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 * Copyright (c) 2014 The FreeBSD Foundation
12 * Portions of this software were developed by Konstantin Belousov
13 * under sponsorship from the FreeBSD Foundation.
15 * Redistribution and use in source and binary forms, with or without
16 * modification, are permitted provided that the following conditions
18 * 1. Redistributions of source code must retain the above copyright
19 * notice, this list of conditions and the following disclaimer.
20 * 2. Redistributions in binary form must reproduce the above copyright
21 * notice, this list of conditions and the following disclaimer in the
22 * documentation and/or other materials provided with the distribution.
23 * 4. Neither the name of the University nor the names of its contributors
24 * may be used to endorse or promote products derived from this software
25 * without specific prior written permission.
27 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
28 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
29 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
30 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
31 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
32 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
33 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
34 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
35 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
36 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
39 * @(#)kern_subr.c 8.3 (Berkeley) 1/21/94
42 #include <sys/cdefs.h>
43 __FBSDID("$FreeBSD$");
45 #include <sys/param.h>
46 #include <sys/systm.h>
47 #include <sys/kernel.h>
48 #include <sys/limits.h>
52 #include <sys/resourcevar.h>
53 #include <sys/rwlock.h>
54 #include <sys/sched.h>
55 #include <sys/sysctl.h>
56 #include <sys/vnode.h>
59 #include <vm/vm_param.h>
60 #include <vm/vm_extern.h>
61 #include <vm/vm_page.h>
62 #include <vm/vm_pageout.h>
63 #include <vm/vm_map.h>
65 SYSCTL_INT(_kern, KERN_IOV_MAX, iov_max, CTLFLAG_RD, SYSCTL_NULL_INT_PTR, UIO_MAXIOV,
66 "Maximum number of elements in an I/O vector; sysconf(_SC_IOV_MAX)");
68 static int uiomove_faultflag(void *cp, int n, struct uio *uio, int nofault);
71 copyin_nofault(const void *udaddr, void *kaddr, size_t len)
75 save = vm_fault_disable_pagefaults();
76 error = copyin(udaddr, kaddr, len);
77 vm_fault_enable_pagefaults(save);
82 copyout_nofault(const void *kaddr, void *udaddr, size_t len)
86 save = vm_fault_disable_pagefaults();
87 error = copyout(kaddr, udaddr, len);
88 vm_fault_enable_pagefaults(save);
92 #define PHYS_PAGE_COUNT(len) (howmany(len, PAGE_SIZE) + 1)
95 physcopyin(void *src, vm_paddr_t dst, size_t len)
97 vm_page_t m[PHYS_PAGE_COUNT(len)];
102 iov[0].iov_base = src;
103 iov[0].iov_len = len;
108 uio.uio_segflg = UIO_SYSSPACE;
109 uio.uio_rw = UIO_WRITE;
110 for (i = 0; i < PHYS_PAGE_COUNT(len); i++, dst += PAGE_SIZE)
111 m[i] = PHYS_TO_VM_PAGE(dst);
112 return (uiomove_fromphys(m, dst & PAGE_MASK, len, &uio));
116 physcopyout(vm_paddr_t src, void *dst, size_t len)
118 vm_page_t m[PHYS_PAGE_COUNT(len)];
123 iov[0].iov_base = dst;
124 iov[0].iov_len = len;
129 uio.uio_segflg = UIO_SYSSPACE;
130 uio.uio_rw = UIO_READ;
131 for (i = 0; i < PHYS_PAGE_COUNT(len); i++, src += PAGE_SIZE)
132 m[i] = PHYS_TO_VM_PAGE(src);
133 return (uiomove_fromphys(m, src & PAGE_MASK, len, &uio));
136 #undef PHYS_PAGE_COUNT
139 uiomove(void *cp, int n, struct uio *uio)
142 return (uiomove_faultflag(cp, n, uio, 0));
146 uiomove_nofault(void *cp, int n, struct uio *uio)
149 return (uiomove_faultflag(cp, n, uio, 1));
153 uiomove_faultflag(void *cp, int n, struct uio *uio, int nofault)
158 int error, newflags, save;
163 KASSERT(uio->uio_rw == UIO_READ || uio->uio_rw == UIO_WRITE,
165 KASSERT(uio->uio_segflg != UIO_USERSPACE || uio->uio_td == td,
168 WITNESS_WARN(WARN_GIANTOK | WARN_SLEEPOK, NULL,
169 "Calling uiomove()");
171 /* XXX does it make a sense to set TDP_DEADLKTREAT for UIO_SYSSPACE ? */
172 newflags = TDP_DEADLKTREAT;
173 if (uio->uio_segflg == UIO_USERSPACE && nofault) {
175 * Fail if a non-spurious page fault occurs.
177 newflags |= TDP_NOFAULTING | TDP_RESETSPUR;
179 save = curthread_pflags_set(newflags);
181 while (n > 0 && uio->uio_resid) {
192 switch (uio->uio_segflg) {
196 if (uio->uio_rw == UIO_READ)
197 error = copyout(cp, iov->iov_base, cnt);
199 error = copyin(iov->iov_base, cp, cnt);
205 if (uio->uio_rw == UIO_READ)
206 bcopy(cp, iov->iov_base, cnt);
208 bcopy(iov->iov_base, cp, cnt);
213 iov->iov_base = (char *)iov->iov_base + cnt;
215 uio->uio_resid -= cnt;
216 uio->uio_offset += cnt;
217 cp = (char *)cp + cnt;
221 curthread_pflags_restore(save);
226 * Wrapper for uiomove() that validates the arguments against a known-good
227 * kernel buffer. Currently, uiomove accepts a signed (n) argument, which
228 * is almost definitely a bad thing, so we catch that here as well. We
229 * return a runtime failure, but it might be desirable to generate a runtime
230 * assertion failure instead.
