git://git.qemu.org / qemu.git / blob
? search:
summary | shortlog | log | commit | commitdiff | tree
blame | history | raw | HEAD
linux-user/: fix some comment spelling errors
[qemu.git] / linux-user / syscall.c
1 /*
2 * Linux syscalls
3 *
4 * Copyright (c) 2003 Fabrice Bellard
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 #define _ATFILE_SOURCE
20 #include "qemu/osdep.h"
21 #include "qemu/cutils.h"
22 #include "qemu/path.h"
23 #include "qemu/memfd.h"
24 #include "qemu/queue.h"
25 #include <elf.h>
26 #include <endian.h>
27 #include <grp.h>
28 #include <sys/ipc.h>
29 #include <sys/msg.h>
30 #include <sys/wait.h>
31 #include <sys/mount.h>
32 #include <sys/file.h>
33 #include <sys/fsuid.h>
34 #include <sys/personality.h>
35 #include <sys/prctl.h>
36 #include <sys/resource.h>
37 #include <sys/swap.h>
38 #include <linux/capability.h>
39 #include <sched.h>
40 #include <sys/timex.h>
41 #include <sys/socket.h>
42 #include <linux/sockios.h>
43 #include <sys/un.h>
44 #include <sys/uio.h>
45 #include <poll.h>
46 #include <sys/times.h>
47 #include <sys/shm.h>
48 #include <sys/sem.h>
49 #include <sys/statfs.h>
50 #include <utime.h>
51 #include <sys/sysinfo.h>
52 #include <sys/signalfd.h>
53 //#include <sys/user.h>
54 #include <netinet/ip.h>
55 #include <netinet/tcp.h>
56 #include <linux/wireless.h>
57 #include <linux/icmp.h>
58 #include <linux/icmpv6.h>
59 #include <linux/errqueue.h>
60 #include <linux/random.h>
61 #ifdef CONFIG_TIMERFD
62 #include <sys/timerfd.h>
63 #endif
64 #ifdef CONFIG_EVENTFD
65 #include <sys/eventfd.h>
66 #endif
67 #ifdef CONFIG_EPOLL
68 #include <sys/epoll.h>
69 #endif
70 #ifdef CONFIG_ATTR
71 #include "qemu/xattr.h"
72 #endif
73 #ifdef CONFIG_SENDFILE
74 #include <sys/sendfile.h>
75 #endif
76 #ifdef CONFIG_KCOV
77 #include <sys/kcov.h>
78 #endif
79
80 #define termios host_termios
81 #define winsize host_winsize
82 #define termio host_termio
83 #define sgttyb host_sgttyb /* same as target */
84 #define tchars host_tchars /* same as target */
85 #define ltchars host_ltchars /* same as target */
86
87 #include <linux/termios.h>
88 #include <linux/unistd.h>
89 #include <linux/cdrom.h>
90 #include <linux/hdreg.h>
91 #include <linux/soundcard.h>
92 #include <linux/kd.h>
93 #include <linux/mtio.h>
94 #include <linux/fs.h>
95 #include <linux/fd.h>
96 #if defined(CONFIG_FIEMAP)
97 #include <linux/fiemap.h>
98 #endif
99 #include <linux/fb.h>
100 #if defined(CONFIG_USBFS)
101 #include <linux/usbdevice_fs.h>
102 #include <linux/usb/ch9.h>
103 #endif
104 #include <linux/vt.h>
105 #include <linux/dm-ioctl.h>
106 #include <linux/reboot.h>
107 #include <linux/route.h>
108 #include <linux/filter.h>
109 #include <linux/blkpg.h>
110 #include <netpacket/packet.h>
111 #include <linux/netlink.h>
112 #include <linux/if_alg.h>
113 #include <linux/rtc.h>
114 #include <sound/asound.h>
115 #ifdef CONFIG_BTRFS
116 #include <linux/btrfs.h>
117 #endif
118 #ifdef HAVE_DRM_H
119 #include <libdrm/drm.h>
120 #include <libdrm/i915_drm.h>
121 #endif
122 #include "linux_loop.h"
123 #include "uname.h"
124
125 #include "qemu.h"
126 #include "qemu/guest-random.h"
127 #include "qemu/selfmap.h"
128 #include "user/syscall-trace.h"
129 #include "qapi/error.h"
130 #include "fd-trans.h"
131 #include "tcg/tcg.h"
132
133 #ifndef CLONE_IO
134 #define CLONE_IO 0x80000000 /* Clone io context */
135 #endif
136
137 /* We can't directly call the host clone syscall, because this will
138 * badly confuse libc (breaking mutexes, for example). So we must
139 * divide clone flags into:
140 * * flag combinations that look like pthread_create()
141 * * flag combinations that look like fork()
142 * * flags we can implement within QEMU itself
143 * * flags we can't support and will return an error for
144 */
145 /* For thread creation, all these flags must be present; for
146 * fork, none must be present.
147 */
148 #define CLONE_THREAD_FLAGS \
149 (CLONE_VM | CLONE_FS | CLONE_FILES | \
150 CLONE_SIGHAND | CLONE_THREAD | CLONE_SYSVSEM)
151
152 /* These flags are ignored:
153 * CLONE_DETACHED is now ignored by the kernel;
154 * CLONE_IO is just an optimisation hint to the I/O scheduler
155 */
156 #define CLONE_IGNORED_FLAGS \
157 (CLONE_DETACHED | CLONE_IO)
158
159 /* Flags for fork which we can implement within QEMU itself */
160 #define CLONE_OPTIONAL_FORK_FLAGS \
161 (CLONE_SETTLS | CLONE_PARENT_SETTID | \
162 CLONE_CHILD_CLEARTID | CLONE_CHILD_SETTID)
163
164 /* Flags for thread creation which we can implement within QEMU itself */
165 #define CLONE_OPTIONAL_THREAD_FLAGS \
166 (CLONE_SETTLS | CLONE_PARENT_SETTID | \
167 CLONE_CHILD_CLEARTID | CLONE_CHILD_SETTID | CLONE_PARENT)
168
169 #define CLONE_INVALID_FORK_FLAGS \
170 (~(CSIGNAL | CLONE_OPTIONAL_FORK_FLAGS | CLONE_IGNORED_FLAGS))
171
172 #define CLONE_INVALID_THREAD_FLAGS \
173 (~(CSIGNAL | CLONE_THREAD_FLAGS | CLONE_OPTIONAL_THREAD_FLAGS | \
174 CLONE_IGNORED_FLAGS))
175
176 /* CLONE_VFORK is special cased early in do_fork(). The other flag bits
177 * have almost all been allocated. We cannot support any of
178 * CLONE_NEWNS, CLONE_NEWCGROUP, CLONE_NEWUTS, CLONE_NEWIPC,
179 * CLONE_NEWUSER, CLONE_NEWPID, CLONE_NEWNET, CLONE_PTRACE, CLONE_UNTRACED.
180 * The checks against the invalid thread masks above will catch these.
181 * (The one remaining unallocated bit is 0x1000 which used to be CLONE_PID.)
182 */
183
184 /* Define DEBUG_ERESTARTSYS to force every syscall to be restarted
185 * once. This exercises the codepaths for restart.
186 */
187 //#define DEBUG_ERESTARTSYS
188
189 //#include <linux/msdos_fs.h>
190 #define VFAT_IOCTL_READDIR_BOTH _IOR('r', 1, struct linux_dirent [2])
191 #define VFAT_IOCTL_READDIR_SHORT _IOR('r', 2, struct linux_dirent [2])
192
193 #undef _syscall0
194 #undef _syscall1
195 #undef _syscall2
196 #undef _syscall3
197 #undef _syscall4
198 #undef _syscall5
199 #undef _syscall6
200
201 #define _syscall0(type,name) \
202 static type name (void) \
203 { \
204 return syscall(__NR_##name); \
205 }
206
207 #define _syscall1(type,name,type1,arg1) \
208 static type name (type1 arg1) \
209 { \
210 return syscall(__NR_##name, arg1); \
211 }
212
213 #define _syscall2(type,name,type1,arg1,type2,arg2) \
214 static type name (type1 arg1,type2 arg2) \
215 { \
216 return syscall(__NR_##name, arg1, arg2); \
217 }
218
219 #define _syscall3(type,name,type1,arg1,type2,arg2,type3,arg3) \
220 static type name (type1 arg1,type2 arg2,type3 arg3) \
221 { \
222 return syscall(__NR_##name, arg1, arg2, arg3); \
223 }
224
225 #define _syscall4(type,name,type1,arg1,type2,arg2,type3,arg3,type4,arg4) \
226 static type name (type1 arg1,type2 arg2,type3 arg3,type4 arg4) \
227 { \
228 return syscall(__NR_##name, arg1, arg2, arg3, arg4); \
229 }
230
231 #define _syscall5(type,name,type1,arg1,type2,arg2,type3,arg3,type4,arg4, \
232 type5,arg5) \
233 static type name (type1 arg1,type2 arg2,type3 arg3,type4 arg4,type5 arg5) \
234 { \
235 return syscall(__NR_##name, arg1, arg2, arg3, arg4, arg5); \
236 }
237
238
239 #define _syscall6(type,name,type1,arg1,type2,arg2,type3,arg3,type4,arg4, \
240 type5,arg5,type6,arg6) \
241 static type name (type1 arg1,type2 arg2,type3 arg3,type4 arg4,type5 arg5, \
242 type6 arg6) \
243 { \
244 return syscall(__NR_##name, arg1, arg2, arg3, arg4, arg5, arg6); \
245 }
246
247
248 #define __NR_sys_uname __NR_uname
249 #define __NR_sys_getcwd1 __NR_getcwd
250 #define __NR_sys_getdents __NR_getdents
251 #define __NR_sys_getdents64 __NR_getdents64
252 #define __NR_sys_getpriority __NR_getpriority
253 #define __NR_sys_rt_sigqueueinfo __NR_rt_sigqueueinfo
254 #define __NR_sys_rt_tgsigqueueinfo __NR_rt_tgsigqueueinfo
255 #define __NR_sys_syslog __NR_syslog
256 #if defined(__NR_futex)
257 # define __NR_sys_futex __NR_futex
258 #endif
259 #if defined(__NR_futex_time64)
260 # define __NR_sys_futex_time64 __NR_futex_time64
261 #endif
262 #define __NR_sys_inotify_init __NR_inotify_init
263 #define __NR_sys_inotify_add_watch __NR_inotify_add_watch
264 #define __NR_sys_inotify_rm_watch __NR_inotify_rm_watch
265 #define __NR_sys_statx __NR_statx
266
267 #if defined(__alpha__) || defined(__x86_64__) || defined(__s390x__)
268 #define __NR__llseek __NR_lseek
269 #endif
270
271 /* Newer kernel ports have llseek() instead of _llseek() */
272 #if defined(TARGET_NR_llseek) && !defined(TARGET_NR__llseek)
273 #define TARGET_NR__llseek TARGET_NR_llseek
274 #endif
275
276 #define __NR_sys_gettid __NR_gettid
277 _syscall0(int, sys_gettid)
278
279 /* For the 64-bit guest on 32-bit host case we must emulate
280 * getdents using getdents64, because otherwise the host
281 * might hand us back more dirent records than we can fit
282 * into the guest buffer after structure format conversion.
283 * Otherwise we emulate getdents with getdents if the host has it.
284 */
285 #if defined(__NR_getdents) && HOST_LONG_BITS >= TARGET_ABI_BITS
286 #define EMULATE_GETDENTS_WITH_GETDENTS
287 #endif
288
289 #if defined(TARGET_NR_getdents) && defined(EMULATE_GETDENTS_WITH_GETDENTS)
290 _syscall3(int, sys_getdents, uint, fd, struct linux_dirent *, dirp, uint, count);
291 #endif
292 #if (defined(TARGET_NR_getdents) && \
293 !defined(EMULATE_GETDENTS_WITH_GETDENTS)) || \
294 (defined(TARGET_NR_getdents64) && defined(__NR_getdents64))
295 _syscall3(int, sys_getdents64, uint, fd, struct linux_dirent64 *, dirp, uint, count);
296 #endif
297 #if defined(TARGET_NR__llseek) && defined(__NR_llseek)
298 _syscall5(int, _llseek, uint, fd, ulong, hi, ulong, lo,
299 loff_t *, res, uint, wh);
300 #endif
301 _syscall3(int, sys_rt_sigqueueinfo, pid_t, pid, int, sig, siginfo_t *, uinfo)
302 _syscall4(int, sys_rt_tgsigqueueinfo, pid_t, pid, pid_t, tid, int, sig,
303 siginfo_t *, uinfo)
304 _syscall3(int,sys_syslog,int,type,char*,bufp,int,len)
305 #ifdef __NR_exit_group
306 _syscall1(int,exit_group,int,error_code)
307 #endif
308 #if defined(TARGET_NR_set_tid_address) && defined(__NR_set_tid_address)
309 _syscall1(int,set_tid_address,int *,tidptr)
310 #endif
311 #if defined(__NR_futex)
312 _syscall6(int,sys_futex,int *,uaddr,int,op,int,val,
313 const struct timespec *,timeout,int *,uaddr2,int,val3)
314 #endif
315 #if defined(__NR_futex_time64)
316 _syscall6(int,sys_futex_time64,int *,uaddr,int,op,int,val,
317 const struct timespec *,timeout,int *,uaddr2,int,val3)
318 #endif
319 #define __NR_sys_sched_getaffinity __NR_sched_getaffinity
320 _syscall3(int, sys_sched_getaffinity, pid_t, pid, unsigned int, len,
321 unsigned long *, user_mask_ptr);
322 #define __NR_sys_sched_setaffinity __NR_sched_setaffinity
323 _syscall3(int, sys_sched_setaffinity, pid_t, pid, unsigned int, len,
324 unsigned long *, user_mask_ptr);
325 #define __NR_sys_getcpu __NR_getcpu
326 _syscall3(int, sys_getcpu, unsigned *, cpu, unsigned *, node, void *, tcache);
327 _syscall4(int, reboot, int, magic1, int, magic2, unsigned int, cmd,
328 void *, arg);
329 _syscall2(int, capget, struct __user_cap_header_struct *, header,
330 struct __user_cap_data_struct *, data);
331 _syscall2(int, capset, struct __user_cap_header_struct *, header,
332 struct __user_cap_data_struct *, data);
333 #if defined(TARGET_NR_ioprio_get) && defined(__NR_ioprio_get)
334 _syscall2(int, ioprio_get, int, which, int, who)
335 #endif
336 #if defined(TARGET_NR_ioprio_set) && defined(__NR_ioprio_set)
337 _syscall3(int, ioprio_set, int, which, int, who, int, ioprio)
338 #endif
339 #if defined(TARGET_NR_getrandom) && defined(__NR_getrandom)
340 _syscall3(int, getrandom, void *, buf, size_t, buflen, unsigned int, flags)
341 #endif
342
343 #if defined(TARGET_NR_kcmp) && defined(__NR_kcmp)
344 _syscall5(int, kcmp, pid_t, pid1, pid_t, pid2, int, type,
345 unsigned long, idx1, unsigned long, idx2)
346 #endif
347
348 /*
349 * It is assumed that struct statx is architecture independent.
