2 * Copyright (c) 1988 University of Utah.
3 * Copyright (c) 1982, 1986, 1990 The Regents of the University of California.
6 * This code is derived from software contributed to Berkeley by
7 * the Systems Programming Group of the University of Utah Computer
8 * Science Department, and code derived from software contributed to
9 * Berkeley by William Jolitz.
11 * Redistribution and use in source and binary forms, with or without
12 * modification, are permitted provided that the following conditions
14 * 1. Redistributions of source code must retain the above copyright
15 * notice, this list of conditions and the following disclaimer.
16 * 2. Redistributions in binary form must reproduce the above copyright
17 * notice, this list of conditions and the following disclaimer in the
18 * documentation and/or other materials provided with the distribution.
19 * 4. Neither the name of the University nor the names of its contributors
20 * may be used to endorse or promote products derived from this software
21 * without specific prior written permission.
23 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
24 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
25 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
26 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
27 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
28 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
29 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
30 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
31 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
32 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
35 * from: Utah $Hdr: mem.c 1.13 89/10/08$
36 * from: @(#)mem.c 7.2 (Berkeley) 5/9/91
37 * $FreeBSD: src/sys/i386/i386/mem.c,v 1.79.2.9 2003/01/04 22:58:01 njl Exp $
44 #include <sys/param.h>
45 #include <sys/systm.h>
48 #include <sys/fcntl.h>
49 #include <sys/filio.h>
50 #include <sys/kernel.h>
51 #include <sys/malloc.h>
52 #include <sys/memrange.h>
55 #include <sys/random.h>
56 #include <sys/signalvar.h>
58 #include <sys/vnode.h>
60 #include <sys/signal2.h>
61 #include <sys/mplock2.h>
65 #include <vm/vm_extern.h>
68 static d_open_t mmopen;
69 static d_close_t mmclose;
70 static d_read_t mmread;
71 static d_write_t mmwrite;
72 static d_ioctl_t mmioctl;
73 static d_mmap_t memmmap;
74 static d_kqfilter_t mmkqfilter;
77 static struct dev_ops mem_ops = {
78 { "mem", 0, D_MPSAFE },
84 .d_kqfilter = mmkqfilter,
90 static cdev_t zerodev = NULL;
92 MALLOC_DEFINE(M_MEMDESC, "memdesc", "memory range descriptors");
93 static int mem_ioctl (cdev_t, u_long, caddr_t, int, struct ucred *);
94 static int random_ioctl (cdev_t, u_long, caddr_t, int, struct ucred *);
96 struct mem_range_softc mem_range_softc;
100 mmopen(struct dev_open_args *ap)
102 cdev_t dev = ap->a_head.a_dev;
105 switch (minor(dev)) {
108 if (ap->a_oflags & FWRITE) {
109 if (securelevel > 0 || kernel_mem_readonly)
115 error = priv_check_cred(ap->a_cred, PRIV_ROOT, 0);
118 if (securelevel > 0 || kernel_mem_readonly) {
122 error = cpu_set_iopl();
132 mmclose(struct dev_close_args *ap)
134 cdev_t dev = ap->a_head.a_dev;
137 switch (minor(dev)) {
139 error = cpu_clr_iopl();
150 mmrw(cdev_t dev, struct uio *uio, int flags)
160 while (uio->uio_resid > 0 && error == 0) {
162 if (iov->iov_len == 0) {
165 if (uio->uio_iovcnt < 0)
169 switch (minor(dev)) {
172 * minor device 0 is physical memory, /dev/mem
175 v &= ~(long)PAGE_MASK;
176 pmap_kenter((vm_offset_t)ptvmmap, v);
177 o = (int)uio->uio_offset & PAGE_MASK;
178 c = (u_int)(PAGE_SIZE - ((uintptr_t)iov->iov_base & PAGE_MASK));
179 c = min(c, (u_int)(PAGE_SIZE - o));
180 c = min(c, (u_int)iov->iov_len);
181 error = uiomove((caddr_t)&ptvmmap[o], (int)c, uio);
182 pmap_kremove((vm_offset_t)ptvmmap);
187 * minor device 1 is kernel memory, /dev/kmem
189 vm_offset_t saddr, eaddr;
195 * Make sure that all of the pages are currently
196 * resident so that we don't create any zero-fill
199 saddr = trunc_page(uio->uio_offset);
200 eaddr = round_page(uio->uio_offset + c);
205 * Make sure the kernel addresses are mapped.
206 * platform_direct_mapped() can be used to bypass
207 * default mapping via the page table (virtual kernels
208 * contain a lot of out-of-band data).
