1 /*
2 * memory management system call shims and definitions
3 *
4 * Copyright (c) 2013-15 Stacey D. Son
5 *
6 * This program is free software; you can redistribute it and/or modify
7 * it under the terms of the GNU General Public License as published by
8 * the Free Software Foundation; either version 2 of the License, or
9 * (at your option) any later version.
10 *
11 * This program is distributed in the hope that it will be useful,
12 * but WITHOUT ANY WARRANTY; without even the implied warranty of
13 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
14 * GNU General Public License for more details.
15 *
16 * You should have received a copy of the GNU General Public License
17 * along with this program; if not, see <http://www.gnu.org/licenses/>.
18 */
19
20 /*
21 * Copyright (c) 1982, 1986, 1993
22 * The Regents of the University of California. All rights reserved.
23 *
24 * Redistribution and use in source and binary forms, with or without
25 * modification, are permitted provided that the following conditions
26 * are met:
27 * 1. Redistributions of source code must retain the above copyright
28 * notice, this list of conditions and the following disclaimer.
29 * 2. Redistributions in binary form must reproduce the above copyright
30 * notice, this list of conditions and the following disclaimer in the
31 * documentation and/or other materials provided with the distribution.
32 * 4. Neither the name of the University nor the names of its contributors
33 * may be used to endorse or promote products derived from this software
34 * without specific prior written permission.
35 *
36 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
37 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
38 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
39 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
40 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
41 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
42 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
43 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
44 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
45 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
46 * SUCH DAMAGE.
47 */
48
49 #ifndef BSD_USER_BSD_MEM_H
50 #define BSD_USER_BSD_MEM_H
51
52 #include <sys/types.h>
53 #include <sys/ipc.h>
54 #include <sys/mman.h>
55 #include <sys/shm.h>
56 #include <fcntl.h>
57
58 #include "qemu-bsd.h"
59 #include "exec/page-protection.h"
60 #include "user/page-protection.h"
61
62 extern struct bsd_shm_regions bsd_shm_regions[];
63 extern abi_ulong target_brk;
64 extern abi_ulong initial_target_brk;
65
66 /* mmap(2) */
do_bsd_mmap(void * cpu_env,abi_long arg1,abi_long arg2,abi_long arg3,abi_long arg4,abi_long arg5,abi_long arg6,abi_long arg7,abi_long arg8)67 static inline abi_long do_bsd_mmap(void *cpu_env, abi_long arg1, abi_long arg2,
68 abi_long arg3, abi_long arg4, abi_long arg5, abi_long arg6, abi_long arg7,
69 abi_long arg8)
70 {
71 if (regpairs_aligned(cpu_env) != 0) {
72 arg6 = arg7;
73 arg7 = arg8;
74 }
75 return get_errno(target_mmap(arg1, arg2, arg3,
76 target_to_host_bitmask(arg4, mmap_flags_tbl),
77 arg5, target_arg64(arg6, arg7)));
78 }
79
80 /* munmap(2) */
do_bsd_munmap(abi_long arg1,abi_long arg2)81 static inline abi_long do_bsd_munmap(abi_long arg1, abi_long arg2)
82 {