350 */
351 #if defined(TARGET_NR_statx) && defined(__NR_statx)
352 _syscall5(int, sys_statx, int, dirfd, const char *, pathname, int, flags,
353 unsigned int, mask, struct target_statx *, statxbuf)
354 #endif
355 #if defined(TARGET_NR_membarrier) && defined(__NR_membarrier)
356 _syscall2(int, membarrier, int, cmd, int, flags)
357 #endif
358
359 static bitmask_transtbl fcntl_flags_tbl[] = {
360 { TARGET_O_ACCMODE, TARGET_O_WRONLY, O_ACCMODE, O_WRONLY, },
361 { TARGET_O_ACCMODE, TARGET_O_RDWR, O_ACCMODE, O_RDWR, },
362 { TARGET_O_CREAT, TARGET_O_CREAT, O_CREAT, O_CREAT, },
363 { TARGET_O_EXCL, TARGET_O_EXCL, O_EXCL, O_EXCL, },
364 { TARGET_O_NOCTTY, TARGET_O_NOCTTY, O_NOCTTY, O_NOCTTY, },
365 { TARGET_O_TRUNC, TARGET_O_TRUNC, O_TRUNC, O_TRUNC, },
366 { TARGET_O_APPEND, TARGET_O_APPEND, O_APPEND, O_APPEND, },
367 { TARGET_O_NONBLOCK, TARGET_O_NONBLOCK, O_NONBLOCK, O_NONBLOCK, },
368 { TARGET_O_SYNC, TARGET_O_DSYNC, O_SYNC, O_DSYNC, },
369 { TARGET_O_SYNC, TARGET_O_SYNC, O_SYNC, O_SYNC, },
370 { TARGET_FASYNC, TARGET_FASYNC, FASYNC, FASYNC, },
371 { TARGET_O_DIRECTORY, TARGET_O_DIRECTORY, O_DIRECTORY, O_DIRECTORY, },
372 { TARGET_O_NOFOLLOW, TARGET_O_NOFOLLOW, O_NOFOLLOW, O_NOFOLLOW, },
373 #if defined(O_DIRECT)
374 { TARGET_O_DIRECT, TARGET_O_DIRECT, O_DIRECT, O_DIRECT, },
375 #endif
376 #if defined(O_NOATIME)
377 { TARGET_O_NOATIME, TARGET_O_NOATIME, O_NOATIME, O_NOATIME },
378 #endif
379 #if defined(O_CLOEXEC)
380 { TARGET_O_CLOEXEC, TARGET_O_CLOEXEC, O_CLOEXEC, O_CLOEXEC },
381 #endif
382 #if defined(O_PATH)
383 { TARGET_O_PATH, TARGET_O_PATH, O_PATH, O_PATH },
384 #endif
385 #if defined(O_TMPFILE)
386 { TARGET_O_TMPFILE, TARGET_O_TMPFILE, O_TMPFILE, O_TMPFILE },
387 #endif
388 /* Don't terminate the list prematurely on 64-bit host+guest. */
389 #if TARGET_O_LARGEFILE != 0 || O_LARGEFILE != 0
390 { TARGET_O_LARGEFILE, TARGET_O_LARGEFILE, O_LARGEFILE, O_LARGEFILE, },
391 #endif
392 { 0, 0, 0, 0 }
393 };
394
395 _syscall2(int, sys_getcwd1, char *, buf, size_t, size)
396
397 #if defined(TARGET_NR_utimensat) || defined(TARGET_NR_utimensat_time64)
398 #if defined(__NR_utimensat)
399 #define __NR_sys_utimensat __NR_utimensat
400 _syscall4(int,sys_utimensat,int,dirfd,const char *,pathname,
401 const struct timespec *,tsp,int,flags)
402 #else
403 static int sys_utimensat(int dirfd, const char *pathname,
404 const struct timespec times[2], int flags)
405 {
406 errno = ENOSYS;
407 return -1;
408 }
409 #endif
410 #endif /* TARGET_NR_utimensat */
411
412 #ifdef TARGET_NR_renameat2
413 #if defined(__NR_renameat2)
414 #define __NR_sys_renameat2 __NR_renameat2
415 _syscall5(int, sys_renameat2, int, oldfd, const char *, old, int, newfd,
416 const char *, new, unsigned int, flags)
417 #else
418 static int sys_renameat2(int oldfd, const char *old,
419 int newfd, const char *new, int flags)
420 {
421 if (flags == 0) {
422 return renameat(oldfd, old, newfd, new);
423 }
424 errno = ENOSYS;
425 return -1;
426 }
427 #endif
428 #endif /* TARGET_NR_renameat2 */
429
430 #ifdef CONFIG_INOTIFY
431 #include <sys/inotify.h>
432
433 #if defined(TARGET_NR_inotify_init) && defined(__NR_inotify_init)
434 static int sys_inotify_init(void)
435 {
436 return (inotify_init());
437 }
438 #endif
439 #if defined(TARGET_NR_inotify_add_watch) && defined(__NR_inotify_add_watch)
440 static int sys_inotify_add_watch(int fd,const char *pathname, int32_t mask)
441 {
442 return (inotify_add_watch(fd, pathname, mask));
443 }
444 #endif
445 #if defined(TARGET_NR_inotify_rm_watch) && defined(__NR_inotify_rm_watch)
446 static int sys_inotify_rm_watch(int fd, int32_t wd)
447 {
448 return (inotify_rm_watch(fd, wd));
449 }
450 #endif
451 #ifdef CONFIG_INOTIFY1
452 #if defined(TARGET_NR_inotify_init1) && defined(__NR_inotify_init1)
453 static int sys_inotify_init1(int flags)
454 {
455 return (inotify_init1(flags));
456 }
457 #endif
458 #endif
459 #else
460 /* Userspace can usually survive runtime without inotify */
461 #undef TARGET_NR_inotify_init
462 #undef TARGET_NR_inotify_init1
463 #undef TARGET_NR_inotify_add_watch
464 #undef TARGET_NR_inotify_rm_watch
465 #endif /* CONFIG_INOTIFY */
466
467 #if defined(TARGET_NR_prlimit64)
468 #ifndef __NR_prlimit64
469 # define __NR_prlimit64 -1
470 #endif
471 #define __NR_sys_prlimit64 __NR_prlimit64
472 /* The glibc rlimit structure may not be that used by the underlying syscall */
473 struct host_rlimit64 {
474 uint64_t rlim_cur;
475 uint64_t rlim_max;
476 };
477 _syscall4(int, sys_prlimit64, pid_t, pid, int, resource,
478 const struct host_rlimit64 *, new_limit,
479 struct host_rlimit64 *, old_limit)
480 #endif
481
482
483 #if defined(TARGET_NR_timer_create)
484 /* Maximum of 32 active POSIX timers allowed at any one time. */
485 static timer_t g_posix_timers[32] = { 0, } ;
486
487 static inline int next_free_host_timer(void)
488 {
489 int k ;
490 /* FIXME: Does finding the next free slot require a lock? */
491 for (k = 0; k < ARRAY_SIZE(g_posix_timers); k++) {
492 if (g_posix_timers[k] == 0) {
493 g_posix_timers[k] = (timer_t) 1;
494 return k;
495 }
496 }
497 return -1;
498 }
499 #endif
500
501 #define ERRNO_TABLE_SIZE 1200
502
503 /* target_to_host_errno_table[] is initialized from
504 * host_to_target_errno_table[] in syscall_init(). */
505 static uint16_t target_to_host_errno_table[ERRNO_TABLE_SIZE] = {
506 };
507
508 /*
509 * This list is the union of errno values overridden in asm-<arch>/errno.h
510 * minus the errnos that are not actually generic to all archs.
511 */
512 static uint16_t host_to_target_errno_table[ERRNO_TABLE_SIZE] = {
513 [EAGAIN] = TARGET_EAGAIN,
514 [EIDRM] = TARGET_EIDRM,
515 [ECHRNG] = TARGET_ECHRNG,
516 [EL2NSYNC] = TARGET_EL2NSYNC,
517 [EL3HLT] = TARGET_EL3HLT,
518 [EL3RST] = TARGET_EL3RST,
519 [ELNRNG] = TARGET_ELNRNG,
520 [EUNATCH] = TARGET_EUNATCH,
521 [ENOCSI] = TARGET_ENOCSI,
522 [EL2HLT] = TARGET_EL2HLT,
523 [EDEADLK] = TARGET_EDEADLK,
524 [ENOLCK] = TARGET_ENOLCK,
525 [EBADE] = TARGET_EBADE,
526 [EBADR] = TARGET_EBADR,
527 [EXFULL] = TARGET_EXFULL,
528 [ENOANO] = TARGET_ENOANO,
529 [EBADRQC] = TARGET_EBADRQC,
530 [EBADSLT] = TARGET_EBADSLT,
531 [EBFONT] = TARGET_EBFONT,
532 [ENOSTR] = TARGET_ENOSTR,
533 [ENODATA] = TARGET_ENODATA,
534 [ETIME] = TARGET_ETIME,
535 [ENOSR] = TARGET_ENOSR,
536 [ENONET] = TARGET_ENONET,
537 [ENOPKG] = TARGET_ENOPKG,
538 [EREMOTE] = TARGET_EREMOTE,
539 [ENOLINK] = TARGET_ENOLINK,
540 [EADV] = TARGET_EADV,
541 [ESRMNT] = TARGET_ESRMNT,
542 [ECOMM] = TARGET_ECOMM,
543 [EPROTO] = TARGET_EPROTO,
544 [EDOTDOT] = TARGET_EDOTDOT,
545 [EMULTIHOP] = TARGET_EMULTIHOP,
546 [EBADMSG] = TARGET_EBADMSG,
547 [ENAMETOOLONG] = TARGET_ENAMETOOLONG,
548 [EOVERFLOW] = TARGET_EOVERFLOW,
549 [ENOTUNIQ] = TARGET_ENOTUNIQ,
550 [EBADFD] = TARGET_EBADFD,
551 [EREMCHG] = TARGET_EREMCHG,
552 [ELIBACC] = TARGET_ELIBACC,
553 [ELIBBAD] = TARGET_ELIBBAD,
554 [ELIBSCN] = TARGET_ELIBSCN,
555 [ELIBMAX] = TARGET_ELIBMAX,
556 [ELIBEXEC] = TARGET_ELIBEXEC,
557 [EILSEQ] = TARGET_EILSEQ,
558 [ENOSYS] = TARGET_ENOSYS,
559 [ELOOP] = TARGET_ELOOP,
560 [ERESTART] = TARGET_ERESTART,
561 [ESTRPIPE] = TARGET_ESTRPIPE,
562 [ENOTEMPTY] = TARGET_ENOTEMPTY,
563 [EUSERS] = TARGET_EUSERS,
564 [ENOTSOCK] = TARGET_ENOTSOCK,
565 [EDESTADDRREQ] = TARGET_EDESTADDRREQ,
566 [EMSGSIZE] = TARGET_EMSGSIZE,
567 [EPROTOTYPE] = TARGET_EPROTOTYPE,
568 [ENOPROTOOPT] = TARGET_ENOPROTOOPT,
569 [EPROTONOSUPPORT] = TARGET_EPROTONOSUPPORT,
570 [ESOCKTNOSUPPORT] = TARGET_ESOCKTNOSUPPORT,
571 [EOPNOTSUPP] = TARGET_EOPNOTSUPP,
572 [EPFNOSUPPORT] = TARGET_EPFNOSUPPORT,
573 [EAFNOSUPPORT] = TARGET_EAFNOSUPPORT,
574 [EADDRINUSE] = TARGET_EADDRINUSE,
575 [EADDRNOTAVAIL] = TARGET_EADDRNOTAVAIL,
576 [ENETDOWN] = TARGET_ENETDOWN,
577 [ENETUNREACH] = TARGET_ENETUNREACH,
578 [ENETRESET] = TARGET_ENETRESET,
579 [ECONNABORTED] = TARGET_ECONNABORTED,
580 [ECONNRESET] = TARGET_ECONNRESET,
581 [ENOBUFS] = TARGET_ENOBUFS,
582 [EISCONN] = TARGET_EISCONN,
583 [ENOTCONN] = TARGET_ENOTCONN,
584 [EUCLEAN] = TARGET_EUCLEAN,
585 [ENOTNAM] = TARGET_ENOTNAM,
586 [ENAVAIL] = TARGET_ENAVAIL,
587 [EISNAM] = TARGET_EISNAM,
588 [EREMOTEIO] = TARGET_EREMOTEIO,
589 [EDQUOT] = TARGET_EDQUOT,
590 [ESHUTDOWN] = TARGET_ESHUTDOWN,
591 [ETOOMANYREFS] = TARGET_ETOOMANYREFS,
592 [ETIMEDOUT] = TARGET_ETIMEDOUT,
593 [ECONNREFUSED] = TARGET_ECONNREFUSED,
594 [EHOSTDOWN] = TARGET_EHOSTDOWN,
595 [EHOSTUNREACH] = TARGET_EHOSTUNREACH,
596 [EALREADY] = TARGET_EALREADY,
597 [EINPROGRESS] = TARGET_EINPROGRESS,
598 [ESTALE] = TARGET_ESTALE,
599 [ECANCELED] = TARGET_ECANCELED,
600 [ENOMEDIUM] = TARGET_ENOMEDIUM,
601 [EMEDIUMTYPE] = TARGET_EMEDIUMTYPE,
602 #ifdef ENOKEY
603 [ENOKEY] = TARGET_ENOKEY,
604 #endif
605 #ifdef EKEYEXPIRED
606 [EKEYEXPIRED] = TARGET_EKEYEXPIRED,
607 #endif
608 #ifdef EKEYREVOKED
609 [EKEYREVOKED] = TARGET_EKEYREVOKED,
610 #endif
611 #ifdef EKEYREJECTED
612 [EKEYREJECTED] = TARGET_EKEYREJECTED,
613 #endif
614 #ifdef EOWNERDEAD
615 [EOWNERDEAD] = TARGET_EOWNERDEAD,
616 #endif
617 #ifdef ENOTRECOVERABLE
618 [ENOTRECOVERABLE] = TARGET_ENOTRECOVERABLE,
619 #endif
620 #ifdef ENOMSG
621 [ENOMSG] = TARGET_ENOMSG,
622 #endif
623 #ifdef ERKFILL
624 [ERFKILL] = TARGET_ERFKILL,
625 #endif
626 #ifdef EHWPOISON
627 [EHWPOISON] = TARGET_EHWPOISON,
628 #endif
629 };
630
631 static inline int host_to_target_errno(int err)
632 {
633 if (err >= 0 && err < ERRNO_TABLE_SIZE &&
634 host_to_target_errno_table[err]) {
635 return host_to_target_errno_table[err];
636 }
637 return err;
638 }
639
640 static inline int target_to_host_errno(int err)
641 {
642 if (err >= 0 && err < ERRNO_TABLE_SIZE &&
643 target_to_host_errno_table[err]) {
644 return target_to_host_errno_table[err];
645 }
646 return err;
647 }
648
649 static inline abi_long get_errno(abi_long ret)
650 {
651 if (ret == -1)
652 return -host_to_target_errno(errno);
653 else
654 return ret;
655 }
656
657 const char *target_strerror(int err)
658 {
659 if (err == TARGET_ERESTARTSYS) {
660 return "To be restarted";
661 }
662 if (err == TARGET_QEMU_ESIGRETURN) {
663 return "Successful exit from sigreturn";
664 }
665
666 if ((err >= ERRNO_TABLE_SIZE) || (err < 0)) {
667 return NULL;
668 }
669 return strerror(target_to_host_errno(err));
670 }
671
672 #define safe_syscall0(type, name) \
673 static type safe_##name(void) \
674 { \
675 return safe_syscall(__NR_##name); \
676 }
677
678 #define safe_syscall1(type, name, type1, arg1) \
