master
h 431 lines 11.6 KB
Raw
1 /*
2 * memory management system call shims and definitions
3 *
4 * Copyright (c) 2013-2015 Stacey D. Son
5 *
6 * SPDX-License-Identifier: GPL-2.0-or-later
7 */
8
9 /*
10 * Copyright (c) 1982, 1986, 1993
11 * The Regents of the University of California. All rights reserved.
12 *
13 * Redistribution and use in source and binary forms, with or without
14 * modification, are permitted provided that the following conditions
15 * are met:
16 * 1. Redistributions of source code must retain the above copyright
17 * notice, this list of conditions and the following disclaimer.
18 * 2. Redistributions in binary form must reproduce the above copyright
19 * notice, this list of conditions and the following disclaimer in the
20 * documentation and/or other materials provided with the distribution.
21 * 4. Neither the name of the University nor the names of its contributors
22 * may be used to endorse or promote products derived from this software
23 * without specific prior written permission.
24 *
25 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
26 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
27 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
28 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
29 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
30 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
31 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
32 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
33 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
34 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
35 * SUCH DAMAGE.
36 */
37
38 #ifndef BSD_USER_BSD_MEM_H
39 #define BSD_USER_BSD_MEM_H
40
41 #include <sys/ipc.h>
42 #include <sys/shm.h>
43
44 #include "qemu-bsd.h"
45 #include "exec/mmap-lock.h"
46 #include "exec/page-protection.h"
47 #include "user/page-protection.h"
48
49 extern struct bsd_shm_regions bsd_shm_regions[];
50 extern abi_ulong target_brk;
51 extern abi_ulong initial_target_brk;
52
53 /* mmap(2) */
54 static inline abi_long do_bsd_mmap(void *cpu_env, abi_long arg1, abi_long arg2,
55 abi_long arg3, abi_long arg4, abi_long arg5, abi_long arg6, abi_long arg7,
56 abi_long arg8)
57 {
58 if (regpairs_aligned(cpu_env) != 0) {
59 arg6 = arg7;
60 arg7 = arg8;
61 }
62 return get_errno(target_mmap(arg1, arg2, arg3,
63 target_to_host_bitmask(arg4, mmap_flags_tbl),
64 arg5, target_arg64(arg6, arg7)));
65 }
66
67 /* munmap(2) */
68 static inline abi_long do_bsd_munmap(abi_long arg1, abi_long arg2)
69 {
70 return get_errno(target_munmap(arg1, arg2));
71 }
72
73 /* mprotect(2) */
74 static inline abi_long do_bsd_mprotect(abi_long arg1, abi_long arg2,
75 abi_long arg3)
76 {
77 return get_errno(target_mprotect(arg1, arg2, arg3));
78 }
79
80 /* msync(2) */
81 static inline abi_long do_bsd_msync(abi_long addr, abi_long len, abi_long flags)
82 {
83 if (!guest_range_valid_untagged(addr, len)) {
84 /* It seems odd, but POSIX wants this to be ENOMEM */
85 return -TARGET_ENOMEM;
86 }
87
88 return get_errno(msync(g2h_untagged(addr), len, flags));
89 }
90
91 /* mlock(2) */
92 static inline abi_long do_bsd_mlock(abi_long arg1, abi_long arg2)
93 {
94 if (!guest_range_valid_untagged(arg1, arg2)) {
95 return -TARGET_EINVAL;
96 }
97 return get_errno(mlock(g2h_untagged(arg1), arg2));
98 }
99
100 /* munlock(2) */
101 static inline abi_long do_bsd_munlock(abi_long arg1, abi_long arg2)
102 {
103 if (!guest_range_valid_untagged(arg1, arg2)) {
104 return -TARGET_EINVAL;
105 }
106 return get_errno(munlock(g2h_untagged(arg1), arg2));
107 }
108
109 /* mlockall(2) */
110 static inline abi_long do_bsd_mlockall(abi_long arg1)
111 {
112 return get_errno(mlockall(arg1));
113 }
114
115 /* munlockall(2) */
116 static inline abi_long do_bsd_munlockall(void)
117 {
118 return get_errno(munlockall());
119 }
120
121 /* madvise(2) */
122 static inline abi_long do_bsd_madvise(abi_long arg1, abi_long arg2,
123 abi_long arg3)
124 {
125 abi_ulong len;
126 int ret = 0;
127 abi_long start = arg1;
128 abi_long len_in = arg2;
129 abi_long advice = arg3;
130
131 if (start & ~TARGET_PAGE_MASK) {
132 return -TARGET_EINVAL;
133 }
134 if (len_in == 0) {
135 return 0;
136 }
137 len = TARGET_PAGE_ALIGN(len_in);
138 if (len == 0 || !guest_range_valid_untagged(start, len)) {
139 return -TARGET_EINVAL;
140 }
141
142 /*
143 * Most advice values are hints, so ignoring and returning success is ok.
