xref: /openbmc/qemu/bsd-user/freebsd/os-syscall.c (revision ed3a06b1)
1 /*
2  *  BSD syscalls
3  *
4  *  Copyright (c) 2003-2008 Fabrice Bellard
5  *  Copyright (c) 2013-2014 Stacey D. Son
6  *
7  *  This program is free software; you can redistribute it and/or modify
8  *  it under the terms of the GNU General Public License as published by
9  *  the Free Software Foundation; either version 2 of the License, or
10  *  (at your option) any later version.
11  *
12  *  This program is distributed in the hope that it will be useful,
13  *  but WITHOUT ANY WARRANTY; without even the implied warranty of
14  *  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
15  *  GNU General Public License for more details.
16  *
17  *  You should have received a copy of the GNU General Public License
18  *  along with this program; if not, see <http://www.gnu.org/licenses/>.
19  */
20 
21 /*
22  * We need the FreeBSD "legacy" definitions. Rust needs the FreeBSD 11 system
23  * calls since it doesn't use libc at all, so we have to emulate that despite
24  * FreeBSD 11 being EOL'd.
25  */
26 #define _WANT_FREEBSD11_STAT
27 #define _WANT_FREEBSD11_STATFS
28 #define _WANT_FREEBSD11_DIRENT
29 #define _WANT_KERNEL_ERRNO
30 #define _WANT_SEMUN
31 #include "qemu/osdep.h"
32 #include "qemu/cutils.h"
33 #include "qemu/path.h"
34 #include <sys/syscall.h>
35 #include <sys/param.h>
36 #include <sys/sysctl.h>
37 #include <utime.h>
38 
39 #include "qemu.h"
40 #include "signal-common.h"
41 #include "user/syscall-trace.h"
42 
43 #include "bsd-file.h"
44 #include "bsd-proc.h"
45 
46 /* I/O */
47 safe_syscall3(int, open, const char *, path, int, flags, mode_t, mode);
48 safe_syscall4(int, openat, int, fd, const char *, path, int, flags, mode_t,
49     mode);
50 
51 safe_syscall3(ssize_t, read, int, fd, void *, buf, size_t, nbytes);
52 safe_syscall4(ssize_t, pread, int, fd, void *, buf, size_t, nbytes, off_t,
53     offset);
54 safe_syscall3(ssize_t, readv, int, fd, const struct iovec *, iov, int, iovcnt);
55 safe_syscall4(ssize_t, preadv, int, fd, const struct iovec *, iov, int, iovcnt,
56     off_t, offset);
57 
58 safe_syscall3(ssize_t, write, int, fd, void *, buf, size_t, nbytes);
59 safe_syscall4(ssize_t, pwrite, int, fd, void *, buf, size_t, nbytes, off_t,
60     offset);
61 safe_syscall3(ssize_t, writev, int, fd, const struct iovec *, iov, int, iovcnt);
62 safe_syscall4(ssize_t, pwritev, int, fd, const struct iovec *, iov, int, iovcnt,
63     off_t, offset);
64 
65 void target_set_brk(abi_ulong new_brk)
66 {
67 }
68 
69 /*
70  * errno conversion.
71  */
72 abi_long get_errno(abi_long ret)
73 {
74     if (ret == -1) {
75         return -host_to_target_errno(errno);
76     } else {
77         return ret;
78     }
79 }
80 
81 int host_to_target_errno(int err)
82 {
83     /*
84      * All the BSDs have the property that the error numbers are uniform across
85      * all architectures for a given BSD, though they may vary between different
86      * BSDs.
87      */
88     return err;
89 }
90 
91 bool is_error(abi_long ret)
92 {
93     return (abi_ulong)ret >= (abi_ulong)(-4096);
94 }
95 
96 /*
97  * Unlocks a iovec. Unlike unlock_iovec, it assumes the tvec array itself is
98  * already locked from target_addr. It will be unlocked as well as all the iovec
99  * elements.