679 static type safe_##name(type1 arg1) \
680 { \
681 return safe_syscall(__NR_##name, arg1); \
682 }
683
684 #define safe_syscall2(type, name, type1, arg1, type2, arg2) \
685 static type safe_##name(type1 arg1, type2 arg2) \
686 { \
687 return safe_syscall(__NR_##name, arg1, arg2); \
688 }
689
690 #define safe_syscall3(type, name, type1, arg1, type2, arg2, type3, arg3) \
691 static type safe_##name(type1 arg1, type2 arg2, type3 arg3) \
692 { \
693 return safe_syscall(__NR_##name, arg1, arg2, arg3); \
694 }
695
696 #define safe_syscall4(type, name, type1, arg1, type2, arg2, type3, arg3, \
697 type4, arg4) \
698 static type safe_##name(type1 arg1, type2 arg2, type3 arg3, type4 arg4) \
699 { \
700 return safe_syscall(__NR_##name, arg1, arg2, arg3, arg4); \
701 }
702
703 #define safe_syscall5(type, name, type1, arg1, type2, arg2, type3, arg3, \
704 type4, arg4, type5, arg5) \
705 static type safe_##name(type1 arg1, type2 arg2, type3 arg3, type4 arg4, \
706 type5 arg5) \
707 { \
708 return safe_syscall(__NR_##name, arg1, arg2, arg3, arg4, arg5); \
709 }
710
711 #define safe_syscall6(type, name, type1, arg1, type2, arg2, type3, arg3, \
712 type4, arg4, type5, arg5, type6, arg6) \
713 static type safe_##name(type1 arg1, type2 arg2, type3 arg3, type4 arg4, \
714 type5 arg5, type6 arg6) \
715 { \
716 return safe_syscall(__NR_##name, arg1, arg2, arg3, arg4, arg5, arg6); \
717 }
718
719 safe_syscall3(ssize_t, read, int, fd, void *, buff, size_t, count)
720 safe_syscall3(ssize_t, write, int, fd, const void *, buff, size_t, count)
721 safe_syscall4(int, openat, int, dirfd, const char *, pathname, \
722 int, flags, mode_t, mode)
723 #if defined(TARGET_NR_wait4) || defined(TARGET_NR_waitpid)
724 safe_syscall4(pid_t, wait4, pid_t, pid, int *, status, int, options, \
725 struct rusage *, rusage)
726 #endif
727 safe_syscall5(int, waitid, idtype_t, idtype, id_t, id, siginfo_t *, infop, \
728 int, options, struct rusage *, rusage)
729 safe_syscall3(int, execve, const char *, filename, char **, argv, char **, envp)
730 #if defined(TARGET_NR_select) || defined(TARGET_NR__newselect) || \
731 defined(TARGET_NR_pselect6) || defined(TARGET_NR_pselect6_time64)
732 safe_syscall6(int, pselect6, int, nfds, fd_set *, readfds, fd_set *, writefds, \
733 fd_set *, exceptfds, struct timespec *, timeout, void *, sig)
734 #endif
735 #if defined(TARGET_NR_ppoll) || defined(TARGET_NR_ppoll_time64)
736 safe_syscall5(int, ppoll, struct pollfd *, ufds, unsigned int, nfds,
737 struct timespec *, tsp, const sigset_t *, sigmask,
738 size_t, sigsetsize)
739 #endif
740 safe_syscall6(int, epoll_pwait, int, epfd, struct epoll_event *, events,
741 int, maxevents, int, timeout, const sigset_t *, sigmask,
742 size_t, sigsetsize)
743 #if defined(__NR_futex)
744 safe_syscall6(int,futex,int *,uaddr,int,op,int,val, \
745 const struct timespec *,timeout,int *,uaddr2,int,val3)
746 #endif
747 #if defined(__NR_futex_time64)
748 safe_syscall6(int,futex_time64,int *,uaddr,int,op,int,val, \
749 const struct timespec *,timeout,int *,uaddr2,int,val3)
750 #endif
751 safe_syscall2(int, rt_sigsuspend, sigset_t *, newset, size_t, sigsetsize)
752 safe_syscall2(int, kill, pid_t, pid, int, sig)
753 safe_syscall2(int, tkill, int, tid, int, sig)
754 safe_syscall3(int, tgkill, int, tgid, int, pid, int, sig)
755 safe_syscall3(ssize_t, readv, int, fd, const struct iovec *, iov, int, iovcnt)
756 safe_syscall3(ssize_t, writev, int, fd, const struct iovec *, iov, int, iovcnt)
757 safe_syscall5(ssize_t, preadv, int, fd, const struct iovec *, iov, int, iovcnt,
758 unsigned long, pos_l, unsigned long, pos_h)
759 safe_syscall5(ssize_t, pwritev, int, fd, const struct iovec *, iov, int, iovcnt,
760 unsigned long, pos_l, unsigned long, pos_h)
761 safe_syscall3(int, connect, int, fd, const struct sockaddr *, addr,
762 socklen_t, addrlen)
763 safe_syscall6(ssize_t, sendto, int, fd, const void *, buf, size_t, len,
764 int, flags, const struct sockaddr *, addr, socklen_t, addrlen)
765 safe_syscall6(ssize_t, recvfrom, int, fd, void *, buf, size_t, len,
766 int, flags, struct sockaddr *, addr, socklen_t *, addrlen)
767 safe_syscall3(ssize_t, sendmsg, int, fd, const struct msghdr *, msg, int, flags)
768 safe_syscall3(ssize_t, recvmsg, int, fd, struct msghdr *, msg, int, flags)
769 safe_syscall2(int, flock, int, fd, int, operation)
770 #if defined(TARGET_NR_rt_sigtimedwait) || defined(TARGET_NR_rt_sigtimedwait_time64)
771 safe_syscall4(int, rt_sigtimedwait, const sigset_t *, these, siginfo_t *, uinfo,
772 const struct timespec *, uts, size_t, sigsetsize)
773 #endif
774 safe_syscall4(int, accept4, int, fd, struct sockaddr *, addr, socklen_t *, len,
775 int, flags)
776 #if defined(TARGET_NR_nanosleep)
777 safe_syscall2(int, nanosleep, const struct timespec *, req,
778 struct timespec *, rem)
779 #endif
780 #if defined(TARGET_NR_clock_nanosleep) || \
781 defined(TARGET_NR_clock_nanosleep_time64)
782 safe_syscall4(int, clock_nanosleep, const clockid_t, clock, int, flags,
783 const struct timespec *, req, struct timespec *, rem)
784 #endif
785 #ifdef __NR_ipc
786 #ifdef __s390x__
787 safe_syscall5(int, ipc, int, call, long, first, long, second, long, third,
788 void *, ptr)
789 #else
790 safe_syscall6(int, ipc, int, call, long, first, long, second, long, third,
791 void *, ptr, long, fifth)
792 #endif
793 #endif
794 #ifdef __NR_msgsnd
795 safe_syscall4(int, msgsnd, int, msgid, const void *, msgp, size_t, sz,
796 int, flags)
797 #endif
798 #ifdef __NR_msgrcv
799 safe_syscall5(int, msgrcv, int, msgid, void *, msgp, size_t, sz,
800 long, msgtype, int, flags)
801 #endif
802 #ifdef __NR_semtimedop
803 safe_syscall4(int, semtimedop, int, semid, struct sembuf *, tsops,
804 unsigned, nsops, const struct timespec *, timeout)
805 #endif
806 #if defined(TARGET_NR_mq_timedsend) || \
807 defined(TARGET_NR_mq_timedsend_time64)
808 safe_syscall5(int, mq_timedsend, int, mqdes, const char *, msg_ptr,
809 size_t, len, unsigned, prio, const struct timespec *, timeout)
810 #endif
811 #if defined(TARGET_NR_mq_timedreceive) || \
812 defined(TARGET_NR_mq_timedreceive_time64)
813 safe_syscall5(int, mq_timedreceive, int, mqdes, char *, msg_ptr,
814 size_t, len, unsigned *, prio, const struct timespec *, timeout)
815 #endif
816 /* We do ioctl like this rather than via safe_syscall3 to preserve the
817 * "third argument might be integer or pointer or not present" behaviour of
818 * the libc function.
819 */
820 #define safe_ioctl(...) safe_syscall(__NR_ioctl, __VA_ARGS__)
821 /* Similarly for fcntl. Note that callers must always:
822 * pass the F_GETLK64 etc constants rather than the unsuffixed F_GETLK
823 * use the flock64 struct rather than unsuffixed flock
824 * This will then work and use a 64-bit offset for both 32-bit and 64-bit hosts.
825 */
826 #ifdef __NR_fcntl64
827 #define safe_fcntl(...) safe_syscall(__NR_fcntl64, __VA_ARGS__)
828 #else
829 #define safe_fcntl(...) safe_syscall(__NR_fcntl, __VA_ARGS__)
830 #endif
831
832 static inline int host_to_target_sock_type(int host_type)
833 {
834 int target_type;
835
836 switch (host_type & 0xf /* SOCK_TYPE_MASK */) {
837 case SOCK_DGRAM:
838 target_type = TARGET_SOCK_DGRAM;
839 break;
840 case SOCK_STREAM:
841 target_type = TARGET_SOCK_STREAM;
842 break;
843 default:
844 target_type = host_type & 0xf /* SOCK_TYPE_MASK */;
845 break;
846 }
847
848 #if defined(SOCK_CLOEXEC)
849 if (host_type & SOCK_CLOEXEC) {
850 target_type |= TARGET_SOCK_CLOEXEC;
851 }
852 #endif
853
854 #if defined(SOCK_NONBLOCK)
855 if (host_type & SOCK_NONBLOCK) {
856 target_type |= TARGET_SOCK_NONBLOCK;
857 }
858 #endif
859
860 return target_type;
861 }
862
863 static abi_ulong target_brk;
864 static abi_ulong target_original_brk;
865 static abi_ulong brk_page;
866
867 void target_set_brk(abi_ulong new_brk)
868 {
869 target_original_brk = target_brk = HOST_PAGE_ALIGN(new_brk);
870 brk_page = HOST_PAGE_ALIGN(target_brk);
871 }
872
873 //#define DEBUGF_BRK(message, args...) do { fprintf(stderr, (message), ## args); } while (0)
874 #define DEBUGF_BRK(message, args...)
875
876 /* do_brk() must return target values and target errnos. */
877 abi_long do_brk(abi_ulong new_brk)
878 {
879 abi_long mapped_addr;
880 abi_ulong new_alloc_size;
881
882 DEBUGF_BRK("do_brk(" TARGET_ABI_FMT_lx ") -> ", new_brk);
883
884 if (!new_brk) {
885 DEBUGF_BRK(TARGET_ABI_FMT_lx " (!new_brk)\n", target_brk);
886 return target_brk;
887 }
888 if (new_brk < target_original_brk) {
889 DEBUGF_BRK(TARGET_ABI_FMT_lx " (new_brk < target_original_brk)\n",
890 target_brk);
891 return target_brk;
892 }
893
894 /* If the new brk is less than the highest page reserved to the
895 * target heap allocation, set it and we're almost done... */
896 if (new_brk <= brk_page) {
897 /* Heap contents are initialized to zero, as for anonymous
898 * mapped pages. */
899 if (new_brk > target_brk) {
900 memset(g2h(target_brk), 0, new_brk - target_brk);
901 }
902 target_brk = new_brk;
903 DEBUGF_BRK(TARGET_ABI_FMT_lx " (new_brk <= brk_page)\n", target_brk);
904 return target_brk;
905 }
906
907 /* We need to allocate more memory after the brk... Note that
908 * we don't use MAP_FIXED because that will map over the top of
909 * any existing mapping (like the one with the host libc or qemu
910 * itself); instead we treat "mapped but at wrong address" as
911 * a failure and unmap again.
912 */
913 new_alloc_size = HOST_PAGE_ALIGN(new_brk - brk_page);
914 mapped_addr = get_errno(target_mmap(brk_page, new_alloc_size,
915 PROT_READ|PROT_WRITE,
916 MAP_ANON|MAP_PRIVATE, 0, 0));
917
918 if (mapped_addr == brk_page) {
919 /* Heap contents are initialized to zero, as for anonymous
920 * mapped pages. Technically the new pages are already
921 * initialized to zero since they *are* anonymous mapped
922 * pages, however we have to take care with the contents that
923 * come from the remaining part of the previous page: it may
924 * contains garbage data due to a previous heap usage (grown
925 * then shrunken). */
926 memset(g2h(target_brk), 0, brk_page - target_brk);
927
928 target_brk = new_brk;
929 brk_page = HOST_PAGE_ALIGN(target_brk);
930 DEBUGF_BRK(TARGET_ABI_FMT_lx " (mapped_addr == brk_page)\n",
931 target_brk);
932 return target_brk;
933 } else if (mapped_addr != -1) {
934 /* Mapped but at wrong address, meaning there wasn't actually
935 * enough space for this brk.