211 if (uio->uio_rw != UIO_READ)
212 prot |= VM_PROT_WRITE;
213 error = kvm_access_check(saddr, eaddr, prot);
216 error = uiomove((caddr_t)(vm_offset_t)uio->uio_offset,
222 * minor device 2 (/dev/null) is EOF/RATHOLE
224 if (uio->uio_rw == UIO_READ)
230 * minor device 3 (/dev/random) is source of filth
231 * on read, seeder on write
234 buf = kmalloc(PAGE_SIZE, M_TEMP, M_WAITOK);
235 c = min(iov->iov_len, PAGE_SIZE);
236 if (uio->uio_rw == UIO_WRITE) {
237 error = uiomove(buf, (int)c, uio);
239 error = add_buffer_randomness(buf, c);
241 poolsize = read_random(buf, c);
245 if ((flags & IO_NDELAY) != 0)
246 return (EWOULDBLOCK);
249 c = min(c, poolsize);
250 error = uiomove(buf, (int)c, uio);
255 * minor device 4 (/dev/urandom) is source of muck
256 * on read, writes are disallowed.
258 c = min(iov->iov_len, PAGE_SIZE);
259 if (uio->uio_rw == UIO_WRITE) {
263 if (CURSIG(curthread->td_lwp) != 0) {
265 * Use tsleep() to get the error code right.
266 * It should return immediately.
268 error = tsleep(&rand_bolt, PCATCH, "urand", 1);
269 if (error != 0 && error != EWOULDBLOCK)
273 buf = kmalloc(PAGE_SIZE, M_TEMP, M_WAITOK);
274 poolsize = read_random_unlimited(buf, c);
275 c = min(c, poolsize);
276 error = uiomove(buf, (int)c, uio);
280 * minor device 12 (/dev/zero) is source of nulls
281 * on read, write are disallowed.
283 if (uio->uio_rw == UIO_WRITE) {
288 zbuf = (caddr_t)kmalloc(PAGE_SIZE, M_TEMP,
291 c = min(iov->iov_len, PAGE_SIZE);
292 error = uiomove(zbuf, (int)c, uio);
299 iov->iov_base = (char *)iov->iov_base + c;
301 uio->uio_offset += c;
310 mmread(struct dev_read_args *ap)
312 return(mmrw(ap->a_head.a_dev, ap->a_uio, ap->a_ioflag));
316 mmwrite(struct dev_write_args *ap)
318 return(mmrw(ap->a_head.a_dev, ap->a_uio, ap->a_ioflag));
325 /*******************************************************\
326 * allow user processes to MMAP some memory sections *
327 * instead of going through read/write *
328 \*******************************************************/
331 memmmap(struct dev_mmap_args *ap)
333 cdev_t dev = ap->a_head.a_dev;
335 switch (minor(dev)) {
338 * minor device 0 is physical memory
340 #if defined(__i386__)
341 ap->a_result = i386_btop(ap->a_offset);
342 #elif defined(__x86_64__)
343 ap->a_result = x86_64_btop(ap->a_offset);
348 * minor device 1 is kernel memory
350 #if defined(__i386__)
351 ap->a_result = i386_btop(vtophys(ap->a_offset));
352 #elif defined(__x86_64__)
353 ap->a_result = x86_64_btop(vtophys(ap->a_offset));
363 mmioctl(struct dev_ioctl_args *ap)
365 cdev_t dev = ap->a_head.a_dev;
370 switch (minor(dev)) {
372 error = mem_ioctl(dev, ap->a_cmd, ap->a_data,
373 ap->a_fflag, ap->a_cred);
377 error = random_ioctl(dev, ap->a_cmd, ap->a_data,
378 ap->a_fflag, ap->a_cred);
390 * Operations for changing memory attributes.
392 * This is basically just an ioctl shim for mem_range_attr_get
393 * and mem_range_attr_set.
396 mem_ioctl(cdev_t dev, u_long cmd, caddr_t data, int flags, struct ucred *cred)
399 struct mem_range_op *mo = (struct mem_range_op *)data;
400 struct mem_range_desc *md;
402 /* is this for us? */
403 if ((cmd != MEMRANGE_GET) &&
404 (cmd != MEMRANGE_SET))
407 /* any chance we can handle this? */
408 if (mem_range_softc.mr_op == NULL)
411 /* do we have any descriptors? */
412 if (mem_range_softc.mr_ndesc == 0)
417 nd = imin(mo->mo_arg[0], mem_range_softc.mr_ndesc);
419 md = (struct mem_range_desc *)
420 kmalloc(nd * sizeof(struct mem_range_desc),
421 M_MEMDESC, M_WAITOK);
422 error = mem_range_attr_get(md, &nd);
424 error = copyout(md, mo->mo_desc,
425 nd * sizeof(struct mem_range_desc));
426 kfree(md, M_MEMDESC);
428 nd = mem_range_softc.mr_ndesc;
434 md = (struct mem_range_desc *)kmalloc(sizeof(struct mem_range_desc),
435 M_MEMDESC, M_WAITOK);
436 error = copyin(mo->mo_desc, md, sizeof(struct mem_range_desc));
437 /* clamp description string */
438 md->mr_owner[sizeof(md->mr_owner) - 1] = 0;
440 error = mem_range_attr_set(md, &mo->mo_arg[0]);
441 kfree(md, M_MEMDESC);
448 * Implementation-neutral, kernel-callable functions for manipulating
449 * memory range attributes.