233 uiomove_frombuf(void *buf, int buflen, struct uio *uio)
237 if (uio->uio_offset < 0 || uio->uio_resid < 0 ||
238 (offset = uio->uio_offset) != uio->uio_offset)
240 if (buflen <= 0 || offset >= buflen)
242 if ((n = buflen - offset) > IOSIZE_MAX)
244 return (uiomove((char *)buf + offset, n, uio));
248 * Give next character to user as result of read.
251 ureadc(int c, struct uio *uio)
256 WITNESS_WARN(WARN_GIANTOK | WARN_SLEEPOK, NULL,
260 if (uio->uio_iovcnt == 0 || uio->uio_resid == 0)
263 if (iov->iov_len == 0) {
268 switch (uio->uio_segflg) {
271 if (subyte(iov->iov_base, c) < 0)
276 iov_base = iov->iov_base;
283 iov->iov_base = (char *)iov->iov_base + 1;
291 copyinfrom(const void * __restrict src, void * __restrict dst, size_t len,
298 error = copyin(src, dst, len);
301 bcopy(src, dst, len);
304 panic("copyinfrom: bad seg %d\n", seg);
310 copyinstrfrom(const void * __restrict src, void * __restrict dst, size_t len,
311 size_t * __restrict copied, int seg)
317 error = copyinstr(src, dst, len, copied);
320 error = copystr(src, dst, len, copied);
323 panic("copyinstrfrom: bad seg %d\n", seg);
329 copyiniov(const struct iovec *iovp, u_int iovcnt, struct iovec **iov, int error)
334 if (iovcnt > UIO_MAXIOV)
336 iovlen = iovcnt * sizeof (struct iovec);
337 *iov = malloc(iovlen, M_IOV, M_WAITOK);
338 error = copyin(iovp, *iov, iovlen);
347 copyinuio(const struct iovec *iovp, u_int iovcnt, struct uio **uiop)
355 if (iovcnt > UIO_MAXIOV)
357 iovlen = iovcnt * sizeof (struct iovec);
358 uio = malloc(iovlen + sizeof *uio, M_IOV, M_WAITOK);
359 iov = (struct iovec *)(uio + 1);
360 error = copyin(iovp, iov, iovlen);
366 uio->uio_iovcnt = iovcnt;
367 uio->uio_segflg = UIO_USERSPACE;
368 uio->uio_offset = -1;
370 for (i = 0; i < iovcnt; i++) {
371 if (iov->iov_len > IOSIZE_MAX - uio->uio_resid) {
375 uio->uio_resid += iov->iov_len;
383 cloneuio(struct uio *uiop)
388 iovlen = uiop->uio_iovcnt * sizeof (struct iovec);
389 uio = malloc(iovlen + sizeof *uio, M_IOV, M_WAITOK);
391 uio->uio_iov = (struct iovec *)(uio + 1);
392 bcopy(uiop->uio_iov, uio->uio_iov, iovlen);
397 * Map some anonymous memory in user space of size sz, rounded up to the page
401 copyout_map(struct thread *td, vm_offset_t *addr, size_t sz)
407 vms = td->td_proc->p_vmspace;
410 * Map somewhere after heap in process memory.
412 PROC_LOCK(td->td_proc);
413 *addr = round_page((vm_offset_t)vms->vm_daddr +
414 lim_max(td->td_proc, RLIMIT_DATA));
415 PROC_UNLOCK(td->td_proc);
417 /* round size up to page boundry */
418 size = (vm_size_t)round_page(sz);
420 error = vm_mmap(&vms->vm_map, addr, size, PROT_READ | PROT_WRITE,
421 VM_PROT_ALL, MAP_PRIVATE | MAP_ANON, OBJT_DEFAULT, NULL, 0);
427 * Unmap memory in user space.
430 copyout_unmap(struct thread *td, vm_offset_t addr, size_t sz)
438 map = &td->td_proc->p_vmspace->vm_map;
439 size = (vm_size_t)round_page(sz);
441 if (vm_map_remove(map, addr, addr + size) != KERN_SUCCESS)
449 * XXXKIB The temporal implementation of fue*() functions which do not
450 * handle usermode -1 properly, mixing it with the fault code. Keep
451 * this until MD code is written. Currently sparc64, mips and arm do
452 * not have proper implementation.
456 fueword(volatile const void *base, long *val)
468 fueword32(volatile const void *base, int32_t *val)
472 res = fuword32(base);
481 fueword64(volatile const void *base, int64_t *val)
485 res = fuword64(base);
494 casueword32(volatile uint32_t *base, uint32_t oldval, uint32_t *oldvalp,
499 ov = casuword32(base, oldval, newval);
507 casueword(volatile u_long *p, u_long oldval, u_long *oldvalp, u_long newval)
511 ov = casuword(p, oldval, newval);
517 #else /* NO_FUEWORD */
519 fuword32(volatile const void *addr)
524 rv = fueword32(addr, &val);
525 return (rv == -1 ? -1 : val);
530 fuword64(volatile const void *addr)
535 rv = fueword64(addr, &val);
536 return (rv == -1 ? -1 : val);
541 fuword(volatile const void *addr)
546 rv = fueword(addr, &val);
547 return (rv == -1 ? -1 : val);
551 casuword32(volatile uint32_t *addr, uint32_t old, uint32_t new)
556 rv = casueword32(addr, old, &val, new);
557 return (rv == -1 ? -1 : val);
561 casuword(volatile u_long *addr, u_long old, u_long new)
566 rv = casueword(addr, old, &val, new);
567 return (rv == -1 ? -1 : val);
570 #endif /* NO_FUEWORD */