936 */
937 target_munmap(mapped_addr, new_alloc_size);
938 mapped_addr = -1;
939 DEBUGF_BRK(TARGET_ABI_FMT_lx " (mapped_addr != -1)\n", target_brk);
940 }
941 else {
942 DEBUGF_BRK(TARGET_ABI_FMT_lx " (otherwise)\n", target_brk);
943 }
944
945 #if defined(TARGET_ALPHA)
946 /* We (partially) emulate OSF/1 on Alpha, which requires we
947 return a proper errno, not an unchanged brk value. */
948 return -TARGET_ENOMEM;
949 #endif
950 /* For everything else, return the previous break. */
951 return target_brk;
952 }
953
954 #if defined(TARGET_NR_select) || defined(TARGET_NR__newselect) || \
955 defined(TARGET_NR_pselect6) || defined(TARGET_NR_pselect6_time64)
956 static inline abi_long copy_from_user_fdset(fd_set *fds,
957 abi_ulong target_fds_addr,
958 int n)
959 {
960 int i, nw, j, k;
961 abi_ulong b, *target_fds;
962
963 nw = DIV_ROUND_UP(n, TARGET_ABI_BITS);
964 if (!(target_fds = lock_user(VERIFY_READ,
965 target_fds_addr,
966 sizeof(abi_ulong) * nw,
967 1)))
968 return -TARGET_EFAULT;
969
970 FD_ZERO(fds);
971 k = 0;
972 for (i = 0; i < nw; i++) {
973 /* grab the abi_ulong */
974 __get_user(b, &target_fds[i]);
975 for (j = 0; j < TARGET_ABI_BITS; j++) {
976 /* check the bit inside the abi_ulong */
977 if ((b >> j) & 1)
978 FD_SET(k, fds);
979 k++;
980 }
981 }
982
983 unlock_user(target_fds, target_fds_addr, 0);
984
985 return 0;
986 }
987
988 static inline abi_ulong copy_from_user_fdset_ptr(fd_set *fds, fd_set **fds_ptr,
989 abi_ulong target_fds_addr,
990 int n)
991 {
992 if (target_fds_addr) {
993 if (copy_from_user_fdset(fds, target_fds_addr, n))
994 return -TARGET_EFAULT;
995 *fds_ptr = fds;
996 } else {
997 *fds_ptr = NULL;
998 }
999 return 0;
1000 }
1001
1002 static inline abi_long copy_to_user_fdset(abi_ulong target_fds_addr,
1003 const fd_set *fds,
1004 int n)
1005 {
1006 int i, nw, j, k;
1007 abi_long v;
1008 abi_ulong *target_fds;
1009
1010 nw = DIV_ROUND_UP(n, TARGET_ABI_BITS);
1011 if (!(target_fds = lock_user(VERIFY_WRITE,
1012 target_fds_addr,
1013 sizeof(abi_ulong) * nw,
1014 0)))
1015 return -TARGET_EFAULT;
1016
1017 k = 0;
1018 for (i = 0; i < nw; i++) {
1019 v = 0;
1020 for (j = 0; j < TARGET_ABI_BITS; j++) {
1021 v |= ((abi_ulong)(FD_ISSET(k, fds) != 0) << j);
1022 k++;
1023 }
1024 __put_user(v, &target_fds[i]);
1025 }
1026
1027 unlock_user(target_fds, target_fds_addr, sizeof(abi_ulong) * nw);
1028
1029 return 0;
1030 }
1031 #endif
1032
1033 #if defined(__alpha__)
1034 #define HOST_HZ 1024
1035 #else
1036 #define HOST_HZ 100
1037 #endif
1038
1039 static inline abi_long host_to_target_clock_t(long ticks)
1040 {
1041 #if HOST_HZ == TARGET_HZ
1042 return ticks;
1043 #else
1044 return ((int64_t)ticks * TARGET_HZ) / HOST_HZ;
1045 #endif
1046 }
1047
1048 static inline abi_long host_to_target_rusage(abi_ulong target_addr,
1049 const struct rusage *rusage)
1050 {
1051 struct target_rusage *target_rusage;
1052
1053 if (!lock_user_struct(VERIFY_WRITE, target_rusage, target_addr, 0))
1054 return -TARGET_EFAULT;
1055 target_rusage->ru_utime.tv_sec = tswapal(rusage->ru_utime.tv_sec);
1056 target_rusage->ru_utime.tv_usec = tswapal(rusage->ru_utime.tv_usec);
1057 target_rusage->ru_stime.tv_sec = tswapal(rusage->ru_stime.tv_sec);
1058 target_rusage->ru_stime.tv_usec = tswapal(rusage->ru_stime.tv_usec);
1059 target_rusage->ru_maxrss = tswapal(rusage->ru_maxrss);
1060 target_rusage->ru_ixrss = tswapal(rusage->ru_ixrss);
1061 target_rusage->ru_idrss = tswapal(rusage->ru_idrss);
1062 target_rusage->ru_isrss = tswapal(rusage->ru_isrss);
1063 target_rusage->ru_minflt = tswapal(rusage->ru_minflt);
1064 target_rusage->ru_majflt = tswapal(rusage->ru_majflt);
1065 target_rusage->ru_nswap = tswapal(rusage->ru_nswap);
1066 target_rusage->ru_inblock = tswapal(rusage->ru_inblock);
1067 target_rusage->ru_oublock = tswapal(rusage->ru_oublock);
1068 target_rusage->ru_msgsnd = tswapal(rusage->ru_msgsnd);
1069 target_rusage->ru_msgrcv = tswapal(rusage->ru_msgrcv);
1070 target_rusage->ru_nsignals = tswapal(rusage->ru_nsignals);
1071 target_rusage->ru_nvcsw = tswapal(rusage->ru_nvcsw);
1072 target_rusage->ru_nivcsw = tswapal(rusage->ru_nivcsw);
1073 unlock_user_struct(target_rusage, target_addr, 1);
1074
1075 return 0;
1076 }
1077
1078 #ifdef TARGET_NR_setrlimit
1079 static inline rlim_t target_to_host_rlim(abi_ulong target_rlim)
1080 {
1081 abi_ulong target_rlim_swap;
1082 rlim_t result;
1083
1084 target_rlim_swap = tswapal(target_rlim);
1085 if (target_rlim_swap == TARGET_RLIM_INFINITY)
1086 return RLIM_INFINITY;
1087
1088 result = target_rlim_swap;
1089 if (target_rlim_swap != (rlim_t)result)
1090 return RLIM_INFINITY;
1091
1092 return result;
1093 }
1094 #endif
1095
1096 #if defined(TARGET_NR_getrlimit) || defined(TARGET_NR_ugetrlimit)
1097 static inline abi_ulong host_to_target_rlim(rlim_t rlim)
1098 {
1099 abi_ulong target_rlim_swap;
1100 abi_ulong result;
1101
1102 if (rlim == RLIM_INFINITY || rlim != (abi_long)rlim)
1103 target_rlim_swap = TARGET_RLIM_INFINITY;
1104 else
1105 target_rlim_swap = rlim;
1106 result = tswapal(target_rlim_swap);
1107
1108 return result;
1109 }
1110 #endif
1111
1112 static inline int target_to_host_resource(int code)
1113 {
1114 switch (code) {
1115 case TARGET_RLIMIT_AS:
1116 return RLIMIT_AS;
1117 case TARGET_RLIMIT_CORE:
1118 return RLIMIT_CORE;
1119 case TARGET_RLIMIT_CPU:
1120 return RLIMIT_CPU;
1121 case TARGET_RLIMIT_DATA:
1122 return RLIMIT_DATA;
1123 case TARGET_RLIMIT_FSIZE:
1124 return RLIMIT_FSIZE;
1125 case TARGET_RLIMIT_LOCKS:
1126 return RLIMIT_LOCKS;
1127 case TARGET_RLIMIT_MEMLOCK:
1128 return RLIMIT_MEMLOCK;
1129 case TARGET_RLIMIT_MSGQUEUE:
1130 return RLIMIT_MSGQUEUE;
1131 case TARGET_RLIMIT_NICE:
1132 return RLIMIT_NICE;
1133 case TARGET_RLIMIT_NOFILE:
1134 return RLIMIT_NOFILE;
1135 case TARGET_RLIMIT_NPROC:
1136 return RLIMIT_NPROC;
1137 case TARGET_RLIMIT_RSS:
1138 return RLIMIT_RSS;
1139 case TARGET_RLIMIT_RTPRIO:
1140 return RLIMIT_RTPRIO;
1141 case TARGET_RLIMIT_SIGPENDING:
1142 return RLIMIT_SIGPENDING;
1143 case TARGET_RLIMIT_STACK:
1144 return RLIMIT_STACK;
1145 default:
1146 return code;
1147 }
1148 }
1149
1150 static inline abi_long copy_from_user_timeval(struct timeval *tv,
1151 abi_ulong target_tv_addr)
1152 {
1153 struct target_timeval *target_tv;
1154
1155 if (!lock_user_struct(VERIFY_READ, target_tv, target_tv_addr, 1)) {
1156 return -TARGET_EFAULT;
1157 }
1158
1159 __get_user(tv->tv_sec, &target_tv->tv_sec);
1160 __get_user(tv->tv_usec, &target_tv->tv_usec);
1161
1162 unlock_user_struct(target_tv, target_tv_addr, 0);
1163
1164 return 0;
1165 }
1166
1167 static inline abi_long copy_to_user_timeval(abi_ulong target_tv_addr,
1168 const struct timeval *tv)
1169 {
1170 struct target_timeval *target_tv;
1171
1172 if (!lock_user_struct(VERIFY_WRITE, target_tv, target_tv_addr, 0)) {
1173 return -TARGET_EFAULT;
1174 }
1175
1176 __put_user(tv->tv_sec, &target_tv->tv_sec);
1177 __put_user(tv->tv_usec, &target_tv->tv_usec);
1178
1179 unlock_user_struct(target_tv, target_tv_addr, 1);
1180
1181 return 0;
1182 }
1183
1184 #if defined(TARGET_NR_clock_adjtime64) && defined(CONFIG_CLOCK_ADJTIME)
1185 static inline abi_long copy_from_user_timeval64(struct timeval *tv,
1186 abi_ulong target_tv_addr)
1187 {
1188 struct target__kernel_sock_timeval *target_tv;
1189
1190 if (!lock_user_struct(VERIFY_READ, target_tv, target_tv_addr, 1)) {
1191 return -TARGET_EFAULT;
1192 }
1193
1194 __get_user(tv->tv_sec, &target_tv->tv_sec);
1195 __get_user(tv->tv_usec, &target_tv->tv_usec);
1196
1197 unlock_user_struct(target_tv, target_tv_addr, 0);
1198
1199 return 0;
1200 }
1201 #endif
1202
1203 static inline abi_long copy_to_user_timeval64(abi_ulong target_tv_addr,
1204 const struct timeval *tv)
1205 {
1206 struct target__kernel_sock_timeval *target_tv;
1207
1208 if (!lock_user_struct(VERIFY_WRITE, target_tv, target_tv_addr, 0)) {
1209 return -TARGET_EFAULT;
1210 }
1211
1212 __put_user(tv->tv_sec, &target_tv->tv_sec);
1213 __put_user(tv->tv_usec, &target_tv->tv_usec);
1214
1215 unlock_user_struct(target_tv, target_tv_addr, 1);
1216
1217 return 0;
1218 }
1219
1220 #if defined(TARGET_NR_futex) || \
1221 defined(TARGET_NR_rt_sigtimedwait) || \
1222 defined(TARGET_NR_pselect6) || defined(TARGET_NR_pselect6) || \
1223 defined(TARGET_NR_nanosleep) || defined(TARGET_NR_clock_settime) || \
1224 defined(TARGET_NR_utimensat) || defined(TARGET_NR_mq_timedsend) || \
1225 defined(TARGET_NR_mq_timedreceive) || defined(TARGET_NR_ipc) || \
1226 defined(TARGET_NR_semop) || defined(TARGET_NR_semtimedop) || \
1227 defined(TARGET_NR_timer_settime) || \
1228 (defined(TARGET_NR_timerfd_settime) && defined(CONFIG_TIMERFD))
1229 static inline abi_long target_to_host_timespec(struct timespec *host_ts,
1230 abi_ulong target_addr)
1231 {
1232 struct target_timespec *target_ts;
1233
1234 if (!lock_user_struct(VERIFY_READ, target_ts, target_addr, 1)) {
1235 return -TARGET_EFAULT;
1236 }
1237 __get_user(host_ts->tv_sec, &target_ts->tv_sec);
1238 __get_user(host_ts->tv_nsec, &target_ts->tv_nsec);
1239 unlock_user_struct(target_ts, target_addr, 0);
1240 return 0;
1241 }
1242 #endif
1243
1244 #if defined(TARGET_NR_clock_settime64) || defined(TARGET_NR_futex_time64) || \
1245 defined(TARGET_NR_timer_settime64) || \
1246 defined(TARGET_NR_mq_timedsend_time64) || \
1247 defined(TARGET_NR_mq_timedreceive_time64) || \
1248 (defined(TARGET_NR_timerfd_settime64) && defined(CONFIG_TIMERFD)) || \
1249 defined(TARGET_NR_clock_nanosleep_time64) || \
1250 defined(TARGET_NR_rt_sigtimedwait_time64) || \
1251 defined(TARGET_NR_utimensat) || \
1252 defined(TARGET_NR_utimensat_time64) || \
1253 defined(TARGET_NR_semtimedop_time64) || \
1254 defined(TARGET_NR_pselect6_time64) || defined(TARGET_NR_ppoll_time64)
1255 static inline abi_long target_to_host_timespec64(struct timespec *host_ts,
1256 abi_ulong target_addr)
1257 {
1258 struct target__kernel_timespec *target_ts;
1259
1260 if (!lock_user_struct(VERIFY_READ, target_ts, target_addr, 1)) {
1261 return -TARGET_EFAULT;
1262 }
1263 __get_user(host_ts->tv_sec, &target_ts->tv_sec);
1264 __get_user(host_ts->tv_nsec, &target_ts->tv_nsec);
1265 /* in 32bit mode, this drops the padding */
1266 host_ts->tv_nsec = (long)(abi_long)host_ts->tv_nsec;
1267 unlock_user_struct(target_ts, target_addr, 0);
1268 return 0;
1269 }
1270 #endif
1271
1272 static inline abi_long host_to_target_timespec(abi_ulong target_addr,
1273 struct timespec *host_ts)
1274 {
1275 struct target_timespec *target_ts;
1276
1277 if (!lock_user_struct(VERIFY_WRITE, target_ts, target_addr, 0)) {
1278 return -TARGET_EFAULT;
1279 }
1280 __put_user(host_ts->tv_sec, &target_ts->tv_sec);
1281 __put_user(host_ts->tv_nsec, &target_ts->tv_nsec);
1282 unlock_user_struct(target_ts, target_addr, 1);
1283 return 0;
1284 }
1285
1286 static inline abi_long host_to_target_timespec64(abi_ulong target_addr,
1287 struct timespec *host_ts)
1288 {
1289 struct target__kernel_timespec *target_ts;
1290
1291 if (!lock_user_struct(VERIFY_WRITE, target_ts, target_addr, 0)) {
1292 return -TARGET_EFAULT;
1293 }
1294 __put_user(host_ts->tv_sec, &target_ts->tv_sec);
1295 __put_user(host_ts->tv_nsec, &target_ts->tv_nsec);
1296 unlock_user_struct(target_ts, target_addr, 1);
1297 return 0;
1298 }
1299
1300 #if defined(TARGET_NR_gettimeofday)
1301 static inline abi_long copy_to_user_timezone(abi_ulong target_tz_addr,
1302 struct timezone *tz)
1303 {
1304 struct target_timezone *target_tz;
1305
1306 if (!lock_user_struct(VERIFY_WRITE, target_tz, target_tz_addr, 1)) {
1307 return -TARGET_EFAULT;
1308 }
1309
1310 __put_user(tz->tz_minuteswest, &target_tz->tz_minuteswest);
1311 __put_user(tz->tz_dsttime, &target_tz->tz_dsttime);
1312
1313 unlock_user_struct(target_tz, target_tz_addr, 1);
1314
1315 return 0;
1316 }
1317 #endif
1318
1319 #if defined(TARGET_NR_settimeofday)
1320 static inline abi_long copy_from_user_timezone(struct timezone *tz,
1321 abi_ulong target_tz_addr)
1322 {
1323 struct target_timezone *target_tz;
1324
1325 if (!lock_user_struct(VERIFY_READ, target_tz, target_tz_addr, 1)) {
1326 return -TARGET_EFAULT;
1327 }
1328
1329 __get_user(tz->tz_minuteswest, &target_tz->tz_minuteswest);
1330 __get_user(tz->tz_dsttime, &target_tz->tz_dsttime);
1331
1332 unlock_user_struct(target_tz, target_tz_addr, 0);
1333
1334 return 0;
1335 }
1336 #endif
1337
1338 #if defined(TARGET_NR_mq_open) && defined(__NR_mq_open)
