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 * 3. All advertising materials mentioning features or use of this software
20 * must display the following acknowledgement:
21 * This product includes software developed by the University of
22 * California, Berkeley and its contributors.
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 * from: Utah $Hdr: mem.c 1.13 89/10/08$
40 * from: @(#)mem.c 7.2 (Berkeley) 5/9/91
41 * $FreeBSD: src/sys/i386/i386/mem.c,v 1.79.2.9 2003/01/04 22:58:01 njl Exp $
42 * $DragonFly: src/sys/kern/kern_memio.c,v 1.32 2008/07/23 16:39:28 dillon Exp $
49 #include <sys/param.h>
50 #include <sys/systm.h>
53 #include <sys/fcntl.h>
54 #include <sys/filio.h>
55 #include <sys/kernel.h>
56 #include <sys/malloc.h>
57 #include <sys/memrange.h>
60 #include <sys/random.h>
61 #include <sys/signalvar.h>
62 #include <sys/signal2.h>
64 #include <sys/vnode.h>
68 #include <vm/vm_extern.h>
71 static d_open_t mmopen;
72 static d_close_t mmclose;
73 static d_read_t mmread;
74 static d_write_t mmwrite;
75 static d_ioctl_t mmioctl;
76 static d_mmap_t memmmap;
77 static d_kqfilter_t mmkqfilter;
80 static struct dev_ops mem_ops = {
81 { "mem", CDEV_MAJOR, D_MEM | D_MPSAFE_READ | D_MPSAFE_WRITE | D_KQFILTER },
87 .d_kqfilter = mmkqfilter,
93 static cdev_t zerodev = NULL;
95 MALLOC_DEFINE(M_MEMDESC, "memdesc", "memory range descriptors");
96 static int mem_ioctl (cdev_t, u_long, caddr_t, int, struct ucred *);
97 static int random_ioctl (cdev_t, u_long, caddr_t, int, struct ucred *);
99 struct mem_range_softc mem_range_softc;
103 mmopen(struct dev_open_args *ap)
105 cdev_t dev = ap->a_head.a_dev;
108 switch (minor(dev)) {
111 if (ap->a_oflags & FWRITE) {
112 if (securelevel > 0 || kernel_mem_readonly)
118 error = priv_check_cred(ap->a_cred, PRIV_ROOT, 0);
121 if (securelevel > 0 || kernel_mem_readonly) {
125 error = cpu_set_iopl();
135 mmclose(struct dev_close_args *ap)
137 cdev_t dev = ap->a_head.a_dev;
140 switch (minor(dev)) {
142 error = cpu_clr_iopl();
153 mmrw(cdev_t dev, struct uio *uio, int flags)
162 while (uio->uio_resid > 0 && error == 0) {
164 if (iov->iov_len == 0) {
167 if (uio->uio_iovcnt < 0)
171 switch (minor(dev)) {
174 * minor device 0 is physical memory, /dev/mem
178 pmap_kenter((vm_offset_t)ptvmmap, v);
179 o = (int)uio->uio_offset & PAGE_MASK;
180 c = (u_int)(PAGE_SIZE - ((uintptr_t)iov->iov_base & PAGE_MASK));
181 c = min(c, (u_int)(PAGE_SIZE - o));
182 c = min(c, (u_int)iov->iov_len);
183 error = uiomove((caddr_t)&ptvmmap[o], (int)c, uio);
184 pmap_kremove((vm_offset_t)ptvmmap);
189 * minor device 1 is kernel memory, /dev/kmem
191 vm_offset_t saddr, eaddr;
197 * Make sure that all of the pages are currently
198 * resident so that we don't create any zero-fill
201 saddr = trunc_page(uio->uio_offset);
202 eaddr = round_page(uio->uio_offset + c);
207 * Make sure the kernel addresses are mapped.
208 * platform_direct_mapped() can be used to bypass
209 * default mapping via the page table (virtual kernels
210 * contain a lot of out-of-band data).