100  */
101 static void helper_unlock_iovec(struct target_iovec *target_vec,
102                                 abi_ulong target_addr, struct iovec *vec,
103                                 int count, int copy)
104 {
105     for (int i = 0; i < count; i++) {
106         abi_ulong base = tswapal(target_vec[i].iov_base);
107 
108         if (vec[i].iov_base) {
109             unlock_user(vec[i].iov_base, base, copy ? vec[i].iov_len : 0);
110         }
111     }
112     unlock_user(target_vec, target_addr, 0);
113 }
114 
115 struct iovec *lock_iovec(int type, abi_ulong target_addr,
116         int count, int copy)
117 {
118     struct target_iovec *target_vec;
119     struct iovec *vec;
120     abi_ulong total_len, max_len;
121     int i;
122     int err = 0;
123 
124     if (count == 0) {
125         errno = 0;
126         return NULL;
127     }
128     if (count < 0 || count > IOV_MAX) {
129         errno = EINVAL;
130         return NULL;
131     }
132 
133     vec = g_try_new0(struct iovec, count);
134     if (vec == NULL) {
135         errno = ENOMEM;
136         return NULL;
137     }
138 
139     target_vec = lock_user(VERIFY_READ, target_addr,
140                            count * sizeof(struct target_iovec), 1);
141     if (target_vec == NULL) {
142         err = EFAULT;
143         goto fail2;
144     }
145 
146     max_len = 0x7fffffff & MIN(TARGET_PAGE_MASK, PAGE_MASK);
147     total_len = 0;
148 
149     for (i = 0; i < count; i++) {
150         abi_ulong base = tswapal(target_vec[i].iov_base);
151         abi_long len = tswapal(target_vec[i].iov_len);
152 
153         if (len < 0) {
154             err = EINVAL;
155             goto fail;
156         } else if (len == 0) {
157             /* Zero length pointer is ignored. */
158             vec[i].iov_base = 0;
159         } else {
160             vec[i].iov_base = lock_user(type, base, len, copy);
161             /*
162              * If the first buffer pointer is bad, this is a fault.  But
163              * subsequent bad buffers will result in a partial write; this is
164              * realized by filling the vector with null pointers and zero
165              * lengths.
166              */
167             if (!vec[i].iov_base) {
168                 if (i == 0) {
169                     err = EFAULT;
170                     goto fail;
171                 } else {
172                     /*
173                      * Fail all the subsequent addresses, they are already
174                      * zero'd.
175                      */
176                     goto out;
177                 }
178             }
179             if (len > max_len - total_len) {
180                 len = max_len - total_len;
181             }
182         }
183         vec[i].iov_len = len;
184         total_len += len;
185     }
186 out:
187     unlock_user(target_vec, target_addr, 0);
188     return vec;
189 
190 fail:
191     helper_unlock_iovec(target_vec, target_addr, vec, i, copy);
192 fail2:
193     g_free(vec);
194     errno = err;
195     return NULL;
196 }
197 
198 void unlock_iovec(struct iovec *vec, abi_ulong target_addr,
199         int count, int copy)
200 {
201     struct target_iovec *target_vec;
202 
203     target_vec = lock_user(VERIFY_READ, target_addr,
204                            count * sizeof(struct target_iovec), 1);
205     if (target_vec) {
206         helper_unlock_iovec(target_vec, target_addr, vec, count, copy);
207     }
208 
209     g_free(vec);
210 }
211 
212 /*
213  * All errnos that freebsd_syscall() returns must be -TARGET_<errcode>.
214  */
215 static abi_long freebsd_syscall(void *cpu_env, int num, abi_long arg1,
216                                 abi_long arg2, abi_long arg3, abi_long arg4,
217                                 abi_long arg5, abi_long arg6, abi_long arg7,
218                                 abi_long arg8)
219 {
220     abi_long ret;
221 
222     switch (num) {
223         /*
224          * process system calls
225          */
226     case TARGET_FREEBSD_NR_exit: /* exit(2) */
227         ret = do_bsd_exit(cpu_env, arg1);
228         break;
229 
230         /*
231          * File system calls.