83 return get_errno(target_munmap(arg1, arg2));
84 }
85
86 /* mprotect(2) */
do_bsd_mprotect(abi_long arg1,abi_long arg2,abi_long arg3)87 static inline abi_long do_bsd_mprotect(abi_long arg1, abi_long arg2,
88 abi_long arg3)
89 {
90 return get_errno(target_mprotect(arg1, arg2, arg3));
91 }
92
93 /* msync(2) */
do_bsd_msync(abi_long addr,abi_long len,abi_long flags)94 static inline abi_long do_bsd_msync(abi_long addr, abi_long len, abi_long flags)
95 {
96 if (!guest_range_valid_untagged(addr, len)) {
97 /* It seems odd, but POSIX wants this to be ENOMEM */
98 return -TARGET_ENOMEM;
99 }
100
101 return get_errno(msync(g2h_untagged(addr), len, flags));
102 }
103
104 /* mlock(2) */
do_bsd_mlock(abi_long arg1,abi_long arg2)105 static inline abi_long do_bsd_mlock(abi_long arg1, abi_long arg2)
106 {
107 if (!guest_range_valid_untagged(arg1, arg2)) {
108 return -TARGET_EINVAL;
109 }
110 return get_errno(mlock(g2h_untagged(arg1), arg2));
111 }
112
113 /* munlock(2) */
do_bsd_munlock(abi_long arg1,abi_long arg2)114 static inline abi_long do_bsd_munlock(abi_long arg1, abi_long arg2)
115 {
116 if (!guest_range_valid_untagged(arg1, arg2)) {
117 return -TARGET_EINVAL;
118 }
119 return get_errno(munlock(g2h_untagged(arg1), arg2));
120 }
121
122 /* mlockall(2) */
do_bsd_mlockall(abi_long arg1)123 static inline abi_long do_bsd_mlockall(abi_long arg1)
124 {
125 return get_errno(mlockall(arg1));
126 }
127
128 /* munlockall(2) */
do_bsd_munlockall(void)129 static inline abi_long do_bsd_munlockall(void)
130 {
131 return get_errno(munlockall());
132 }
133
134 /* madvise(2) */
do_bsd_madvise(abi_long arg1,abi_long arg2,abi_long arg3)135 static inline abi_long do_bsd_madvise(abi_long arg1, abi_long arg2,
136 abi_long arg3)
137 {
138 abi_ulong len;
139 int ret = 0;
140 abi_long start = arg1;
141 abi_long len_in = arg2;
142 abi_long advice = arg3;
143
144 if (start & ~TARGET_PAGE_MASK) {
145 return -TARGET_EINVAL;
146 }
147 if (len_in == 0) {
148 return 0;
149 }
150 len = TARGET_PAGE_ALIGN(len_in);
151 if (len == 0 || !guest_range_valid_untagged(start, len)) {
152 return -TARGET_EINVAL;
153 }
154
155 /*
156 * Most advice values are hints, so ignoring and returning success is ok.
157 *
158 * However, some advice values such as MADV_DONTNEED, are not hints and
159 * need to be emulated.
160 *
161 * A straight passthrough for those may not be safe because qemu sometimes
162 * turns private file-backed mappings into anonymous mappings.
163 * If all guest pages have PAGE_PASSTHROUGH set, mappings have the
164 * same semantics for the host as for the guest.
165 *
166 * MADV_DONTNEED is passed through, if possible.
167 * If passthrough isn't possible, we nevertheless (wrongly!) return
168 * success, which is broken but some userspace programs fail to work
169 * otherwise. Completely implementing such emulation is quite complicated
170 * though.
171 */
172 mmap_lock();
173 switch (advice) {
174 case MADV_DONTNEED:
175 if (page_check_range(start, len, PAGE_PASSTHROUGH)) {
176 ret = get_errno(madvise(g2h_untagged(start), len, advice));
177 if (ret == 0) {
178 page_reset_target_data(start, start + len - 1);
179 }
180 }
181 }
182 mmap_unlock();
183
184 return ret;
185 }
186
187 /* minherit(2) */
do_bsd_minherit(abi_long addr,abi_long len,abi_long inherit)188 static inline abi_long do_bsd_minherit(abi_long addr, abi_long len,