213 if (uio->uio_rw != UIO_READ)
214 prot |= VM_PROT_WRITE;
215 error = kvm_access_check(saddr, eaddr, prot);
218 error = uiomove((caddr_t)(vm_offset_t)uio->uio_offset,
224 * minor device 2 is EOF/RATHOLE
226 if (uio->uio_rw == UIO_READ)
232 * minor device 3 (/dev/random) is source of filth
233 * on read, seeder on write
236 buf = kmalloc(PAGE_SIZE, M_TEMP, M_WAITOK);
237 c = min(iov->iov_len, PAGE_SIZE);
238 if (uio->uio_rw == UIO_WRITE) {
239 error = uiomove(buf, (int)c, uio);
241 error = add_buffer_randomness(buf, c);
243 poolsize = read_random(buf, c);
247 if ((flags & IO_NDELAY) != 0)
248 return (EWOULDBLOCK);
251 c = min(c, poolsize);
252 error = uiomove(buf, (int)c, uio);
257 * minor device 4 (/dev/urandom) is source of muck
258 * on read, writes are disallowed.
260 c = min(iov->iov_len, PAGE_SIZE);
261 if (uio->uio_rw == UIO_WRITE) {
265 if (CURSIG(curthread->td_lwp) != 0) {
267 * Use tsleep() to get the error code right.
268 * It should return immediately.
270 error = tsleep(&rand_bolt, PCATCH, "urand", 1);
271 if (error != 0 && error != EWOULDBLOCK)
275 buf = kmalloc(PAGE_SIZE, M_TEMP, M_WAITOK);
276 poolsize = read_random_unlimited(buf, c);
277 c = min(c, poolsize);
278 error = uiomove(buf, (int)c, uio);
282 * minor device 12 (/dev/zero) is source of nulls
283 * on read, write are disallowed.
285 if (uio->uio_rw == UIO_WRITE) {
290 zbuf = (caddr_t)kmalloc(PAGE_SIZE, M_TEMP,
293 c = min(iov->iov_len, PAGE_SIZE);
294 error = uiomove(zbuf, (int)c, uio);
301 iov->iov_base = (char *)iov->iov_base + c;
303 uio->uio_offset += c;
312 mmread(struct dev_read_args *ap)
314 return(mmrw(ap->a_head.a_dev, ap->a_uio, ap->a_ioflag));
318 mmwrite(struct dev_write_args *ap)
320 return(mmrw(ap->a_head.a_dev, ap->a_uio, ap->a_ioflag));
327 /*******************************************************\
328 * allow user processes to MMAP some memory sections *
329 * instead of going through read/write *
330 \*******************************************************/
333 memmmap(struct dev_mmap_args *ap)
335 cdev_t dev = ap->a_head.a_dev;
337 switch (minor(dev)) {
340 * minor device 0 is physical memory
342 #if defined(__i386__)
343 ap->a_result = i386_btop(ap->a_offset);
344 #elif defined(__x86_64__)
345 ap->a_result = x86_64_btop(ap->a_offset);
350 * minor device 1 is kernel memory
352 #if defined(__i386__)
353 ap->a_result = i386_btop(vtophys(ap->a_offset));
354 #elif defined(__x86_64__)
355 ap->a_result = x86_64_btop(vtophys(ap->a_offset));
365 mmioctl(struct dev_ioctl_args *ap)
367 cdev_t dev = ap->a_head.a_dev;
369 switch (minor(dev)) {
371 return mem_ioctl(dev, ap->a_cmd, ap->a_data,
372 ap->a_fflag, ap->a_cred);
375 return random_ioctl(dev, ap->a_cmd, ap->a_data,
376 ap->a_fflag, ap->a_cred);
382 * Operations for changing memory attributes.
384 * This is basically just an ioctl shim for mem_range_attr_get
385 * and mem_range_attr_set.
388 mem_ioctl(cdev_t dev, u_long cmd, caddr_t data, int flags, struct ucred *cred)
391 struct mem_range_op *mo = (struct mem_range_op *)data;
392 struct mem_range_desc *md;
394 /* is this for us? */
395 if ((cmd != MEMRANGE_GET) &&
396 (cmd != MEMRANGE_SET))
399 /* any chance we can handle this? */
400 if (mem_range_softc.mr_op == NULL)
403 /* do we have any descriptors? */
404 if (mem_range_softc.mr_ndesc == 0)
409 nd = imin(mo->mo_arg[0], mem_range_softc.mr_ndesc);
411 md = (struct mem_range_desc *)
412 kmalloc(nd * sizeof(struct mem_range_desc),
413 M_MEMDESC, M_WAITOK);
414 error = mem_range_attr_get(md, &nd);
416 error = copyout(md, mo->mo_desc,
417 nd * sizeof(struct mem_range_desc));
418 kfree(md, M_MEMDESC);
420 nd = mem_range_softc.mr_ndesc;
426 md = (struct mem_range_desc *)kmalloc(sizeof(struct mem_range_desc),
427 M_MEMDESC, M_WAITOK);
428 error = copyin(mo->mo_desc, md, sizeof(struct mem_range_desc));
429 /* clamp description string */
430 md->mr_owner[sizeof(md->mr_owner) - 1] = 0;
432 error = mem_range_attr_set(md, &mo->mo_arg[0]);
433 kfree(md, M_MEMDESC);
440 * Implementation-neutral, kernel-callable functions for manipulating
441 * memory range attributes.