1339 #include <mqueue.h>
1340
1341 static inline abi_long copy_from_user_mq_attr(struct mq_attr *attr,
1342 abi_ulong target_mq_attr_addr)
1343 {
1344 struct target_mq_attr *target_mq_attr;
1345
1346 if (!lock_user_struct(VERIFY_READ, target_mq_attr,
1347 target_mq_attr_addr, 1))
1348 return -TARGET_EFAULT;
1349
1350 __get_user(attr->mq_flags, &target_mq_attr->mq_flags);
1351 __get_user(attr->mq_maxmsg, &target_mq_attr->mq_maxmsg);
1352 __get_user(attr->mq_msgsize, &target_mq_attr->mq_msgsize);
1353 __get_user(attr->mq_curmsgs, &target_mq_attr->mq_curmsgs);
1354
1355 unlock_user_struct(target_mq_attr, target_mq_attr_addr, 0);
1356
1357 return 0;
1358 }
1359
1360 static inline abi_long copy_to_user_mq_attr(abi_ulong target_mq_attr_addr,
1361 const struct mq_attr *attr)
1362 {
1363 struct target_mq_attr *target_mq_attr;
1364
1365 if (!lock_user_struct(VERIFY_WRITE, target_mq_attr,
1366 target_mq_attr_addr, 0))
1367 return -TARGET_EFAULT;
1368
1369 __put_user(attr->mq_flags, &target_mq_attr->mq_flags);
1370 __put_user(attr->mq_maxmsg, &target_mq_attr->mq_maxmsg);
1371 __put_user(attr->mq_msgsize, &target_mq_attr->mq_msgsize);
1372 __put_user(attr->mq_curmsgs, &target_mq_attr->mq_curmsgs);
1373
1374 unlock_user_struct(target_mq_attr, target_mq_attr_addr, 1);
1375
1376 return 0;
1377 }
1378 #endif
1379
1380 #if defined(TARGET_NR_select) || defined(TARGET_NR__newselect)
1381 /* do_select() must return target values and target errnos. */
1382 static abi_long do_select(int n,
1383 abi_ulong rfd_addr, abi_ulong wfd_addr,
1384 abi_ulong efd_addr, abi_ulong target_tv_addr)
1385 {
1386 fd_set rfds, wfds, efds;
1387 fd_set *rfds_ptr, *wfds_ptr, *efds_ptr;
1388 struct timeval tv;
1389 struct timespec ts, *ts_ptr;
1390 abi_long ret;
1391
1392 ret = copy_from_user_fdset_ptr(&rfds, &rfds_ptr, rfd_addr, n);
1393 if (ret) {
1394 return ret;
1395 }
1396 ret = copy_from_user_fdset_ptr(&wfds, &wfds_ptr, wfd_addr, n);
1397 if (ret) {
1398 return ret;
1399 }
1400 ret = copy_from_user_fdset_ptr(&efds, &efds_ptr, efd_addr, n);
1401 if (ret) {
1402 return ret;
1403 }
1404
1405 if (target_tv_addr) {
1406 if (copy_from_user_timeval(&tv, target_tv_addr))
1407 return -TARGET_EFAULT;
1408 ts.tv_sec = tv.tv_sec;
1409 ts.tv_nsec = tv.tv_usec * 1000;
1410 ts_ptr = &ts;
1411 } else {
1412 ts_ptr = NULL;
1413 }
1414
1415 ret = get_errno(safe_pselect6(n, rfds_ptr, wfds_ptr, efds_ptr,
1416 ts_ptr, NULL));
1417
1418 if (!is_error(ret)) {
1419 if (rfd_addr && copy_to_user_fdset(rfd_addr, &rfds, n))
1420 return -TARGET_EFAULT;
1421 if (wfd_addr && copy_to_user_fdset(wfd_addr, &wfds, n))
1422 return -TARGET_EFAULT;
1423 if (efd_addr && copy_to_user_fdset(efd_addr, &efds, n))
1424 return -TARGET_EFAULT;
1425
1426 if (target_tv_addr) {
1427 tv.tv_sec = ts.tv_sec;
1428 tv.tv_usec = ts.tv_nsec / 1000;
1429 if (copy_to_user_timeval(target_tv_addr, &tv)) {
1430 return -TARGET_EFAULT;
1431 }
1432 }
1433 }
1434
1435 return ret;
1436 }
1437
1438 #if defined(TARGET_WANT_OLD_SYS_SELECT)
1439 static abi_long do_old_select(abi_ulong arg1)
1440 {
1441 struct target_sel_arg_struct *sel;
1442 abi_ulong inp, outp, exp, tvp;
1443 long nsel;
1444
1445 if (!lock_user_struct(VERIFY_READ, sel, arg1, 1)) {
1446 return -TARGET_EFAULT;
1447 }
1448
1449 nsel = tswapal(sel->n);
1450 inp = tswapal(sel->inp);
1451 outp = tswapal(sel->outp);
1452 exp = tswapal(sel->exp);
1453 tvp = tswapal(sel->tvp);
1454
1455 unlock_user_struct(sel, arg1, 0);
1456
1457 return do_select(nsel, inp, outp, exp, tvp);
1458 }
1459 #endif
1460 #endif
1461
1462 #if defined(TARGET_NR_pselect6) || defined(TARGET_NR_pselect6_time64)
1463 static abi_long do_pselect6(abi_long arg1, abi_long arg2, abi_long arg3,
1464 abi_long arg4, abi_long arg5, abi_long arg6,
1465 bool time64)
1466 {
1467 abi_long rfd_addr, wfd_addr, efd_addr, n, ts_addr;
1468 fd_set rfds, wfds, efds;
1469 fd_set *rfds_ptr, *wfds_ptr, *efds_ptr;
1470 struct timespec ts, *ts_ptr;
1471 abi_long ret;
1472
1473 /*
1474 * The 6th arg is actually two args smashed together,
1475 * so we cannot use the C library.
1476 */
1477 sigset_t set;
1478 struct {
1479 sigset_t *set;
1480 size_t size;
1481 } sig, *sig_ptr;
1482
1483 abi_ulong arg_sigset, arg_sigsize, *arg7;
1484 target_sigset_t *target_sigset;
1485
1486 n = arg1;
1487 rfd_addr = arg2;
1488 wfd_addr = arg3;
1489 efd_addr = arg4;
1490 ts_addr = arg5;
1491
1492 ret = copy_from_user_fdset_ptr(&rfds, &rfds_ptr, rfd_addr, n);
1493 if (ret) {
1494 return ret;
1495 }
1496 ret = copy_from_user_fdset_ptr(&wfds, &wfds_ptr, wfd_addr, n);
1497 if (ret) {
1498 return ret;
1499 }
1500 ret = copy_from_user_fdset_ptr(&efds, &efds_ptr, efd_addr, n);
1501 if (ret) {
1502 return ret;
1503 }
1504
1505 /*
1506 * This takes a timespec, and not a timeval, so we cannot
1507 * use the do_select() helper ...
1508 */
1509 if (ts_addr) {
1510 if (time64) {
1511 if (target_to_host_timespec64(&ts, ts_addr)) {
1512 return -TARGET_EFAULT;
1513 }
1514 } else {
1515 if (target_to_host_timespec(&ts, ts_addr)) {
1516 return -TARGET_EFAULT;
1517 }
1518 }
1519 ts_ptr = &ts;
1520 } else {
1521 ts_ptr = NULL;
1522 }
1523
1524 /* Extract the two packed args for the sigset */
1525 if (arg6) {
1526 sig_ptr = &sig;
1527 sig.size = SIGSET_T_SIZE;
1528
1529 arg7 = lock_user(VERIFY_READ, arg6, sizeof(*arg7) * 2, 1);
1530 if (!arg7) {
1531 return -TARGET_EFAULT;
1532 }
1533 arg_sigset = tswapal(arg7[0]);
1534 arg_sigsize = tswapal(arg7[1]);
1535 unlock_user(arg7, arg6, 0);
1536
1537 if (arg_sigset) {
1538 sig.set = &set;
1539 if (arg_sigsize != sizeof(*target_sigset)) {
1540 /* Like the kernel, we enforce correct size sigsets */
1541 return -TARGET_EINVAL;
1542 }
1543 target_sigset = lock_user(VERIFY_READ, arg_sigset,
1544 sizeof(*target_sigset), 1);
1545 if (!target_sigset) {
1546 return -TARGET_EFAULT;
1547 }
1548 target_to_host_sigset(&set, target_sigset);
1549 unlock_user(target_sigset, arg_sigset, 0);
1550 } else {
1551 sig.set = NULL;
1552 }
1553 } else {
1554 sig_ptr = NULL;
1555 }
1556
1557 ret = get_errno(safe_pselect6(n, rfds_ptr, wfds_ptr, efds_ptr,
1558 ts_ptr, sig_ptr));
1559
1560 if (!is_error(ret)) {
1561 if (rfd_addr && copy_to_user_fdset(rfd_addr, &rfds, n)) {
1562 return -TARGET_EFAULT;
1563 }
1564 if (wfd_addr && copy_to_user_fdset(wfd_addr, &wfds, n)) {
1565 return -TARGET_EFAULT;
1566 }
1567 if (efd_addr && copy_to_user_fdset(efd_addr, &efds, n)) {
1568 return -TARGET_EFAULT;
1569 }
1570 if (time64) {
1571 if (ts_addr && host_to_target_timespec64(ts_addr, &ts)) {
1572 return -TARGET_EFAULT;
1573 }
1574 } else {
1575 if (ts_addr && host_to_target_timespec(ts_addr, &ts)) {
1576 return -TARGET_EFAULT;
1577 }
1578 }
1579 }
1580 return ret;
1581 }
1582 #endif
1583
1584 #if defined(TARGET_NR_poll) || defined(TARGET_NR_ppoll) || \
1585 defined(TARGET_NR_ppoll_time64)
1586 static abi_long do_ppoll(abi_long arg1, abi_long arg2, abi_long arg3,
1587 abi_long arg4, abi_long arg5, bool ppoll, bool time64)
1588 {
1589 struct target_pollfd *target_pfd;
1590 unsigned int nfds = arg2;
1591 struct pollfd *pfd;
1592 unsigned int i;
1593 abi_long ret;
1594
1595 pfd = NULL;
1596 target_pfd = NULL;
1597 if (nfds) {
1598 if (nfds > (INT_MAX / sizeof(struct target_pollfd))) {
1599 return -TARGET_EINVAL;
1600 }
1601 target_pfd = lock_user(VERIFY_WRITE, arg1,
1602 sizeof(struct target_pollfd) * nfds, 1);
1603 if (!target_pfd) {
1604 return -TARGET_EFAULT;
1605 }
1606
1607 pfd = alloca(sizeof(struct pollfd) * nfds);
1608 for (i = 0; i < nfds; i++) {
1609 pfd[i].fd = tswap32(target_pfd[i].fd);
1610 pfd[i].events = tswap16(target_pfd[i].events);
1611 }
1612 }
1613 if (ppoll) {
1614 struct timespec _timeout_ts, *timeout_ts = &_timeout_ts;
1615 target_sigset_t *target_set;
1616 sigset_t _set, *set = &_set;
1617
1618 if (arg3) {
1619 if (time64) {
1620 if (target_to_host_timespec64(timeout_ts, arg3)) {
1621 unlock_user(target_pfd, arg1, 0);
1622 return -TARGET_EFAULT;
1623 }
1624 } else {
1625 if (target_to_host_timespec(timeout_ts, arg3)) {
1626 unlock_user(target_pfd, arg1, 0);
1627 return -TARGET_EFAULT;
1628 }
1629 }
1630 } else {
1631 timeout_ts = NULL;
1632 }
1633
1634 if (arg4) {
1635 if (arg5 != sizeof(target_sigset_t)) {
1636 unlock_user(target_pfd, arg1, 0);
1637 return -TARGET_EINVAL;
1638 }
1639
1640 target_set = lock_user(VERIFY_READ, arg4,
1641 sizeof(target_sigset_t), 1);
1642 if (!target_set) {
1643 unlock_user(target_pfd, arg1, 0);
1644 return -TARGET_EFAULT;
1645 }
1646 target_to_host_sigset(set, target_set);
1647 } else {
1648 set = NULL;
1649 }
1650
1651 ret = get_errno(safe_ppoll(pfd, nfds, timeout_ts,
1652 set, SIGSET_T_SIZE));
1653
1654 if (!is_error(ret) && arg3) {
1655 if (time64) {
1656 if (host_to_target_timespec64(arg3, timeout_ts)) {
1657 return -TARGET_EFAULT;
1658 }
1659 } else {
1660 if (host_to_target_timespec(arg3, timeout_ts)) {
1661 return -TARGET_EFAULT;
1662 }
1663 }
1664 }
1665 if (arg4) {
1666 unlock_user(target_set, arg4, 0);
1667 }
1668 } else {
1669 struct timespec ts, *pts;
1670
1671 if (arg3 >= 0) {
1672 /* Convert ms to secs, ns */
1673 ts.tv_sec = arg3 / 1000;
1674 ts.tv_nsec = (arg3 % 1000) * 1000000LL;
1675 pts = &ts;
1676 } else {
1677 /* -ve poll() timeout means "infinite" */
1678 pts = NULL;
1679 }
1680 ret = get_errno(safe_ppoll(pfd, nfds, pts, NULL, 0));
1681 }
1682
1683 if (!is_error(ret)) {
1684 for (i = 0; i < nfds; i++) {
1685 target_pfd[i].revents = tswap16(pfd[i].revents);
1686 }
1687 }
1688 unlock_user(target_pfd, arg1, sizeof(struct target_pollfd) * nfds);
1689 return ret;
1690 }
1691 #endif
1692
1693 static abi_long do_pipe2(int host_pipe[], int flags)
1694 {
1695 #ifdef CONFIG_PIPE2
1696 return pipe2(host_pipe, flags);
1697 #else
1698 return -ENOSYS;
1699 #endif
1700 }
1701
1702 static abi_long do_pipe(void *cpu_env, abi_ulong pipedes,
1703 int flags, int is_pipe2)
1704 {
1705 int host_pipe[2];
1706 abi_long ret;
1707 ret = flags ? do_pipe2(host_pipe, flags) : pipe(host_pipe);
1708
1709 if (is_error(ret))
1710 return get_errno(ret);
1711
1712 /* Several targets have special calling conventions for the original
1713 pipe syscall, but didn't replicate this into the pipe2 syscall. */
1714 if (!is_pipe2) {
1715 #if defined(TARGET_ALPHA)
1716 ((CPUAlphaState *)cpu_env)->ir[IR_A4] = host_pipe[1];
1717 return host_pipe[0];
1718 #elif defined(TARGET_MIPS)
1719 ((CPUMIPSState*)cpu_env)->active_tc.gpr[3] = host_pipe[1];
1720 return host_pipe[0];
1721 #elif defined(TARGET_SH4)
1722 ((CPUSH4State*)cpu_env)->gregs[1] = host_pipe[1];
1723 return host_pipe[0];
1724 #elif defined(TARGET_SPARC)
1725 ((CPUSPARCState*)cpu_env)->regwptr[1] = host_pipe[1];
1726 return host_pipe[0];
1727 #endif
1728 }
1729
1730 if (put_user_s32(host_pipe[0], pipedes)
1731 || put_user_s32(host_pipe[1], pipedes + sizeof(host_pipe[0])))
1732 return -TARGET_EFAULT;
1733 return get_errno(ret);
1734 }
1735
1736 static inline abi_long target_to_host_ip_mreq(struct ip_mreqn *mreqn,
1737 abi_ulong target_addr,
1738 socklen_t len)
1739 {
1740 struct target_ip_mreqn *target_smreqn;
1741
1742 target_smreqn = lock_user(VERIFY_READ, target_addr, len, 1);
1743 if (!target_smreqn)
1744 return -TARGET_EFAULT;
1745 mreqn->imr_multiaddr.s_addr = target_smreqn->imr_multiaddr.s_addr;
1746 mreqn->imr_address.s_addr = target_smreqn->imr_address.s_addr;
1747 if (len == sizeof(struct target_ip_mreqn))
1748 mreqn->imr_ifindex = tswapal(target_smreqn->imr_ifindex);
1749 unlock_user(target_smreqn, target_addr, 0);
1750
1751 return 0;
1752 }
1753
1754 static inline abi_long target_to_host_sockaddr(int fd, struct sockaddr *addr,
1755 abi_ulong target_addr,
1756 socklen_t len)
1757 {
1758 const socklen_t unix_maxlen = sizeof (struct sockaddr_un);
1759 sa_family_t sa_family;
1760 struct target_sockaddr *target_saddr;
1761
1762 if (fd_trans_target_to_host_addr(fd)) {
1763 return fd_trans_target_to_host_addr(fd)(addr, target_addr, len);
1764 }
1765
1766 target_saddr = lock_user(VERIFY_READ, target_addr, len, 1);
1767 if (!target_saddr)
1768 return -TARGET_EFAULT;
1769
1770 sa_family = tswap16(target_saddr->sa_family);
1771
1772 /* Oops. The caller might send a incomplete sun_path; sun_path
1773 * must be terminated by \0 (see the manual page), but
1774 * unfortunately it is quite common to specify sockaddr_un
1775 * length as "strlen(x->sun_path)" while it should be
1776 * "strlen(...) + 1". We'll fix that here if needed.