189 abi_long inherit)
190 {
191 return get_errno(minherit(g2h_untagged(addr), len, inherit));
192 }
193
194 /* mincore(2) */
do_bsd_mincore(abi_ulong target_addr,abi_ulong len,abi_ulong target_vec)195 static inline abi_long do_bsd_mincore(abi_ulong target_addr, abi_ulong len,
196 abi_ulong target_vec)
197 {
198 abi_long ret;
199 void *p;
200 abi_ulong vec_len = DIV_ROUND_UP(len, TARGET_PAGE_SIZE);
201
202 if (!guest_range_valid_untagged(target_addr, len)
203 || !page_check_range(target_addr, len, PAGE_VALID)) {
204 return -TARGET_EFAULT;
205 }
206
207 p = lock_user(VERIFY_WRITE, target_vec, vec_len, 0);
208 if (p == NULL) {
209 return -TARGET_EFAULT;
210 }
211 ret = get_errno(mincore(g2h_untagged(target_addr), len, p));
212 unlock_user(p, target_vec, vec_len);
213
214 return ret;
215 }
216
217 /* do_brk() must return target values and target errnos. */
do_obreak(abi_ulong brk_val)218 static inline abi_long do_obreak(abi_ulong brk_val)
219 {
220 abi_long mapped_addr;
221 abi_ulong new_brk;
222 abi_ulong old_brk;
223
224 /* brk pointers are always untagged */
225
226 /* do not allow to shrink below initial brk value */
227 if (brk_val < initial_target_brk) {
228 return target_brk;
229 }
230
231 new_brk = TARGET_PAGE_ALIGN(brk_val);
232 old_brk = TARGET_PAGE_ALIGN(target_brk);
233
234 /* new and old target_brk might be on the same page */
235 if (new_brk == old_brk) {
236 target_brk = brk_val;
237 return target_brk;
238 }
239
240 /* Release heap if necessary */
241 if (new_brk < old_brk) {
242 target_munmap(new_brk, old_brk - new_brk);
243
244 target_brk = brk_val;
245 return target_brk;
246 }
247
248 mapped_addr = target_mmap(old_brk, new_brk - old_brk,
249 PROT_READ | PROT_WRITE,
250 MAP_FIXED | MAP_EXCL | MAP_ANON | MAP_PRIVATE,
251 -1, 0);
252
253 if (mapped_addr == old_brk) {
254 target_brk = brk_val;
255 return target_brk;
256 }
257
258 /* For everything else, return the previous break. */
259 return target_brk;
260 }
261
262 /* shm_open(2) */
do_bsd_shm_open(abi_ulong arg1,abi_long arg2,abi_long arg3)263 static inline abi_long do_bsd_shm_open(abi_ulong arg1, abi_long arg2,
264 abi_long arg3)
265 {
266 int ret;
267 void *p;
268
269 if (arg1 == (uintptr_t)SHM_ANON) {
270 p = SHM_ANON;
271 } else {
272 p = lock_user_string(arg1);
273 if (p == NULL) {
274 return -TARGET_EFAULT;
275 }
276 }
277 ret = get_errno(shm_open(p, target_to_host_bitmask(arg2, fcntl_flags_tbl),
278 arg3));
279
280 if (p != SHM_ANON) {
281 unlock_user(p, arg1, 0);
282 }
283
284 return ret;
285 }
286
287 /* shm_unlink(2) */
do_bsd_shm_unlink(abi_ulong arg1)288 static inline abi_long do_bsd_shm_unlink(abi_ulong arg1)
289 {
290 int ret;
291 void *p;
292
293 p = lock_user_string(arg1);
294 if (p == NULL) {
295 return -TARGET_EFAULT;
296 }
297 ret = get_errno(shm_unlink(p)); /* XXX path(p)? */
298 unlock_user(p, arg1, 0);
299
300 return ret;
301 }
302
303 /* shmget(2) */
do_bsd_shmget(abi_long arg1,abi_ulong arg2,abi_long arg3)304 static inline abi_long do_bsd_shmget(abi_long arg1, abi_ulong arg2,
305 abi_long arg3)
306 {
307 return get_errno(shmget(arg1, arg2, arg3));
308 }
309
310 /* shmctl(2) */
do_bsd_shmctl(abi_long shmid,abi_long cmd,abi_ulong buff)311 static inline abi_long do_bsd_shmctl(abi_long shmid, abi_long cmd,
312 abi_ulong buff)
313 {
314 struct shmid_ds dsarg;