144 *
145 * However, some advice values such as MADV_DONTNEED, are not hints and
146 * need to be emulated.
147 *
148 * A straight passthrough for those may not be safe because qemu sometimes
149 * turns private file-backed mappings into anonymous mappings.
150 * If all guest pages have PAGE_PASSTHROUGH set, mappings have the
151 * same semantics for the host as for the guest.
152 *
153 * MADV_DONTNEED is passed through, if possible.
154 * If passthrough isn't possible, we nevertheless (wrongly!) return
155 * success, which is broken but some userspace programs fail to work
156 * otherwise. Completely implementing such emulation is quite complicated
157 * though.
158 */
159 mmap_lock();
160 switch (advice) {
161 case MADV_DONTNEED:
162 if (page_check_range(start, len, PAGE_PASSTHROUGH)) {
163 ret = get_errno(madvise(g2h_untagged(start), len, advice));
164 if (ret == 0) {
165 page_reset_target_data(start, start + len - 1);
166 }
167 }
168 }
169 mmap_unlock();
170
171 return ret;
172 }
173
174 /* minherit(2) */
175 static inline abi_long do_bsd_minherit(abi_long addr, abi_long len,
176 abi_long inherit)
177 {
178 return get_errno(minherit(g2h_untagged(addr), len, inherit));
179 }
180
181 /* mincore(2) */
182 static inline abi_long do_bsd_mincore(abi_ulong target_addr, abi_ulong len,
183 abi_ulong target_vec)
184 {
185 abi_long ret;
186 void *p;
187 abi_ulong vec_len = DIV_ROUND_UP(len, TARGET_PAGE_SIZE);
188
189 if (!guest_range_valid_untagged(target_addr, len)
190 || !page_check_range(target_addr, len, PAGE_VALID)) {
191 return -TARGET_EFAULT;
192 }
193
194 p = lock_user(VERIFY_WRITE, target_vec, vec_len, 0);
195 if (p == NULL) {
196 return -TARGET_EFAULT;
197 }
198 ret = get_errno(mincore(g2h_untagged(target_addr), len, p));
199 unlock_user(p, target_vec, vec_len);
200
201 return ret;
202 }
203
204 /* do_brk() must return target values and target errnos. */
205 static inline abi_long do_obreak(abi_ulong brk_val)
206 {
207 abi_long mapped_addr;
208 abi_ulong new_brk;
209 abi_ulong old_brk;
210
211 /* brk pointers are always untagged */
212
213 /* do not allow to shrink below initial brk value */
214 if (brk_val < initial_target_brk) {
215 return target_brk;
216 }
217
218 new_brk = TARGET_PAGE_ALIGN(brk_val);
219 old_brk = TARGET_PAGE_ALIGN(target_brk);
220
221 /* new and old target_brk might be on the same page */
222 if (new_brk == old_brk) {
223 target_brk = brk_val;
224 return target_brk;
225 }
226
227 /* Release heap if necessary */
228 if (new_brk < old_brk) {
229 target_munmap(new_brk, old_brk - new_brk);
230
231 target_brk = brk_val;
232 return target_brk;
233 }
234
235 mapped_addr = target_mmap(old_brk, new_brk - old_brk,
236 PROT_READ | PROT_WRITE,
237 MAP_FIXED | MAP_EXCL | MAP_ANON | MAP_PRIVATE,
238 -1, 0);
239
240 if (mapped_addr == old_brk) {
241 target_brk = brk_val;
242 return target_brk;
243 }
244
245 /* For everything else, return the previous break. */
246 return target_brk;
247 }
248
249 /* shm_open(2) */
250 static inline abi_long do_bsd_shm_open(abi_ulong arg1, abi_long arg2,
251 abi_long arg3)
252 {
253 int ret;
254 void *p;
255
256 if (arg1 == (uintptr_t)SHM_ANON) {
257 p = SHM_ANON;
258 } else {
259 p = lock_user_string(arg1);
260 if (p == NULL) {
261 return -TARGET_EFAULT;
262 }
263 }
264 ret = get_errno(shm_open(p, target_to_host_bitmask(arg2, fcntl_flags_tbl),
265 arg3));
266
267 if (p != SHM_ANON) {
268 unlock_user(p, arg1, 0);
269 }
270
271 return ret;
272 }
273
274 /* shm_unlink(2) */
275 static inline abi_long do_bsd_shm_unlink(abi_ulong arg1)
