xref: /freebsd/sys/amd64/amd64/mem.c (revision d086ded32300bc0f33fb1574d0bcfccfbc60881d)
1 /*-
2  * Copyright (c) 1988 University of Utah.
3  * Copyright (c) 1982, 1986, 1990 The Regents of the University of California.
4  * All rights reserved.
5  *
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.
10  *
11  * Redistribution and use in source and binary forms, with or without
12  * modification, are permitted provided that the following conditions
13  * are met:
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.
26  *
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
37  * SUCH DAMAGE.
38  *
39  *	from: Utah $Hdr: mem.c 1.13 89/10/08$
40  *	from: @(#)mem.c	7.2 (Berkeley) 5/9/91
41  * $FreeBSD$
42  */
43 
44 /*
45  * Memory special file
46  */
47 
48 #include <sys/param.h>
49 #include <sys/conf.h>
50 #include <sys/fcntl.h>
51 #include <sys/ioccom.h>
52 #include <sys/kernel.h>
53 #include <sys/lock.h>
54 #include <sys/malloc.h>
55 #include <sys/memrange.h>
56 #include <sys/mutex.h>
57 #include <sys/proc.h>
58 #include <sys/signalvar.h>
59 #include <sys/systm.h>
60 #include <sys/uio.h>
61 
62 #include <machine/db_machdep.h>
63 #include <machine/frame.h>
64 #include <machine/psl.h>
65 #include <machine/specialreg.h>
66 #include <machine/vmparam.h>
67 
68 #include <vm/vm.h>
69 #include <vm/pmap.h>
70 #include <vm/vm_extern.h>
71 
72 static dev_t memdev, kmemdev, iodev;
73 
74 static	d_open_t	mmopen;
75 static	d_close_t	mmclose;
76 static	d_read_t	mmrw;
77 static	d_ioctl_t	mmioctl;
78 static	d_mmap_t	memmmap;
79 
80 #define CDEV_MAJOR 2
81 static struct cdevsw mem_cdevsw = {
82 	.d_open =	mmopen,
83 	.d_close =	mmclose,
84 	.d_read =	mmrw,
85 	.d_write =	mmrw,
86 	.d_ioctl =	mmioctl,
87 	.d_mmap =	memmmap,
88 	.d_name =	"mem",
89 	.d_maj =	CDEV_MAJOR,
90 	.d_flags =	D_MEM,
91 };
92 
93 MALLOC_DEFINE(M_MEMDESC, "memdesc", "memory range descriptors");
94 
95 struct mem_range_softc mem_range_softc;
96 
97 static int
98 mmclose(dev_t dev, int flags, int fmt, struct thread *td)
99 {
100 	switch (minor(dev)) {
101 	case 14:
102 		td->td_frame->tf_rflags &= ~PSL_IOPL;
103 	}
104 	return (0);
105 }
106 
107 static int
108 mmopen(dev_t dev, int flags, int fmt, struct thread *td)
109 {
110 	int error;
111 
112 	switch (minor(dev)) {
113 	case 0:
114 	case 1:
115 		if (flags & FWRITE) {
116 			error = securelevel_gt(td->td_ucred, 0);
117 			if (error != 0)
118 				return (error);
119 		}
120 		break;
121 	case 14:
122 		error = suser(td);
123 		if (error != 0)
124 			return (error);
125 		error = securelevel_gt(td->td_ucred, 0);
126 		if (error != 0)
127 			return (error);
128 		td->td_frame->tf_rflags |= PSL_IOPL;
129 		break;
130 	}
131 	return (0);
132 }
133 
134 /*ARGSUSED*/
135 static int
136 mmrw(dev_t dev, struct uio *uio, int flags)
137 {
138 	int o;
139 	u_long c = 0, v;
140 	struct iovec *iov;
141 	int error = 0;
142 	vm_offset_t addr, eaddr;
143 
144 	GIANT_REQUIRED;
145 
146 	while (uio->uio_resid > 0 && error == 0) {
147 		iov = uio->uio_iov;
148 		if (iov->iov_len == 0) {
149 			uio->uio_iov++;
150 			uio->uio_iovcnt--;
151 			if (uio->uio_iovcnt < 0)
152 				panic("mmrw");
153 			continue;
154 		}
155 		switch (minor(dev)) {
156 
157 /* minor device 0 is physical memory */
158 		case 0:
159 			v = uio->uio_offset;
160 			v &= ~PAGE_MASK;
161 			pmap_kenter((vm_offset_t)ptvmmap, v);
162 			o = (int)uio->uio_offset & PAGE_MASK;
163 			c = (u_long)(PAGE_SIZE - ((long)iov->iov_base & PAGE_MASK));
164 			c = min(c, (u_int)(PAGE_SIZE - o));
165 			c = min(c, (u_int)iov->iov_len);
166 			error = uiomove((caddr_t)&ptvmmap[o], (int)c, uio);
167 			pmap_qremove((vm_offset_t)ptvmmap, 1);
168 			continue;
169 
170 /* minor device 1 is kernel memory */
171 		case 1:
172 			c = iov->iov_len;
173 
174 			/*
175 			 * Make sure that all of the pages are currently resident so
176 			 * that we don't create any zero-fill pages.
177 			 */
178 			addr = trunc_page(uio->uio_offset);
179 			eaddr = round_page(uio->uio_offset + c);
180 
181 			if (addr < (vm_offset_t)KERNBASE)
182 				return (EFAULT);
183 			for (; addr < eaddr; addr += PAGE_SIZE)
184 				if (pmap_extract(kernel_pmap, addr) == 0)
185 					return (EFAULT);
186 
187 			if (!kernacc((caddr_t)(long)uio->uio_offset, c,
188 			    uio->uio_rw == UIO_READ ?