444 mem_range_attr_get(struct mem_range_desc *mrd, int *arg)
446 /* can we handle this? */
447 if (mem_range_softc.mr_op == NULL)
451 *arg = mem_range_softc.mr_ndesc;
453 bcopy(mem_range_softc.mr_desc, mrd, (*arg) * sizeof(struct mem_range_desc));
459 mem_range_attr_set(struct mem_range_desc *mrd, int *arg)
461 /* can we handle this? */
462 if (mem_range_softc.mr_op == NULL)
465 return (mem_range_softc.mr_op->set(&mem_range_softc, mrd, arg));
470 mem_range_AP_init(void)
472 if (mem_range_softc.mr_op && mem_range_softc.mr_op->initAP)
473 return (mem_range_softc.mr_op->initAP(&mem_range_softc));
478 random_ioctl(cdev_t dev, u_long cmd, caddr_t data, int flags, struct ucred *cred)
484 * Even inspecting the state is privileged, since it gives a hint
485 * about how easily the randomness might be guessed.
490 /* Really handled in upper layer */
494 intr = *(int16_t *)data;
495 if ((error = priv_check_cred(cred, PRIV_ROOT, 0)) != 0)
497 if (intr < 0 || intr >= MAX_INTS)
499 register_randintr(intr);
502 intr = *(int16_t *)data;
503 if ((error = priv_check_cred(cred, PRIV_ROOT, 0)) != 0)
505 if (intr < 0 || intr >= MAX_INTS)
507 unregister_randintr(intr);
513 intr = *(int16_t *)data;
514 if ((error = priv_check_cred(cred, PRIV_ROOT, 0)) != 0)
516 if (intr < 0 || intr >= MAX_INTS)
518 intr = next_registered_randintr(intr);
519 if (intr == MAX_INTS)
521 *(u_int16_t *)data = intr;
531 mm_filter_read(struct knote *kn, long hint)
537 dummy_filter_detach(struct knote *kn) {}
539 static struct filterops random_read_filtops =
540 { FILTEROP_ISFD, NULL, dummy_filter_detach, random_filter_read };
542 static struct filterops mm_read_filtops =
543 { FILTEROP_ISFD, NULL, dummy_filter_detach, mm_filter_read };
546 mmkqfilter(struct dev_kqfilter_args *ap)
548 struct knote *kn = ap->a_kn;
549 cdev_t dev = ap->a_head.a_dev;
552 switch (kn->kn_filter) {
554 switch (minor(dev)) {
556 kn->kn_fop = &random_read_filtops;
559 kn->kn_fop = &mm_read_filtops;
564 ap->a_result = EOPNOTSUPP;
572 iszerodev(cdev_t dev)
574 return (zerodev == dev);
578 mem_drvinit(void *unused)
581 /* Initialise memory range handling */
582 if (mem_range_softc.mr_op != NULL)
583 mem_range_softc.mr_op->init(&mem_range_softc);
585 make_dev(&mem_ops, 0, UID_ROOT, GID_KMEM, 0640, "mem");
586 make_dev(&mem_ops, 1, UID_ROOT, GID_KMEM, 0640, "kmem");
587 make_dev(&mem_ops, 2, UID_ROOT, GID_WHEEL, 0666, "null");
588 make_dev(&mem_ops, 3, UID_ROOT, GID_WHEEL, 0644, "random");
589 make_dev(&mem_ops, 4, UID_ROOT, GID_WHEEL, 0644, "urandom");
590 zerodev = make_dev(&mem_ops, 12, UID_ROOT, GID_WHEEL, 0666, "zero");
591 make_dev(&mem_ops, 14, UID_ROOT, GID_WHEEL, 0600, "io");
594 SYSINIT(memdev,SI_SUB_DRIVERS,SI_ORDER_MIDDLE+CDEV_MAJOR,mem_drvinit,NULL)