xref: /freebsd/sys/amd64/amd64/mem.c (revision 6b3455a7665208c366849f0b2b3bc916fb97516e)
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  * 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.
22  *
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
33  * SUCH DAMAGE.
34  *
35  *	from: Utah $Hdr: mem.c 1.13 89/10/08$
36  *	from: @(#)mem.c	7.2 (Berkeley) 5/9/91
37  */
38 
39 #include <sys/cdefs.h>
40 __FBSDID("$FreeBSD$");
41 
42 /*
43  * Memory special file
44  */
45 
46 #include <sys/param.h>
47 #include <sys/conf.h>
48 #include <sys/fcntl.h>
49 #include <sys/ioccom.h>
50 #include <sys/kernel.h>
51 #include <sys/lock.h>
52 #include <sys/malloc.h>
53 #include <sys/memrange.h>
54 #include <sys/module.h>
55 #include <sys/mutex.h>
56 #include <sys/proc.h>
57 #include <sys/signalvar.h>
58 #include <sys/systm.h>
59 #include <sys/uio.h>
60 
61 #include <machine/db_machdep.h>
62 #include <machine/frame.h>
63 #include <machine/psl.h>
64 #include <machine/specialreg.h>
65 #include <machine/vmparam.h>
66 
67 #include <vm/vm.h>
68 #include <vm/pmap.h>
69 #include <vm/vm_extern.h>
70 
71 static struct cdev *memdev, *kmemdev, *iodev;
72 
73 static	d_open_t	mmopen;
74 static	d_close_t	mmclose;
75 static	d_read_t	mmrw;
76 static	d_ioctl_t	mmioctl;
77 static	d_mmap_t	memmmap;
78 
79 #define CDEV_MAJOR 2
80 static struct cdevsw mem_cdevsw = {
81 	.d_version =	D_VERSION,
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 | D_NEEDGIANT,
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(struct cdev *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(struct cdev *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(struct cdev *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 kmemphys:
161 			o = v & PAGE_MASK;
162 			c = min(uio->uio_resid, (u_int)(PAGE_SIZE - o));
163 			error = uiomove((void *)PHYS_TO_DMAP(v), (int)c, uio);
164 			continue;
165 
166 /* minor device 1 is kernel memory */
167 		case 1:
168 			v = uio->uio_offset;
169 
170 			if (v >= DMAP_MIN_ADDRESS && v < DMAP_MAX_ADDRESS) {
171 				v = DMAP_TO_PHYS(v);
172 				goto kmemphys;
173 			}
174 
175 			c = iov->iov_len;
176 
177 			/*
178 			 * Make sure that all of the pages are currently resident so
179 			 * that we don't create any zero-fill pages.
180 			 */
181 			addr = trunc_page(v);
182 			eaddr = round_page(v + c);
183 
184 			if (addr < (vm_offset_t)KERNBASE)
185 				return (EFAULT);
186 			for (; addr < eaddr; addr += PAGE_SIZE)
187 				if (pmap_extract(kernel_pmap, addr) == 0)
188 					return (EFAULT);
189 
190 			if (!kernacc((caddr_t)(long)v, c,
191 			    uio->uio_rw == UIO_READ ?
192 			    VM_PROT_READ : VM_PROT_WRITE))
193 				return (EFAULT);
194 
195 			error = uiomove((caddr_t)(long)v, (int)c, uio);
196 			continue;
197 
198 		default:
199 			return (ENODEV);
200 		}
201 
202 		if (error)
203 			break;
204 		iov->iov_base = (char *)iov->iov_base + c;
205 		iov->iov_len -= c;
206 		uio->uio_offset += c;
207 		uio->uio_resid -= c;
208 	}
209 	return (error);
210 }
211 
212 /*******************************************************\
213 * allow user processes to MMAP some memory sections	*
214 * instead of going through read/write			*
215 \*******************************************************/
216 static int
217 memmmap(struct cdev *dev, vm_offset_t offset, vm_paddr_t *paddr, int prot)
218 {
219 	switch (minor(dev))
220 	{
221 
222 	/* minor device 0 is physical memory */
223 	case 0:
224 		*paddr = offset;
225 		break;
226 
227 	/* minor device 1 is kernel memory */
228 	case 1:
229         	*paddr = vtophys(offset);
230 		break;
231 
232 	default:
233 		return (-1);
234 	}
235 	return (0);
236 }
237 
238 /*
239  * Operations for changing memory attributes.
240  *
241  * This is basically just an ioctl shim for mem_range_attr_get
242  * and mem_range_attr_set.