1777 * Linux kernel has a similar feature.
1778 */
1779
1780 if (sa_family == AF_UNIX) {
1781 if (len < unix_maxlen && len > 0) {
1782 char *cp = (char*)target_saddr;
1783
1784 if ( cp[len-1] && !cp[len] )
1785 len++;
1786 }
1787 if (len > unix_maxlen)
1788 len = unix_maxlen;
1789 }
1790
1791 memcpy(addr, target_saddr, len);
1792 addr->sa_family = sa_family;
1793 if (sa_family == AF_NETLINK) {
1794 struct sockaddr_nl *nladdr;
1795
1796 nladdr = (struct sockaddr_nl *)addr;
1797 nladdr->nl_pid = tswap32(nladdr->nl_pid);
1798 nladdr->nl_groups = tswap32(nladdr->nl_groups);
1799 } else if (sa_family == AF_PACKET) {
1800 struct target_sockaddr_ll *lladdr;
1801
1802 lladdr = (struct target_sockaddr_ll *)addr;
1803 lladdr->sll_ifindex = tswap32(lladdr->sll_ifindex);
1804 lladdr->sll_hatype = tswap16(lladdr->sll_hatype);
1805 }
1806 unlock_user(target_saddr, target_addr, 0);
1807
1808 return 0;
1809 }
1810
1811 static inline abi_long host_to_target_sockaddr(abi_ulong target_addr,
1812 struct sockaddr *addr,
1813 socklen_t len)
1814 {
1815 struct target_sockaddr *target_saddr;
1816
1817 if (len == 0) {
1818 return 0;
1819 }
1820 assert(addr);
1821
1822 target_saddr = lock_user(VERIFY_WRITE, target_addr, len, 0);
1823 if (!target_saddr)
1824 return -TARGET_EFAULT;
1825 memcpy(target_saddr, addr, len);
1826 if (len >= offsetof(struct target_sockaddr, sa_family) +
1827 sizeof(target_saddr->sa_family)) {
1828 target_saddr->sa_family = tswap16(addr->sa_family);
1829 }
1830 if (addr->sa_family == AF_NETLINK &&
1831 len >= sizeof(struct target_sockaddr_nl)) {
1832 struct target_sockaddr_nl *target_nl =
1833 (struct target_sockaddr_nl *)target_saddr;
1834 target_nl->nl_pid = tswap32(target_nl->nl_pid);
1835 target_nl->nl_groups = tswap32(target_nl->nl_groups);
1836 } else if (addr->sa_family == AF_PACKET) {
1837 struct sockaddr_ll *target_ll = (struct sockaddr_ll *)target_saddr;
1838 target_ll->sll_ifindex = tswap32(target_ll->sll_ifindex);
1839 target_ll->sll_hatype = tswap16(target_ll->sll_hatype);
1840 } else if (addr->sa_family == AF_INET6 &&
1841 len >= sizeof(struct target_sockaddr_in6)) {
1842 struct target_sockaddr_in6 *target_in6 =
1843 (struct target_sockaddr_in6 *)target_saddr;
1844 target_in6->sin6_scope_id = tswap16(target_in6->sin6_scope_id);
1845 }
1846 unlock_user(target_saddr, target_addr, len);
1847
1848 return 0;
1849 }
1850
1851 static inline abi_long target_to_host_cmsg(struct msghdr *msgh,
1852 struct target_msghdr *target_msgh)
1853 {
1854 struct cmsghdr *cmsg = CMSG_FIRSTHDR(msgh);
1855 abi_long msg_controllen;
1856 abi_ulong target_cmsg_addr;
1857 struct target_cmsghdr *target_cmsg, *target_cmsg_start;
1858 socklen_t space = 0;
1859
1860 msg_controllen = tswapal(target_msgh->msg_controllen);
1861 if (msg_controllen < sizeof (struct target_cmsghdr))
1862 goto the_end;
1863 target_cmsg_addr = tswapal(target_msgh->msg_control);
1864 target_cmsg = lock_user(VERIFY_READ, target_cmsg_addr, msg_controllen, 1);
1865 target_cmsg_start = target_cmsg;
1866 if (!target_cmsg)
1867 return -TARGET_EFAULT;
1868
1869 while (cmsg && target_cmsg) {
1870 void *data = CMSG_DATA(cmsg);
1871 void *target_data = TARGET_CMSG_DATA(target_cmsg);
1872
1873 int len = tswapal(target_cmsg->cmsg_len)
1874 - sizeof(struct target_cmsghdr);
1875
1876 space += CMSG_SPACE(len);
1877 if (space > msgh->msg_controllen) {
1878 space -= CMSG_SPACE(len);
1879 /* This is a QEMU bug, since we allocated the payload
1880 * area ourselves (unlike overflow in host-to-target
1881 * conversion, which is just the guest giving us a buffer
1882 * that's too small). It can't happen for the payload types
1883 * we currently support; if it becomes an issue in future
1884 * we would need to improve our allocation strategy to
1885 * something more intelligent than "twice the size of the
1886 * target buffer we're reading from".
1887 */
1888 qemu_log_mask(LOG_UNIMP,
1889 ("Unsupported ancillary data %d/%d: "
1890 "unhandled msg size\n"),
1891 tswap32(target_cmsg->cmsg_level),
1892 tswap32(target_cmsg->cmsg_type));
1893 break;
1894 }
1895
1896 if (tswap32(target_cmsg->cmsg_level) == TARGET_SOL_SOCKET) {
1897 cmsg->cmsg_level = SOL_SOCKET;
1898 } else {
1899 cmsg->cmsg_level = tswap32(target_cmsg->cmsg_level);
1900 }
1901 cmsg->cmsg_type = tswap32(target_cmsg->cmsg_type);
1902 cmsg->cmsg_len = CMSG_LEN(len);
1903
1904 if (cmsg->cmsg_level == SOL_SOCKET && cmsg->cmsg_type == SCM_RIGHTS) {
1905 int *fd = (int *)data;
1906 int *target_fd = (int *)target_data;
1907 int i, numfds = len / sizeof(int);
1908
1909 for (i = 0; i < numfds; i++) {
1910 __get_user(fd[i], target_fd + i);
1911 }
1912 } else if (cmsg->cmsg_level == SOL_SOCKET
1913 && cmsg->cmsg_type == SCM_CREDENTIALS) {
1914 struct ucred *cred = (struct ucred *)data;
1915 struct target_ucred *target_cred =
1916 (struct target_ucred *)target_data;
1917
1918 __get_user(cred->pid, &target_cred->pid);
1919 __get_user(cred->uid, &target_cred->uid);
1920 __get_user(cred->gid, &target_cred->gid);
1921 } else {
1922 qemu_log_mask(LOG_UNIMP, "Unsupported ancillary data: %d/%d\n",
1923 cmsg->cmsg_level, cmsg->cmsg_type);
1924 memcpy(data, target_data, len);
1925 }
1926
1927 cmsg = CMSG_NXTHDR(msgh, cmsg);
1928 target_cmsg = TARGET_CMSG_NXTHDR(target_msgh, target_cmsg,
1929 target_cmsg_start);
1930 }
1931 unlock_user(target_cmsg, target_cmsg_addr, 0);
1932 the_end:
1933 msgh->msg_controllen = space;
1934 return 0;
1935 }
1936
1937 static inline abi_long host_to_target_cmsg(struct target_msghdr *target_msgh,
1938 struct msghdr *msgh)
1939 {
1940 struct cmsghdr *cmsg = CMSG_FIRSTHDR(msgh);
1941 abi_long msg_controllen;
1942 abi_ulong target_cmsg_addr;
1943 struct target_cmsghdr *target_cmsg, *target_cmsg_start;
1944 socklen_t space = 0;
1945
1946 msg_controllen = tswapal(target_msgh->msg_controllen);
1947 if (msg_controllen < sizeof (struct target_cmsghdr))
1948 goto the_end;
1949 target_cmsg_addr = tswapal(target_msgh->msg_control);
1950 target_cmsg = lock_user(VERIFY_WRITE, target_cmsg_addr, msg_controllen, 0);
1951 target_cmsg_start = target_cmsg;
1952 if (!target_cmsg)
1953 return -TARGET_EFAULT;
1954
1955 while (cmsg && target_cmsg) {
1956 void *data = CMSG_DATA(cmsg);
1957 void *target_data = TARGET_CMSG_DATA(target_cmsg);
1958
1959 int len = cmsg->cmsg_len - sizeof(struct cmsghdr);
1960 int tgt_len, tgt_space;
1961
1962 /* We never copy a half-header but may copy half-data;
1963 * this is Linux's behaviour in put_cmsg(). Note that
1964 * truncation here is a guest problem (which we report
1965 * to the guest via the CTRUNC bit), unlike truncation
1966 * in target_to_host_cmsg, which is a QEMU bug.
1967 */
1968 if (msg_controllen < sizeof(struct target_cmsghdr)) {
1969 target_msgh->msg_flags |= tswap32(MSG_CTRUNC);
1970 break;
1971 }
1972
1973 if (cmsg->cmsg_level == SOL_SOCKET) {
1974 target_cmsg->cmsg_level = tswap32(TARGET_SOL_SOCKET);
1975 } else {
1976 target_cmsg->cmsg_level = tswap32(cmsg->cmsg_level);
1977 }
1978 target_cmsg->cmsg_type = tswap32(cmsg->cmsg_type);
1979
1980 /* Payload types which need a different size of payload on
1981 * the target must adjust tgt_len here.
1982 */
1983 tgt_len = len;
1984 switch (cmsg->cmsg_level) {
1985 case SOL_SOCKET:
1986 switch (cmsg->cmsg_type) {
1987 case SO_TIMESTAMP:
1988 tgt_len = sizeof(struct target_timeval);
1989 break;
1990 default:
1991 break;
1992 }
1993 break;
1994 default:
1995 break;
1996 }
1997
1998 if (msg_controllen < TARGET_CMSG_LEN(tgt_len)) {
1999 target_msgh->msg_flags |= tswap32(MSG_CTRUNC);
2000 tgt_len = msg_controllen - sizeof(struct target_cmsghdr);
2001 }
2002
2003 /* We must now copy-and-convert len bytes of payload
2004 * into tgt_len bytes of destination space. Bear in mind
2005 * that in both source and destination we may be dealing
2006 * with a truncated value!