232          */
233     case TARGET_FREEBSD_NR_read: /* read(2) */
234         ret = do_bsd_read(arg1, arg2, arg3);
235         break;
236 
237     case TARGET_FREEBSD_NR_pread: /* pread(2) */
238         ret = do_bsd_pread(cpu_env, arg1, arg2, arg3, arg4, arg5, arg6);
239         break;
240 
241     case TARGET_FREEBSD_NR_readv: /* readv(2) */
242         ret = do_bsd_readv(arg1, arg2, arg3);
243         break;
244 
245     case TARGET_FREEBSD_NR_preadv: /* preadv(2) */
246         ret = do_bsd_preadv(cpu_env, arg1, arg2, arg3, arg4, arg5, arg6);
247 
248     case TARGET_FREEBSD_NR_write: /* write(2) */
249         ret = do_bsd_write(arg1, arg2, arg3);
250         break;
251 
252     case TARGET_FREEBSD_NR_pwrite: /* pwrite(2) */
253         ret = do_bsd_pwrite(cpu_env, arg1, arg2, arg3, arg4, arg5, arg6);
254         break;
255 
256     case TARGET_FREEBSD_NR_writev: /* writev(2) */
257         ret = do_bsd_writev(arg1, arg2, arg3);
258         break;
259 
260     case TARGET_FREEBSD_NR_pwritev: /* pwritev(2) */
261         ret = do_bsd_pwritev(cpu_env, arg1, arg2, arg3, arg4, arg5, arg6);
262         break;
263 
264     case TARGET_FREEBSD_NR_open: /* open(2) */
265         ret = do_bsd_open(arg1, arg2, arg3);
266         break;
267 
268     case TARGET_FREEBSD_NR_openat: /* openat(2) */
269         ret = do_bsd_openat(arg1, arg2, arg3, arg4);
270         break;
271 
272     case TARGET_FREEBSD_NR_close: /* close(2) */
273         ret = do_bsd_close(arg1);
274         break;
275 
276     case TARGET_FREEBSD_NR_fdatasync: /* fdatasync(2) */
277         ret = do_bsd_fdatasync(arg1);
278         break;
279 
280     case TARGET_FREEBSD_NR_fsync: /* fsync(2) */
281         ret = do_bsd_fsync(arg1);
282         break;
283 
284     case TARGET_FREEBSD_NR_freebsd12_closefrom: /* closefrom(2) */
285         ret = do_bsd_closefrom(arg1);
286         break;
287 
288     case TARGET_FREEBSD_NR_revoke: /* revoke(2) */
289         ret = do_bsd_revoke(arg1);
290         break;
291 
292     case TARGET_FREEBSD_NR_access: /* access(2) */
293         ret = do_bsd_access(arg1, arg2);
294         break;
295 
296     case TARGET_FREEBSD_NR_eaccess: /* eaccess(2) */
297         ret = do_bsd_eaccess(arg1, arg2);
298         break;
299 
300     case TARGET_FREEBSD_NR_faccessat: /* faccessat(2) */
301         ret = do_bsd_faccessat(arg1, arg2, arg3, arg4);
302         break;
303 
304     case TARGET_FREEBSD_NR_chdir: /* chdir(2) */
305         ret = do_bsd_chdir(arg1);
306         break;
307 
308     case TARGET_FREEBSD_NR_fchdir: /* fchdir(2) */
309         ret = do_bsd_fchdir(arg1);
310         break;
311 
312     case TARGET_FREEBSD_NR_rename: /* rename(2) */
313         ret = do_bsd_rename(arg1, arg2);
314         break;
315 
316     case TARGET_FREEBSD_NR_renameat: /* renameat(2) */
317         ret = do_bsd_renameat(arg1, arg2, arg3, arg4);
318         break;
319 
320     case TARGET_FREEBSD_NR_link: /* link(2) */
321         ret = do_bsd_link(arg1, arg2);
322         break;
323 
324     case TARGET_FREEBSD_NR_linkat: /* linkat(2) */