276 {
277 int ret;
278 void *p;
279
280 p = lock_user_string(arg1);
281 if (p == NULL) {
282 return -TARGET_EFAULT;
283 }
284 ret = get_errno(shm_unlink(p)); /* XXX path(p)? */
285 unlock_user(p, arg1, 0);
286
287 return ret;
288 }
289
290 /* shmget(2) */
291 static inline abi_long do_bsd_shmget(abi_long arg1, abi_ulong arg2,
292 abi_long arg3)
293 {
294 return get_errno(shmget(arg1, arg2, arg3));
295 }
296
297 /* shmctl(2) */
298 static inline abi_long do_bsd_shmctl(abi_long shmid, abi_long cmd,
299 abi_ulong buff)
300 {
301 struct shmid_ds dsarg;
302 abi_long ret = -TARGET_EINVAL;
303
304 cmd &= 0xff;
305
306 switch (cmd) {
307 case IPC_STAT:
308 if (target_to_host_shmid_ds(&dsarg, buff)) {
309 return -TARGET_EFAULT;
310 }
311 ret = get_errno(shmctl(shmid, cmd, &dsarg));
312 if (host_to_target_shmid_ds(buff, &dsarg)) {
313 return -TARGET_EFAULT;
314 }
315 break;
316
317 case IPC_SET:
318 if (target_to_host_shmid_ds(&dsarg, buff)) {
319 return -TARGET_EFAULT;
320 }
321 ret = get_errno(shmctl(shmid, cmd, &dsarg));
322 break;
323
324 case IPC_RMID:
325 ret = get_errno(shmctl(shmid, cmd, NULL));
326 break;
327
328 default:
329 ret = -TARGET_EINVAL;
330 break;
331 }
332
333 return ret;
334 }
335
336 /* shmat(2) */
337 static inline abi_long do_bsd_shmat(int shmid, abi_ulong shmaddr, int shmflg)
338 {
339 abi_ulong raddr;
340 abi_long ret;
341 struct shmid_ds shm_info;
342
343 /* Find out the length of the shared memory segment. */
344 ret = get_errno(shmctl(shmid, IPC_STAT, &shm_info));
345 if (is_error(ret)) {
346 /* Can't get the length */
347 return ret;
348 }
349
350 if (!guest_range_valid_untagged(shmaddr, shm_info.shm_segsz)) {
351 return -TARGET_EINVAL;
352 }
353
354 WITH_MMAP_LOCK_GUARD() {
355 void *host_raddr;
356
357 if (shmaddr) {
358 host_raddr = shmat(shmid, (void *)g2h_untagged(shmaddr), shmflg);
359 } else {
360 abi_ulong alignment;
361 abi_ulong mmap_start;
362
363 alignment = 0; /* alignment above page size not required */
364 mmap_start = mmap_find_vma(0, shm_info.shm_segsz, alignment);
365
366 if (mmap_start == -1) {
367 return -TARGET_ENOMEM;
368 }
369 host_raddr = shmat(shmid, g2h_untagged(mmap_start),
370 shmflg | SHM_REMAP);
371 }
372
373 if (host_raddr == (void *)-1) {
374 return get_errno(-1);
375 }
376 raddr = h2g(host_raddr);
377
378 page_set_flags(raddr, raddr + shm_info.shm_segsz - 1,
379 PAGE_VALID | PAGE_READ |
380 (shmflg & SHM_RDONLY ? 0 : PAGE_WRITE),
381 PAGE_VALID);
382
383 for (int i = 0; i < N_BSD_SHM_REGIONS; i++) {
384 if (bsd_shm_regions[i].start == 0) {
385 bsd_shm_regions[i].start = raddr;
386 bsd_shm_regions[i].size = shm_info.shm_segsz;
387 break;
388 }
389 }
390 }
391
392 return raddr;
393 }
394
395 /* shmdt(2) */
396 static inline abi_long do_bsd_shmdt(abi_ulong shmaddr)
397 {
398 abi_long ret;
399
400 WITH_MMAP_LOCK_GUARD() {
401 int i;
402
403 for (i = 0; i < N_BSD_SHM_REGIONS; ++i) {
404 if (bsd_shm_regions[i].start == shmaddr) {
405 break;
406 }
407 }
408
409 if (i == N_BSD_SHM_REGIONS) {
410 return -TARGET_EINVAL;
411 }
412
413 ret = get_errno(shmdt(g2h_untagged(shmaddr)));
414 if (ret == 0) {
415 abi_ulong size = bsd_shm_regions[i].size;
416
417 bsd_shm_regions[i].start = 0;
418 page_set_flags(shmaddr, shmaddr + size - 1, 0, PAGE_VALID);
419 mmap_reserve(shmaddr, size);
420 }
421 }
422
423 return ret;
424 }
425
426 static inline abi_long do_bsd_vadvise(void)
427 {
428 /* See sys_ovadvise() in vm_unix.c */
429 return -TARGET_EINVAL;
430 }
431 #endif /* BSD_USER_BSD_MEM_H */