2007 */
2008 switch (cmsg->cmsg_level) {
2009 case SOL_SOCKET:
2010 switch (cmsg->cmsg_type) {
2011 case SCM_RIGHTS:
2012 {
2013 int *fd = (int *)data;
2014 int *target_fd = (int *)target_data;
2015 int i, numfds = tgt_len / sizeof(int);
2016
2017 for (i = 0; i < numfds; i++) {
2018 __put_user(fd[i], target_fd + i);
2019 }
2020 break;
2021 }
2022 case SO_TIMESTAMP:
2023 {
2024 struct timeval *tv = (struct timeval *)data;
2025 struct target_timeval *target_tv =
2026 (struct target_timeval *)target_data;
2027
2028 if (len != sizeof(struct timeval) ||
2029 tgt_len != sizeof(struct target_timeval)) {
2030 goto unimplemented;
2031 }
2032
2033 /* copy struct timeval to target */
2034 __put_user(tv->tv_sec, &target_tv->tv_sec);
2035 __put_user(tv->tv_usec, &target_tv->tv_usec);
2036 break;
2037 }
2038 case SCM_CREDENTIALS:
2039 {
2040 struct ucred *cred = (struct ucred *)data;
2041 struct target_ucred *target_cred =
2042 (struct target_ucred *)target_data;
2043
2044 __put_user(cred->pid, &target_cred->pid);
2045 __put_user(cred->uid, &target_cred->uid);
2046 __put_user(cred->gid, &target_cred->gid);
2047 break;
2048 }
2049 default:
2050 goto unimplemented;
2051 }
2052 break;
2053
2054 case SOL_IP:
2055 switch (cmsg->cmsg_type) {
2056 case IP_TTL:
2057 {
2058 uint32_t *v = (uint32_t *)data;
2059 uint32_t *t_int = (uint32_t *)target_data;
2060
2061 if (len != sizeof(uint32_t) ||
2062 tgt_len != sizeof(uint32_t)) {
2063 goto unimplemented;
2064 }
2065 __put_user(*v, t_int);
2066 break;
2067 }
2068 case IP_RECVERR:
2069 {
2070 struct errhdr_t {
2071 struct sock_extended_err ee;
2072 struct sockaddr_in offender;
2073 };
2074 struct errhdr_t *errh = (struct errhdr_t *)data;
2075 struct errhdr_t *target_errh =
2076 (struct errhdr_t *)target_data;
2077
2078 if (len != sizeof(struct errhdr_t) ||
2079 tgt_len != sizeof(struct errhdr_t)) {
2080 goto unimplemented;
2081 }
2082 __put_user(errh->ee.ee_errno, &target_errh->ee.ee_errno);
2083 __put_user(errh->ee.ee_origin, &target_errh->ee.ee_origin);
2084 __put_user(errh->ee.ee_type, &target_errh->ee.ee_type);
2085 __put_user(errh->ee.ee_code, &target_errh->ee.ee_code);
2086 __put_user(errh->ee.ee_pad, &target_errh->ee.ee_pad);
2087 __put_user(errh->ee.ee_info, &target_errh->ee.ee_info);
2088 __put_user(errh->ee.ee_data, &target_errh->ee.ee_data);
2089 host_to_target_sockaddr((unsigned long) &target_errh->offender,
2090 (void *) &errh->offender, sizeof(errh->offender));
2091 break;
2092 }
2093 default:
2094 goto unimplemented;
2095 }
2096 break;
2097
2098 case SOL_IPV6:
2099 switch (cmsg->cmsg_type) {
2100 case IPV6_HOPLIMIT:
2101 {
2102 uint32_t *v = (uint32_t *)data;
2103 uint32_t *t_int = (uint32_t *)target_data;
2104
2105 if (len != sizeof(uint32_t) ||
2106 tgt_len != sizeof(uint32_t)) {
2107 goto unimplemented;
2108 }
2109 __put_user(*v, t_int);
2110 break;
2111 }
2112 case IPV6_RECVERR:
2113 {
2114 struct errhdr6_t {
2115 struct sock_extended_err ee;
2116 struct sockaddr_in6 offender;
2117 };
2118 struct errhdr6_t *errh = (struct errhdr6_t *)data;
2119 struct errhdr6_t *target_errh =
2120 (struct errhdr6_t *)target_data;
2121
2122 if (len != sizeof(struct errhdr6_t) ||
2123 tgt_len != sizeof(struct errhdr6_t)) {
2124 goto unimplemented;
2125 }
2126 __put_user(errh->ee.ee_errno, &target_errh->ee.ee_errno);
2127 __put_user(errh->ee.ee_origin, &target_errh->ee.ee_origin);
2128 __put_user(errh->ee.ee_type, &target_errh->ee.ee_type);
2129 __put_user(errh->ee.ee_code, &target_errh->ee.ee_code);
2130 __put_user(errh->ee.ee_pad, &target_errh->ee.ee_pad);
2131 __put_user(errh->ee.ee_info, &target_errh->ee.ee_info);
2132 __put_user(errh->ee.ee_data, &target_errh->ee.ee_data);
2133 host_to_target_sockaddr((unsigned long) &target_errh->offender,
2134 (void *) &errh->offender, sizeof(errh->offender));
2135 break;
2136 }
2137 default:
2138 goto unimplemented;
2139 }
2140 break;
2141
2142 default:
2143 unimplemented:
2144 qemu_log_mask(LOG_UNIMP, "Unsupported ancillary data: %d/%d\n",
2145 cmsg->cmsg_level, cmsg->cmsg_type);
2146 memcpy(target_data, data, MIN(len, tgt_len));
2147 if (tgt_len > len) {
2148 memset(target_data + len, 0, tgt_len - len);
2149 }
2150 }
2151
2152 target_cmsg->cmsg_len = tswapal(TARGET_CMSG_LEN(tgt_len));
2153 tgt_space = TARGET_CMSG_SPACE(tgt_len);
2154 if (msg_controllen < tgt_space) {
2155 tgt_space = msg_controllen;
2156 }
2157 msg_controllen -= tgt_space;
2158 space += tgt_space;
2159 cmsg = CMSG_NXTHDR(msgh, cmsg);
2160 target_cmsg = TARGET_CMSG_NXTHDR(target_msgh, target_cmsg,
2161 target_cmsg_start);
2162 }
2163 unlock_user(target_cmsg, target_cmsg_addr, space);
2164 the_end:
2165 target_msgh->msg_controllen = tswapal(space);
2166 return 0;
2167 }
2168
2169 /* do_setsockopt() Must return target values and target errnos. */
2170 static abi_long do_setsockopt(int sockfd, int level, int optname,
2171 abi_ulong optval_addr, socklen_t optlen)
2172 {
2173 abi_long ret;
2174 int val;
2175 struct ip_mreqn *ip_mreq;
2176 struct ip_mreq_source *ip_mreq_source;
2177
2178 switch(level) {
2179 case SOL_TCP:
2180 /* TCP options all take an 'int' value. */
2181 if (optlen < sizeof(uint32_t))
2182 return -TARGET_EINVAL;
2183
2184 if (get_user_u32(val, optval_addr))
2185 return -TARGET_EFAULT;
2186 ret = get_errno(setsockopt(sockfd, level, optname, &val, sizeof(val)));
2187 break;
2188 case SOL_IP:
2189 switch(optname) {
2190 case IP_TOS:
2191 case IP_TTL:
2192 case IP_HDRINCL:
2193 case IP_ROUTER_ALERT:
2194 case IP_RECVOPTS:
2195 case IP_RETOPTS:
2196 case IP_PKTINFO:
2197 case IP_MTU_DISCOVER:
2198 case IP_RECVERR:
2199 case IP_RECVTTL:
2200 case IP_RECVTOS:
2201 #ifdef IP_FREEBIND
2202 case IP_FREEBIND:
2203 #endif
2204 case IP_MULTICAST_TTL:
2205 case IP_MULTICAST_LOOP:
2206 val = 0;
2207 if (optlen >= sizeof(uint32_t)) {
2208 if (get_user_u32(val, optval_addr))
2209 return -TARGET_EFAULT;
2210 } else if (optlen >= 1) {
2211 if (get_user_u8(val, optval_addr))
2212 return -TARGET_EFAULT;
2213 }
2214 ret = get_errno(setsockopt(sockfd, level, optname, &val, sizeof(val)));
2215 break;
2216 case IP_ADD_MEMBERSHIP:
2217 case IP_DROP_MEMBERSHIP:
2218 if (optlen < sizeof (struct target_ip_mreq) ||
2219 optlen > sizeof (struct target_ip_mreqn))
2220 return -TARGET_EINVAL;
2221
2222 ip_mreq = (struct ip_mreqn *) alloca(optlen);
2223 target_to_host_ip_mreq(ip_mreq, optval_addr, optlen);
2224 ret = get_errno(setsockopt(sockfd, level, optname, ip_mreq, optlen));
2225 break;
2226
2227 case IP_BLOCK_SOURCE:
2228 case IP_UNBLOCK_SOURCE:
2229 case IP_ADD_SOURCE_MEMBERSHIP:
2230 case IP_DROP_SOURCE_MEMBERSHIP:
2231 if (optlen != sizeof (struct target_ip_mreq_source))
2232 return -TARGET_EINVAL;
2233
2234 ip_mreq_source = lock_user(VERIFY_READ, optval_addr, optlen, 1);
2235 ret = get_errno(setsockopt(sockfd, level, optname, ip_mreq_source, optlen));
2236 unlock_user (ip_mreq_source, optval_addr, 0);
2237 break;
2238
2239 default:
2240 goto unimplemented;
2241 }
2242 break;
2243 case SOL_IPV6:
2244 switch (optname) {
2245 case IPV6_MTU_DISCOVER:
2246 case IPV6_MTU:
2247 case IPV6_V6ONLY:
2248 case IPV6_RECVPKTINFO:
2249 case IPV6_UNICAST_HOPS:
2250 case IPV6_MULTICAST_HOPS:
2251 case IPV6_MULTICAST_LOOP:
2252 case IPV6_RECVERR:
2253 case IPV6_RECVHOPLIMIT:
2254 case IPV6_2292HOPLIMIT:
2255 case IPV6_CHECKSUM:
2256 case IPV6_ADDRFORM:
2257 case IPV6_2292PKTINFO:
2258 case IPV6_RECVTCLASS:
2259 case IPV6_RECVRTHDR:
2260 case IPV6_2292RTHDR:
2261 case IPV6_RECVHOPOPTS:
2262 case IPV6_2292HOPOPTS:
2263 case IPV6_RECVDSTOPTS:
2264 case IPV6_2292DSTOPTS:
2265 case IPV6_TCLASS:
2266 #ifdef IPV6_RECVPATHMTU
2267 case IPV6_RECVPATHMTU:
2268 #endif
2269 #ifdef IPV6_TRANSPARENT
2270 case IPV6_TRANSPARENT:
2271 #endif
2272 #ifdef IPV6_FREEBIND
2273 case IPV6_FREEBIND:
2274 #endif
2275 #ifdef IPV6_RECVORIGDSTADDR
2276 case IPV6_RECVORIGDSTADDR:
2277 #endif
2278 val = 0;
2279 if (optlen < sizeof(uint32_t)) {
2280 return -TARGET_EINVAL;
2281 }
2282 if (get_user_u32(val, optval_addr)) {
2283 return -TARGET_EFAULT;
2284 }
2285 ret = get_errno(setsockopt(sockfd, level, optname,
2286 &val, sizeof(val)));
2287 break;
2288 case IPV6_PKTINFO:
2289 {
2290 struct in6_pktinfo pki;
2291
2292 if (optlen < sizeof(pki)) {
2293 return -TARGET_EINVAL;
2294 }
2295
2296 if (copy_from_user(&pki, optval_addr, sizeof(pki))) {
2297 return -TARGET_EFAULT;
2298 }
2299
2300 pki.ipi6_ifindex = tswap32(pki.ipi6_ifindex);
2301
2302 ret = get_errno(setsockopt(sockfd, level, optname,
2303 &pki, sizeof(pki)));
2304 break;
2305 }
2306 case IPV6_ADD_MEMBERSHIP:
2307 case IPV6_DROP_MEMBERSHIP:
2308 {
2309 struct ipv6_mreq ipv6mreq;
2310
2311 if (optlen < sizeof(ipv6mreq)) {
2312 return -TARGET_EINVAL;
2313 }
2314
2315 if (copy_from_user(&ipv6mreq, optval_addr, sizeof(ipv6mreq))) {
2316 return -TARGET_EFAULT;
2317 }
2318
2319 ipv6mreq.ipv6mr_interface = tswap32(ipv6mreq.ipv6mr_interface);
2320
2321 ret = get_errno(setsockopt(sockfd, level, optname,
2322 &ipv6mreq, sizeof(ipv6mreq)));
2323 break;
2324 }
2325 default:
2326 goto unimplemented;
2327 }
2328 break;
2329 case SOL_ICMPV6:
2330 switch (optname) {
2331 case ICMPV6_FILTER:
2332 {
2333 struct icmp6_filter icmp6f;
2334
2335 if (optlen > sizeof(icmp6f)) {
2336 optlen = sizeof(icmp6f);
2337 }
2338
2339 if (copy_from_user(&icmp6f, optval_addr, optlen)) {
2340 return -TARGET_EFAULT;
2341 }
2342
2343 for (val = 0; val < 8; val++) {
2344 icmp6f.data[val] = tswap32(icmp6f.data[val]);
2345 }
2346
2347 ret = get_errno(setsockopt(sockfd, level, optname,
2348 &icmp6f, optlen));
2349 break;
2350 }
2351 default:
2352 goto unimplemented;
2353 }
2354 break;
2355 case SOL_RAW:
2356 switch (optname) {
2357 case ICMP_FILTER:
2358 case IPV6_CHECKSUM:
2359 /* those take an u32 value */
2360 if (optlen < sizeof(uint32_t)) {
2361 return -TARGET_EINVAL;
2362 }
2363
2364 if (get_user_u32(val, optval_addr)) {
2365 return -TARGET_EFAULT;
2366 }
2367 ret = get_errno(setsockopt(sockfd, level, optname,
2368 &val, sizeof(val)));
2369 break;
2370
2371 default:
2372 goto unimplemented;
2373 }
2374 break;
2375 #if defined(SOL_ALG) && defined(ALG_SET_KEY) && defined(ALG_SET_AEAD_AUTHSIZE)
2376 case SOL_ALG:
2377 switch (optname) {
2378 case ALG_SET_KEY:
2379 {
2380 char *alg_key = g_malloc(optlen);
2381
2382 if (!alg_key) {
2383 return -TARGET_ENOMEM;
2384 }
2385 if (copy_from_user(alg_key, optval_addr, optlen)) {
2386 g_free(alg_key);
2387 return -TARGET_EFAULT;
2388 }
2389 ret = get_errno(setsockopt(sockfd, level, optname,
2390 alg_key, optlen));
2391 g_free(alg_key);
2392 break;
2393 }
2394 case ALG_SET_AEAD_AUTHSIZE:
2395 {
2396 ret = get_errno(setsockopt(sockfd, level, optname,
2397 NULL, optlen));
2398 break;
2399 }
2400 default:
2401 goto unimplemented;
2402 }
2403 break;
2404 #endif
2405 case TARGET_SOL_SOCKET:
2406 switch (optname) {
2407 case TARGET_SO_RCVTIMEO:
2408 {
2409 struct timeval tv;
2410
2411 optname = SO_RCVTIMEO;
2412
2413 set_timeout:
2414 if (optlen != sizeof(struct target_timeval)) {
2415 return -TARGET_EINVAL;
2416 }
2417
2418 if (copy_from_user_timeval(&tv, optval_addr)) {
2419 return -TARGET_EFAULT;
2420 }
2421
2422 ret = get_errno(setsockopt(sockfd, SOL_SOCKET, optname,
2423 &tv, sizeof(tv)));
2424 return ret;