325         ret = do_bsd_linkat(arg1, arg2, arg3, arg4, arg5);
326         break;
327 
328     case TARGET_FREEBSD_NR_unlink: /* unlink(2) */
329         ret = do_bsd_unlink(arg1);
330         break;
331 
332     case TARGET_FREEBSD_NR_unlinkat: /* unlinkat(2) */
333         ret = do_bsd_unlinkat(arg1, arg2, arg3);
334         break;
335 
336     case TARGET_FREEBSD_NR_mkdir: /* mkdir(2) */
337         ret = do_bsd_mkdir(arg1, arg2);
338         break;
339 
340     case TARGET_FREEBSD_NR_mkdirat: /* mkdirat(2) */
341         ret = do_bsd_mkdirat(arg1, arg2, arg3);
342         break;
343 
344     case TARGET_FREEBSD_NR_rmdir: /* rmdir(2) (XXX no rmdirat()?) */
345         ret = do_bsd_rmdir(arg1);
346         break;
347 
348     case TARGET_FREEBSD_NR___getcwd: /* undocumented __getcwd() */
349         ret = do_bsd___getcwd(arg1, arg2);
350         break;
351 
352     case TARGET_FREEBSD_NR_dup: /* dup(2) */
353         ret = do_bsd_dup(arg1);
354         break;
355 
356     case TARGET_FREEBSD_NR_dup2: /* dup2(2) */
357         ret = do_bsd_dup2(arg1, arg2);
358         break;
359 
360     case TARGET_FREEBSD_NR_truncate: /* truncate(2) */
361         ret = do_bsd_truncate(cpu_env, arg1, arg2, arg3, arg4);
362         break;
363 
364     case TARGET_FREEBSD_NR_ftruncate: /* ftruncate(2) */
365         ret = do_bsd_ftruncate(cpu_env, arg1, arg2, arg3, arg4);
366         break;
367 
368     case TARGET_FREEBSD_NR_acct: /* acct(2) */
369         ret = do_bsd_acct(arg1);
370         break;
371 
372     case TARGET_FREEBSD_NR_sync: /* sync(2) */
373         ret = do_bsd_sync();
374         break;
375 
376     default:
377         qemu_log_mask(LOG_UNIMP, "Unsupported syscall: %d\n", num);
378         ret = -TARGET_ENOSYS;
379         break;
380     }
381 
382     return ret;
383 }
384 
385 /*
386  * do_freebsd_syscall() should always have a single exit point at the end so
387  * that actions, such as logging of syscall results, can be performed. This
388  * as a wrapper around freebsd_syscall() so that actually happens. Since
389  * that is a singleton, modern compilers will inline it anyway...
390  */
391 abi_long do_freebsd_syscall(void *cpu_env, int num, abi_long arg1,
392                             abi_long arg2, abi_long arg3, abi_long arg4,
393                             abi_long arg5, abi_long arg6, abi_long arg7,
394                             abi_long arg8)
395 {
396     CPUState *cpu = env_cpu(cpu_env);
397     int ret;
398 
399     trace_guest_user_syscall(cpu, num, arg1, arg2, arg3, arg4, arg5, arg6, arg7, arg8);
400     if (do_strace) {
401         print_freebsd_syscall(num, arg1, arg2, arg3, arg4, arg5, arg6);
402     }
403 
404     ret = freebsd_syscall(cpu_env, num, arg1, arg2, arg3, arg4, arg5, arg6,
405                           arg7, arg8);
406     if (do_strace) {
407         print_freebsd_syscall_ret(num, ret);
408     }
409     trace_guest_user_syscall_ret(cpu, num, ret);
410 
411     return ret;
412 }
413 
414 void syscall_init(void)
415 {
416 }
417