243  */
244 static int
245 mmioctl(struct cdev *dev, u_long cmd, caddr_t data, int flags, struct thread *td)
246 {
247 	int nd, error = 0;
248 	struct mem_range_op *mo = (struct mem_range_op *)data;
249 	struct mem_range_desc *md;
250 
251 	/* is this for us? */
252 	if ((cmd != MEMRANGE_GET) &&
253 	    (cmd != MEMRANGE_SET))
254 		return (ENOTTY);
255 
256 	/* any chance we can handle this? */
257 	if (mem_range_softc.mr_op == NULL)
258 		return (EOPNOTSUPP);
259 
260 	/* do we have any descriptors? */
261 	if (mem_range_softc.mr_ndesc == 0)
262 		return (ENXIO);
263 
264 	switch (cmd) {
265 	case MEMRANGE_GET:
266 		nd = imin(mo->mo_arg[0], mem_range_softc.mr_ndesc);
267 		if (nd > 0) {
268 			md = (struct mem_range_desc *)
269 				malloc(nd * sizeof(struct mem_range_desc),
270 				       M_MEMDESC, M_WAITOK);
271 			error = mem_range_attr_get(md, &nd);
272 			if (!error)
273 				error = copyout(md, mo->mo_desc,
274 					nd * sizeof(struct mem_range_desc));
275 			free(md, M_MEMDESC);
276 		}
277 		else
278 			nd = mem_range_softc.mr_ndesc;
279 		mo->mo_arg[0] = nd;
280 		break;
281 
282 	case MEMRANGE_SET:
283 		md = (struct mem_range_desc *)malloc(sizeof(struct mem_range_desc),
284 						    M_MEMDESC, M_WAITOK);
285 		error = copyin(mo->mo_desc, md, sizeof(struct mem_range_desc));
286 		/* clamp description string */
287 		md->mr_owner[sizeof(md->mr_owner) - 1] = 0;
288 		if (error == 0)
289 			error = mem_range_attr_set(md, &mo->mo_arg[0]);
290 		free(md, M_MEMDESC);
291 		break;
292 	}
293 	return (error);
294 }
295 
296 /*
297  * Implementation-neutral, kernel-callable functions for manipulating
298  * memory range attributes.
299  */
300 int
301 mem_range_attr_get(struct mem_range_desc *mrd, int *arg)
302 {
303 	/* can we handle this? */
304 	if (mem_range_softc.mr_op == NULL)
305 		return (EOPNOTSUPP);
306 
307 	if (*arg == 0)
308 		*arg = mem_range_softc.mr_ndesc;
309 	else
310 		bcopy(mem_range_softc.mr_desc, mrd,
311 			(*arg) * sizeof(struct mem_range_desc));
312 	return (0);
313 }
314 
315 int
316 mem_range_attr_set(struct mem_range_desc *mrd, int *arg)
317 {
318 	/* can we handle this? */
319 	if (mem_range_softc.mr_op == NULL)
320 		return (EOPNOTSUPP);
321 
322 	return (mem_range_softc.mr_op->set(&mem_range_softc, mrd, arg));
323 }
324 
325 #ifdef SMP
326 void
327 mem_range_AP_init(void)
328 {
329 	if (mem_range_softc.mr_op && mem_range_softc.mr_op->initAP)
330 		(mem_range_softc.mr_op->initAP(&mem_range_softc));
331 }
332 #endif
333 
334 static int
335 mem_modevent(module_t mod, int type, void *data)
336 {
337 	switch(type) {
338 	case MOD_LOAD:
339 		if (bootverbose)
340 			printf("mem: <memory & I/O>\n");
341 		/* Initialise memory range handling */
342 		if (mem_range_softc.mr_op != NULL)
343 			mem_range_softc.mr_op->init(&mem_range_softc);
344 
345 		memdev = make_dev(&mem_cdevsw, 0, UID_ROOT, GID_KMEM,
346 			0640, "mem");
347 		kmemdev = make_dev(&mem_cdevsw, 1, UID_ROOT, GID_KMEM,
348 			0640, "kmem");
349 		iodev = make_dev(&mem_cdevsw, 14, UID_ROOT, GID_WHEEL,
350 			0600, "io");
351 		return (0);
352 
353 	case MOD_UNLOAD:
354 		destroy_dev(memdev);
355 		destroy_dev(kmemdev);
356 		destroy_dev(iodev);
357 		return (0);
358 
359 	case MOD_SHUTDOWN:
360 		return (0);
361 
362 	default:
363 		return (EOPNOTSUPP);
364 	}
365 }
366 
367 DEV_MODULE(mem, mem_modevent, NULL);
368