LIST_INIT(&hashtbl[i]);
352 *nentries = hashsize;
357 * This is a newer version which allocates a hash table of structures
358 * in a prime-number size.
360 * The returned array will be zero'd. The caller is responsible for
361 * initializing the structures.
364 phashinit_ext(int elements, size_t size, struct malloc_type *type,
372 panic("phashinit: bad elements");
373 for (i = 1, hashsize = primes[1]; hashsize <= elements;) {
377 hashsize = primes[i];
379 hashsize = primes[i - 1];
380 hashtbl = kmalloc((size_t)hashsize * size, type, M_WAITOK | M_ZERO);
381 *nentries = hashsize;
386 * Copyin an iovec. If the iovec array fits, use the preallocated small
387 * iovec structure. If it is too big, dynamically allocate an iovec array
388 * of sufficient size.
393 iovec_copyin(struct iovec *uiov, struct iovec **kiov, struct iovec *siov,
394 size_t iov_cnt, size_t *iov_len)
400 if (iov_cnt > UIO_MAXIOV)
402 if (iov_cnt > UIO_SMALLIOV) {
403 *kiov = kmalloc(sizeof(struct iovec) * iov_cnt, M_IOV,
408 error = copyin(uiov, *kiov, iov_cnt * sizeof(struct iovec));
411 for (i = 0, iovp = *kiov; i < iov_cnt; i++, iovp++) {
413 * Check for both *iov_len overflows and out of
414 * range iovp->iov_len's. We limit to the
415 * capabilities of signed integers.
417 * GCC4 - overflow check opt requires assign/test.
419 len = *iov_len + iovp->iov_len;
427 * From userland disallow iovec's which exceed the sized size
428 * limit as the system calls return ssize_t.
430 * NOTE: Internal kernel interfaces can handle the unsigned
433 if (error == 0 && (ssize_t)*iov_len < 0)
437 iovec_free(kiov, siov);
443 * Copyright (c) 2004 Alan L. Cox <alc@cs.rice.edu>
444 * Copyright (c) 1982, 1986, 1991, 1993
445 * The Regents of the University of California. All rights reserved.
446 * (c) UNIX System Laboratories, Inc.
447 * All or some portions of this file are derived from material licensed
448 * to the University of California by American Telephone and Telegraph
449 * Co. or Unix System Laboratories, Inc. and are reproduced herein with
450 * the permission of UNIX System Laboratories, Inc.
452 * Redistribution and use in source and binary forms, with or without
453 * modification, are permitted provided that the following conditions
455 * 1. Redistributions of source code must retain the above copyright
456 * notice, this list of conditions and the following disclaimer.
457 * 2. Redistributions in binary form must reproduce the above copyright
458 * notice, this list of conditions and the following disclaimer in the
459 * documentation and/or other materials provided with the distribution.
460 * 4. Neither the name of the University nor the names of its contributors
461 * may be used to endorse or promote products derived from this software
462 * without specific prior written permission.
464 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
465 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
466 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
467 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
468 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
469 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
470 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
471 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
472 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
473 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
476 * @(#)kern_subr.c 8.3 (Berkeley) 1/21/94
477 * $FreeBSD: src/sys/i386/i386/uio_machdep.c,v 1.1 2004/03/21 20:28:36 alc Exp $
481 * Implement uiomove(9) from physical memory using lwbuf's to reduce
482 * the creation and destruction of ephemeral mappings.
485 uiomove_fromphys(vm_page_t *ma, vm_offset_t offset, size_t n, struct uio *uio)
487 struct lwbuf lwb_cache;
489 struct thread *td = curthread;
492 vm_offset_t page_offset;
498 KASSERT(uio->uio_rw == UIO_READ || uio->uio_rw == UIO_WRITE,
499 ("uiomove_fromphys: mode"));
500 KASSERT(uio->uio_segflg != UIO_USERSPACE || uio->uio_td == curthread,
501 ("uiomove_fromphys proc"));
504 save = td->td_flags & TDF_DEADLKTREAT;
505 td->td_flags |= TDF_DEADLKTREAT;
508 while (n > 0 && uio->uio_resid) {
518 page_offset = offset & PAGE_MASK;
519 cnt = min(cnt, PAGE_SIZE - page_offset);
520 m = ma[offset >> PAGE_SHIFT];
521 lwb = lwbuf_alloc(m, &lwb_cache);
522 cp = (char *)lwbuf_kva(lwb) + page_offset;
523 switch (uio->uio_segflg) {
526 * note: removed uioyield (it was the wrong place to
529 if (uio->uio_rw == UIO_READ)
530 error = copyout(cp, iov->iov_base, cnt);
532 error = copyin(iov->iov_base, cp, cnt);
539 if (uio->uio_rw == UIO_READ)
540 bcopy(cp, iov->iov_base, cnt);
542 bcopy(iov->iov_base, cp, cnt);
548 iov->iov_base = (char *)iov->iov_base + cnt;
550 uio->uio_resid -= cnt;
551 uio->uio_offset += cnt;
558 td->td_flags &= ~TDF_DEADLKTREAT;