452 mem_range_attr_get(struct mem_range_desc *mrd, int *arg)
454 /* can we handle this? */
455 if (mem_range_softc.mr_op == NULL)
459 *arg = mem_range_softc.mr_ndesc;
461 bcopy(mem_range_softc.mr_desc, mrd, (*arg) * sizeof(struct mem_range_desc));
467 mem_range_attr_set(struct mem_range_desc *mrd, int *arg)
469 /* can we handle this? */
470 if (mem_range_softc.mr_op == NULL)
473 return (mem_range_softc.mr_op->set(&mem_range_softc, mrd, arg));
477 mem_range_AP_init(void)
479 if (mem_range_softc.mr_op && mem_range_softc.mr_op->initAP)
480 return (mem_range_softc.mr_op->initAP(&mem_range_softc));
484 random_ioctl(cdev_t dev, u_long cmd, caddr_t data, int flags, struct ucred *cred)
490 * Even inspecting the state is privileged, since it gives a hint
491 * about how easily the randomness might be guessed.
496 /* Really handled in upper layer */
500 intr = *(int16_t *)data;
501 if ((error = priv_check_cred(cred, PRIV_ROOT, 0)) != 0)
503 if (intr < 0 || intr >= MAX_INTS)
505 register_randintr(intr);
508 intr = *(int16_t *)data;
509 if ((error = priv_check_cred(cred, PRIV_ROOT, 0)) != 0)
511 if (intr < 0 || intr >= MAX_INTS)
513 unregister_randintr(intr);
519 intr = *(int16_t *)data;
520 if ((error = priv_check_cred(cred, PRIV_ROOT, 0)) != 0)
522 if (intr < 0 || intr >= MAX_INTS)
524 intr = next_registered_randintr(intr);
525 if (intr == MAX_INTS)
527 *(u_int16_t *)data = intr;
537 mm_filter_read(struct knote *kn, long hint)
543 mm_filter_write(struct knote *kn, long hint)
549 dummy_filter_detach(struct knote *kn) {}
551 /* Implemented in kern_nrandom.c */
552 static struct filterops random_read_filtops =
553 { FILTEROP_ISFD|FILTEROP_MPSAFE, NULL, dummy_filter_detach, random_filter_read };
555 static struct filterops mm_read_filtops =
556 { FILTEROP_ISFD|FILTEROP_MPSAFE, NULL, dummy_filter_detach, mm_filter_read };
558 static struct filterops mm_write_filtops =
559 { FILTEROP_ISFD|FILTEROP_MPSAFE, NULL, dummy_filter_detach, mm_filter_write };
562 mmkqfilter(struct dev_kqfilter_args *ap)
564 struct knote *kn = ap->a_kn;
565 cdev_t dev = ap->a_head.a_dev;
568 switch (kn->kn_filter) {
570 switch (minor(dev)) {
572 kn->kn_fop = &random_read_filtops;
575 kn->kn_fop = &mm_read_filtops;
580 kn->kn_fop = &mm_write_filtops;
583 ap->a_result = EOPNOTSUPP;
591 iszerodev(cdev_t dev)
593 return (zerodev == dev);
597 mem_drvinit(void *unused)
600 /* Initialise memory range handling */
601 if (mem_range_softc.mr_op != NULL)
602 mem_range_softc.mr_op->init(&mem_range_softc);
604 make_dev(&mem_ops, 0, UID_ROOT, GID_KMEM, 0640, "mem");
605 make_dev(&mem_ops, 1, UID_ROOT, GID_KMEM, 0640, "kmem");
606 make_dev(&mem_ops, 2, UID_ROOT, GID_WHEEL, 0666, "null");
607 make_dev(&mem_ops, 3, UID_ROOT, GID_WHEEL, 0644, "random");
608 make_dev(&mem_ops, 4, UID_ROOT, GID_WHEEL, 0644, "urandom");
609 zerodev = make_dev(&mem_ops, 12, UID_ROOT, GID_WHEEL, 0666, "zero");
610 make_dev(&mem_ops, 14, UID_ROOT, GID_WHEEL, 0600, "io");
613 SYSINIT(memdev,SI_SUB_DRIVERS,SI_ORDER_MIDDLE+CDEV_MAJOR,mem_drvinit,NULL)