315 abi_long ret = -TARGET_EINVAL;
316
317 cmd &= 0xff;
318
319 switch (cmd) {
320 case IPC_STAT:
321 if (target_to_host_shmid_ds(&dsarg, buff)) {
322 return -TARGET_EFAULT;
323 }
324 ret = get_errno(shmctl(shmid, cmd, &dsarg));
325 if (host_to_target_shmid_ds(buff, &dsarg)) {
326 return -TARGET_EFAULT;
327 }
328 break;
329
330 case IPC_SET:
331 if (target_to_host_shmid_ds(&dsarg, buff)) {
332 return -TARGET_EFAULT;
333 }
334 ret = get_errno(shmctl(shmid, cmd, &dsarg));
335 break;
336
337 case IPC_RMID:
338 ret = get_errno(shmctl(shmid, cmd, NULL));
339 break;
340
341 default:
342 ret = -TARGET_EINVAL;
343 break;
344 }
345
346 return ret;
347 }
348
349 /* shmat(2) */
do_bsd_shmat(int shmid,abi_ulong shmaddr,int shmflg)350 static inline abi_long do_bsd_shmat(int shmid, abi_ulong shmaddr, int shmflg)
351 {
352 abi_ulong raddr;
353 abi_long ret;
354 struct shmid_ds shm_info;
355
356 /* Find out the length of the shared memory segment. */
357 ret = get_errno(shmctl(shmid, IPC_STAT, &shm_info));
358 if (is_error(ret)) {
359 /* Can't get the length */
360 return ret;
361 }
362
363 if (!guest_range_valid_untagged(shmaddr, shm_info.shm_segsz)) {
364 return -TARGET_EINVAL;
365 }
366
367 WITH_MMAP_LOCK_GUARD() {
368 void *host_raddr;
369
370 if (shmaddr) {
371 host_raddr = shmat(shmid, (void *)g2h_untagged(shmaddr), shmflg);
372 } else {
373 abi_ulong alignment;
374 abi_ulong mmap_start;
375
376 alignment = 0; /* alignment above page size not required */
377 mmap_start = mmap_find_vma(0, shm_info.shm_segsz, alignment);
378
379 if (mmap_start == -1) {
380 return -TARGET_ENOMEM;
381 }
382 host_raddr = shmat(shmid, g2h_untagged(mmap_start),
383 shmflg | SHM_REMAP);
384 }
385
386 if (host_raddr == (void *)-1) {
387 return get_errno(-1);
388 }
389 raddr = h2g(host_raddr);
390
391 page_set_flags(raddr, raddr + shm_info.shm_segsz - 1,
392 PAGE_VALID | PAGE_RESET | PAGE_READ |
393 (shmflg & SHM_RDONLY ? 0 : PAGE_WRITE));
394
395 for (int i = 0; i < N_BSD_SHM_REGIONS; i++) {
396 if (bsd_shm_regions[i].start == 0) {
397 bsd_shm_regions[i].start = raddr;
398 bsd_shm_regions[i].size = shm_info.shm_segsz;
399 break;
400 }
401 }
402 }
403
404 return raddr;
405 }
406
407 /* shmdt(2) */
do_bsd_shmdt(abi_ulong shmaddr)408 static inline abi_long do_bsd_shmdt(abi_ulong shmaddr)
409 {
410 abi_long ret;
411
412 WITH_MMAP_LOCK_GUARD() {
413 int i;
414
415 for (i = 0; i < N_BSD_SHM_REGIONS; ++i) {
416 if (bsd_shm_regions[i].start == shmaddr) {
417 break;
418 }
419 }
420
421 if (i == N_BSD_SHM_REGIONS) {
422 return -TARGET_EINVAL;
423 }
424
425 ret = get_errno(shmdt(g2h_untagged(shmaddr)));
426 if (ret == 0) {
427 abi_ulong size = bsd_shm_regions[i].size;
428
429 bsd_shm_regions[i].start = 0;
430 page_set_flags(shmaddr, shmaddr + size - 1, 0);
431 mmap_reserve(shmaddr, size);
432 }
433 }
434
435 return ret;
436 }
437
do_bsd_vadvise(void)438 static inline abi_long do_bsd_vadvise(void)
439 {
440 /* See sys_ovadvise() in vm_unix.c */
441 return -TARGET_EINVAL;
442 }
443
do_bsd_sbrk(void)444 static inline abi_long do_bsd_sbrk(void)
445 {
446 /* see sys_sbrk() in vm_mmap.c */
447 return -TARGET_EOPNOTSUPP;
448 }
449
do_bsd_sstk(void)450 static inline abi_long do_bsd_sstk(void)
451 {
452 /* see sys_sstk() in vm_mmap.c */
453 return -TARGET_EOPNOTSUPP;
454 }
455
456 #endif /* BSD_USER_BSD_MEM_H */
457