189 			    VM_PROT_READ : VM_PROT_WRITE))
190 				return (EFAULT);
191 			error = uiomove((caddr_t)(long)uio->uio_offset, (int)c, uio);
192 			continue;
193 
194 		default:
195 			return (ENODEV);
196 		}
197 
198 		if (error)
199 			break;
200 		iov->iov_base = (char *)iov->iov_base + c;
201 		iov->iov_len -= c;
202 		uio->uio_offset += c;
203 		uio->uio_resid -= c;
204 	}
205 	return (error);
206 }
207 
208 /*******************************************************\
209 * allow user processes to MMAP some memory sections	*
210 * instead of going through read/write			*
211 \*******************************************************/
212 static int
213 memmmap(dev_t dev, vm_offset_t offset, vm_paddr_t *paddr, int prot)
214 {
215 	switch (minor(dev))
216 	{
217 
218 	/* minor device 0 is physical memory */
219 	case 0:
220 		*paddr = offset;
221 		break;
222 
223 	/* minor device 1 is kernel memory */
224 	case 1:
225         	*paddr = vtophys(offset);
226 		break;
227 
228 	default:
229 		return (-1);
230 	}
231 	return (0);
232 }
233 
234 /*
235  * Operations for changing memory attributes.
236  *
237  * This is basically just an ioctl shim for mem_range_attr_get
238  * and mem_range_attr_set.
239  */
240 static int
241 mmioctl(dev_t dev, u_long cmd, caddr_t data, int flags, struct thread *td)
242 {
243 	int nd, error = 0;
244 	struct mem_range_op *mo = (struct mem_range_op *)data;
245 	struct mem_range_desc *md;
246 
247 	/* is this for us? */
248 	if ((cmd != MEMRANGE_GET) &&
249 	    (cmd != MEMRANGE_SET))
250 		return (ENOTTY);
251 
252 	/* any chance we can handle this? */
253 	if (mem_range_softc.mr_op == NULL)
254 		return (EOPNOTSUPP);
255 
256 	/* do we have any descriptors? */
257 	if (mem_range_softc.mr_ndesc == 0)
258 		return (ENXIO);
259 
260 	switch (cmd) {
261 	case MEMRANGE_GET:
262 		nd = imin(mo->mo_arg[0], mem_range_softc.mr_ndesc);
263 		if (nd > 0) {
264 			md = (struct mem_range_desc *)
265 				malloc(nd * sizeof(struct mem_range_desc),
266 				       M_MEMDESC, M_WAITOK);
267 			error = mem_range_attr_get(md, &nd);
268 			if (!error)
269 				error = copyout(md, mo->mo_desc,
270 					nd * sizeof(struct mem_range_desc));
271 			free(md, M_MEMDESC);
272 		}
273 		else
274 			nd = mem_range_softc.mr_ndesc;
275 		mo->mo_arg[0] = nd;
276 		break;
277 
278 	case MEMRANGE_SET:
279 		md = (struct mem_range_desc *)malloc(sizeof(struct mem_range_desc),
280 						    M_MEMDESC, M_WAITOK);
281 		error = copyin(mo->mo_desc, md, sizeof(struct mem_range_desc));
282 		/* clamp description string */
283 		md->mr_owner[sizeof(md->mr_owner) - 1] = 0;
284 		if (error == 0)
285 			error = mem_range_attr_set(md, &mo->mo_arg[0]);
286 		free(md, M_MEMDESC);
287 		break;
288 	}
289 	return (error);
290 }
291 
292 /*
293  * Implementation-neutral, kernel-callable functions for manipulating
294  * memory range attributes.
295  */
296 int
297 mem_range_attr_get(struct mem_range_desc *mrd, int *arg)
298 {
299 	/* can we handle this? */
300 	if (mem_range_softc.mr_op == NULL)
301 		return (EOPNOTSUPP);
302 
303 	if (*arg == 0)
304 		*arg = mem_range_softc.mr_ndesc;
305 	else
306 		bcopy(mem_range_softc.mr_desc, mrd,
307 			(*arg) * sizeof(struct mem_range_desc));
308 	return (0);
309 }
310 
311 int
312 mem_range_attr_set(struct mem_range_desc *mrd, int *arg)
313 {
314 	/* can we handle this? */
315 	if (mem_range_softc.mr_op == NULL)
316 		return (EOPNOTSUPP);
317 
318 	return (mem_range_softc.mr_op->set(&mem_range_softc, mrd, arg));
319 }
320 
321 static int
322 mem_modevent(module_t mod, int type, void *data)
323 {
324 	switch(type) {
325 	case MOD_LOAD:
326 		if (bootverbose)
327 			printf("mem: <memory & I/O>\n");
328 		/* Initialise memory range handling */
329 		if (mem_range_softc.mr_op != NULL)
330 			mem_range_softc.mr_op->init(&mem_range_softc);
331 
332 		memdev = make_dev(&mem_cdevsw, 0, UID_ROOT, GID_KMEM,
333 			0640, "mem");
334 		kmemdev = make_dev(&mem_cdevsw, 1, UID_ROOT, GID_KMEM,
335 			0640, "kmem");
336 		iodev = make_dev(&mem_cdevsw, 14, UID_ROOT, GID_WHEEL,
337 			0600, "io");
338 		return (0);
339 
340 	case MOD_UNLOAD:
341 		destroy_dev(memdev);
342 		destroy_dev(kmemdev);
343 		destroy_dev(iodev);
344 		return (0);
345 
346 	case MOD_SHUTDOWN:
347 		return (0);
348 
349 	default:
350 		return (EOPNOTSUPP);
351 	}
352 }
353 
354 DEV_MODULE(mem, mem_modevent, NULL);
355