2425 }
2426 case TARGET_SO_SNDTIMEO:
2427 optname = SO_SNDTIMEO;
2428 goto set_timeout;
2429 case TARGET_SO_ATTACH_FILTER:
2430 {
2431 struct target_sock_fprog *tfprog;
2432 struct target_sock_filter *tfilter;
2433 struct sock_fprog fprog;
2434 struct sock_filter *filter;
2435 int i;
2436
2437 if (optlen != sizeof(*tfprog)) {
2438 return -TARGET_EINVAL;
2439 }
2440 if (!lock_user_struct(VERIFY_READ, tfprog, optval_addr, 0)) {
2441 return -TARGET_EFAULT;
2442 }
2443 if (!lock_user_struct(VERIFY_READ, tfilter,
2444 tswapal(tfprog->filter), 0)) {
2445 unlock_user_struct(tfprog, optval_addr, 1);
2446 return -TARGET_EFAULT;
2447 }
2448
2449 fprog.len = tswap16(tfprog->len);
2450 filter = g_try_new(struct sock_filter, fprog.len);
2451 if (filter == NULL) {
2452 unlock_user_struct(tfilter, tfprog->filter, 1);
2453 unlock_user_struct(tfprog, optval_addr, 1);
2454 return -TARGET_ENOMEM;
2455 }
2456 for (i = 0; i < fprog.len; i++) {
2457 filter[i].code = tswap16(tfilter[i].code);
2458 filter[i].jt = tfilter[i].jt;
2459 filter[i].jf = tfilter[i].jf;
2460 filter[i].k = tswap32(tfilter[i].k);
2461 }
2462 fprog.filter = filter;
2463
2464 ret = get_errno(setsockopt(sockfd, SOL_SOCKET,
2465 SO_ATTACH_FILTER, &fprog, sizeof(fprog)));
2466 g_free(filter);
2467
2468 unlock_user_struct(tfilter, tfprog->filter, 1);
2469 unlock_user_struct(tfprog, optval_addr, 1);
2470 return ret;
2471 }
2472 case TARGET_SO_BINDTODEVICE:
2473 {
2474 char *dev_ifname, *addr_ifname;
2475
2476 if (optlen > IFNAMSIZ - 1) {
2477 optlen = IFNAMSIZ - 1;
2478 }
2479 dev_ifname = lock_user(VERIFY_READ, optval_addr, optlen, 1);
2480 if (!dev_ifname) {
2481 return -TARGET_EFAULT;
2482 }
2483 optname = SO_BINDTODEVICE;
2484 addr_ifname = alloca(IFNAMSIZ);
2485 memcpy(addr_ifname, dev_ifname, optlen);
2486 addr_ifname[optlen] = 0;
2487 ret = get_errno(setsockopt(sockfd, SOL_SOCKET, optname,
2488 addr_ifname, optlen));
2489 unlock_user (dev_ifname, optval_addr, 0);
2490 return ret;
2491 }
2492 case TARGET_SO_LINGER:
2493 {
2494 struct linger lg;
2495 struct target_linger *tlg;
2496
2497 if (optlen != sizeof(struct target_linger)) {
2498 return -TARGET_EINVAL;
2499 }
2500 if (!lock_user_struct(VERIFY_READ, tlg, optval_addr, 1)) {
2501 return -TARGET_EFAULT;
2502 }
2503 __get_user(lg.l_onoff, &tlg->l_onoff);
2504 __get_user(lg.l_linger, &tlg->l_linger);
2505 ret = get_errno(setsockopt(sockfd, SOL_SOCKET, SO_LINGER,
2506 &lg, sizeof(lg)));
2507 unlock_user_struct(tlg, optval_addr, 0);
2508 return ret;
2509 }
2510 /* Options with 'int' argument. */
2511 case TARGET_SO_DEBUG:
2512 optname = SO_DEBUG;
2513 break;
2514 case TARGET_SO_REUSEADDR:
2515 optname = SO_REUSEADDR;
2516 break;
2517 #ifdef SO_REUSEPORT
2518 case TARGET_SO_REUSEPORT:
2519 optname = SO_REUSEPORT;
2520 break;
2521 #endif
2522 case TARGET_SO_TYPE:
2523 optname = SO_TYPE;
2524 break;
2525 case TARGET_SO_ERROR:
2526 optname = SO_ERROR;
2527 break;
2528 case TARGET_SO_DONTROUTE:
2529 optname = SO_DONTROUTE;
2530 break;
2531 case TARGET_SO_BROADCAST:
2532 optname = SO_BROADCAST;
2533 break;
2534 case TARGET_SO_SNDBUF:
2535 optname = SO_SNDBUF;
2536 break;
2537 case TARGET_SO_SNDBUFFORCE:
2538 optname = SO_SNDBUFFORCE;
2539 break;
2540 case TARGET_SO_RCVBUF:
2541 optname = SO_RCVBUF;
2542 break;
2543 case TARGET_SO_RCVBUFFORCE:
2544 optname = SO_RCVBUFFORCE;
2545 break;
2546 case TARGET_SO_KEEPALIVE:
2547 optname = SO_KEEPALIVE;
2548 break;
2549 case TARGET_SO_OOBINLINE:
2550 optname = SO_OOBINLINE;
2551 break;
2552 case TARGET_SO_NO_CHECK:
2553 optname = SO_NO_CHECK;
2554 break;
2555 case TARGET_SO_PRIORITY:
2556 optname = SO_PRIORITY;
2557 break;
2558 #ifdef SO_BSDCOMPAT
2559 case TARGET_SO_BSDCOMPAT:
2560 optname = SO_BSDCOMPAT;
2561 break;
2562 #endif
2563 case TARGET_SO_PASSCRED:
2564 optname = SO_PASSCRED;
2565 break;
2566 case TARGET_SO_PASSSEC:
2567 optname = SO_PASSSEC;
2568 break;
2569 case TARGET_SO_TIMESTAMP:
2570 optname = SO_TIMESTAMP;
2571 break;
2572 case TARGET_SO_RCVLOWAT:
2573 optname = SO_RCVLOWAT;
2574 break;
2575 default:
2576 goto unimplemented;
2577 }
2578 if (optlen < sizeof(uint32_t))
2579 return -TARGET_EINVAL;
2580
2581 if (get_user_u32(val, optval_addr))
2582 return -TARGET_EFAULT;
2583 ret = get_errno(setsockopt(sockfd, SOL_SOCKET, optname, &val, sizeof(val)));
2584 break;
2585 #ifdef SOL_NETLINK
2586 case SOL_NETLINK:
2587 switch (optname) {
2588 case NETLINK_PKTINFO:
2589 case NETLINK_ADD_MEMBERSHIP:
2590 case NETLINK_DROP_MEMBERSHIP:
2591 case NETLINK_BROADCAST_ERROR:
2592 case NETLINK_NO_ENOBUFS:
2593 #if LINUX_VERSION_CODE >= KERNEL_VERSION(4, 2, 0)
2594 case NETLINK_LISTEN_ALL_NSID:
2595 case NETLINK_CAP_ACK:
2596 #endif /* LINUX_VERSION_CODE >= KERNEL_VERSION(4, 2, 0) */
2597 #if LINUX_VERSION_CODE >= KERNEL_VERSION(4, 12, 0)
2598 case NETLINK_EXT_ACK:
2599 #endif /* LINUX_VERSION_CODE >= KERNEL_VERSION(4, 12, 0) */
2600 #if LINUX_VERSION_CODE >= KERNEL_VERSION(4, 20, 0)
2601 case NETLINK_GET_STRICT_CHK:
2602 #endif /* LINUX_VERSION_CODE >= KERNEL_VERSION(4, 12, 0) */
2603 break;
2604 default:
2605 goto unimplemented;
2606 }
2607 val = 0;
2608 if (optlen < sizeof(uint32_t)) {
2609 return -TARGET_EINVAL;
2610 }
2611 if (get_user_u32(val, optval_addr)) {
2612 return -TARGET_EFAULT;
2613 }
2614 ret = get_errno(setsockopt(sockfd, SOL_NETLINK, optname, &val,
2615 sizeof(val)));
2616 break;
2617 #endif /* SOL_NETLINK */
2618 default:
2619 unimplemented:
2620 qemu_log_mask(LOG_UNIMP, "Unsupported setsockopt level=%d optname=%d\n",
2621 level, optname);
2622 ret = -TARGET_ENOPROTOOPT;
2623 }
2624 return ret;
2625 }
2626
2627 /* do_getsockopt() Must return target values and target errnos. */
2628 static abi_long do_getsockopt(int sockfd, int level, int optname,
2629 abi_ulong optval_addr, abi_ulong optlen)
2630 {
2631 abi_long ret;
2632 int len, val;
2633 socklen_t lv;
2634
2635 switch(level) {
2636 case TARGET_SOL_SOCKET:
2637 level = SOL_SOCKET;
2638 switch (optname) {
2639 /* These don't just return a single integer */
2640 case TARGET_SO_PEERNAME:
2641 goto unimplemented;
2642 case TARGET_SO_RCVTIMEO: {
2643 struct timeval tv;
2644 socklen_t tvlen;
2645
2646 optname = SO_RCVTIMEO;
2647
2648 get_timeout:
2649 if (get_user_u32(len, optlen)) {
2650 return -TARGET_EFAULT;
2651 }
2652 if (len < 0) {
2653 return -TARGET_EINVAL;
2654 }
2655
2656 tvlen = sizeof(tv);
2657 ret = get_errno(getsockopt(sockfd, level, optname,
2658 &tv, &tvlen));
2659 if (ret < 0) {
2660 return ret;
2661 }
2662 if (len > sizeof(struct target_timeval)) {
2663 len = sizeof(struct target_timeval);
2664 }
2665 if (copy_to_user_timeval(optval_addr, &tv)) {
2666 return -TARGET_EFAULT;
2667 }
2668 if (put_user_u32(len, optlen)) {
2669 return -TARGET_EFAULT;
2670 }
2671 break;
2672 }
2673 case TARGET_SO_SNDTIMEO:
2674 optname = SO_SNDTIMEO;
2675 goto get_timeout;
2676 case TARGET_SO_PEERCRED: {
2677 struct ucred cr;
2678 socklen_t crlen;
2679 struct target_ucred *tcr;
2680
2681 if (get_user_u32(len, optlen)) {
2682 return -TARGET_EFAULT;
2683 }
2684 if (len < 0) {
2685 return -TARGET_EINVAL;
2686 }
2687
2688 crlen = sizeof(cr);
2689 ret = get_errno(getsockopt(sockfd, level, SO_PEERCRED,
2690 &cr, &crlen));
2691 if (ret < 0) {
2692 return ret;
2693 }
2694 if (len > crlen) {
2695 len = crlen;
2696 }
2697 if (!lock_user_struct(VERIFY_WRITE, tcr, optval_addr, 0)) {
2698 return -TARGET_EFAULT;
2699 }
2700 __put_user(cr.pid, &tcr->pid);
2701 __put_user(cr.uid, &tcr->uid);
2702 __put_user(cr.gid, &tcr->gid);
2703 unlock_user_struct(tcr, optval_addr, 1);
2704 if (put_user_u32(len, optlen)) {
2705 return -TARGET_EFAULT;
2706 }
2707 break;
2708 }
2709 case TARGET_SO_PEERSEC: {
2710 char *name;
2711
2712 if (get_user_u32(len, optlen)) {
2713 return -TARGET_EFAULT;
2714 }
2715 if (len < 0) {
2716 return -TARGET_EINVAL;
2717 }
2718 name = lock_user(VERIFY_WRITE, optval_addr, len, 0);
2719 if (!name) {
2720 return -TARGET_EFAULT;
2721 }
2722 lv = len;
2723 ret = get_errno(getsockopt(sockfd, level, SO_PEERSEC,
2724 name, &lv));
2725 if (put_user_u32(lv, optlen)) {
2726 ret = -TARGET_EFAULT;
2727 }
2728 unlock_user(name, optval_addr, lv);
2729 break;
2730 }
2731 case TARGET_SO_LINGER:
2732 {
2733 struct linger lg;
2734 socklen_t lglen;
2735 struct target_linger *tlg;
2736
2737 if (get_user_u32(len, optlen)) {
2738 return -TARGET_EFAULT;
2739 }
2740 if (len < 0) {
2741 return -TARGET_EINVAL;
2742 }
2743
2744 lglen = sizeof(lg);
2745 ret = get_errno(getsockopt(sockfd, level, SO_LINGER,
2746 &lg, &lglen));
2747 if (ret < 0) {
2748 return ret;
2749 }
2750 if (len > lglen) {
2751 len = lglen;
2752 }
2753 if (!lock_user_struct(VERIFY_WRITE, tlg, optval_addr, 0)) {
2754 return -TARGET_EFAULT;
2755 }
2756 __put_user(lg.l_onoff, &tlg->l_onoff);
2757 __put_user(lg.l_linger, &tlg->l_linger);
2758 unlock_user_struct(tlg, optval_addr, 1);
2759 if (put_user_u32(len, optlen)) {
2760 return -TARGET_EFAULT;
2761 }
2762 break;
2763 }
2764 /* Options with 'int' argument. */
2765 case TARGET_SO_DEBUG:
2766 optname = SO_DEBUG;
2767 goto int_case;
2768 case TARGET_SO_REUSEADDR:
2769 optname = SO_REUSEADDR;
2770 goto int_case;
2771 #ifdef SO_REUSEPORT
2772 case TARGET_SO_REUSEPORT:
2773 optname = SO_REUSEPORT;
2774 goto int_case;
2775 #endif
2776 case TARGET_SO_TYPE:
2777 optname = SO_TYPE;
2778 goto int_case;
2779 case TARGET_SO_ERROR:
2780 optname = SO_ERROR;
2781 goto int_case;
2782 case TARGET_SO_DONTROUTE:
2783 optname = SO_DONTROUTE;
2784 goto int_case;
2785 case TARGET_SO_BROADCAST:
2786 optname = SO_BROADCAST;
2787 goto int_case;
2788 case TARGET_SO_SNDBUF:
2789 optname = SO_SNDBUF;
2790 goto int_case;
2791 case TARGET_SO_RCVBUF:
2792 optname = SO_RCVBUF;
2793 goto int_case;
2794 case TARGET_SO_KEEPALIVE:
2795 optname = SO_KEEPALIVE;
2796 goto int_case;
2797 case TARGET_SO_OOBINLINE:
2798 optname = SO_OOBINLINE;
2799 goto int_case;
2800 case TARGET_SO_NO_CHECK:
2801 optname = SO_NO_CHECK;
2802 goto int_case;
2803 case TARGET_SO_PRIORITY:
2804 optname = SO_PRIORITY;
2805 goto int_case;
2806 #ifdef SO_BSDCOMPAT
2807 case TARGET_SO_BSDCOMPAT:
2808 optname = SO_BSDCOMPAT;
2809 goto int_case;
2810 #endif
2811 case TARGET_SO_PASSCRED:
2812 optname = SO_PASSCRED;
2813 goto int_case;
2814 case TARGET_SO_TIMESTAMP:
2815 optname = SO_TIMESTAMP;
2816 goto int_case;
2817 case TARGET_SO_RCVLOWAT:
2818 optname = SO_RCVLOWAT;
2819 goto int_case;
2820 case TARGET_SO_ACCEPTCONN:
2821 optname = SO_ACCEPTCONN;
2822 goto int_case;
2823 default:
2824 goto int_case;
2825 }
2826 break;
2827 case SOL_TCP:
2828 /* TCP options all take an 'int' value. */
2829 int_case:
2830 if (get_user_u32(len, optlen))
2831 return -TARGET_EFAULT;
2832 if (len < 0)
2833 return -TARGET_EINVAL;
2834 lv = sizeof(lv);
2835 ret = get_errno(getsockopt(sockfd, level, optname, &val, &lv));
2836 if (ret < 0)
2837 return ret;
2838 if (optname == SO_TYPE) {
2839 val = host_to_target_sock_type(val);
2840 }
2841 if (len > lv)
2842 len = lv;
2843 if (len == 4) {
2844 if (put_user_u32(val, optval_addr))
2845 return -TARGET_EFAULT;
2846 } else {
2847 if (put_user_u8(val, optval_addr))
2848 return -TARGET_EFAULT;
2849 }
2850 if (put_user_u32(len, optlen))
2851 return -TARGET_EFAULT;
2852 break;
2853 case SOL_IP:
2854 switch(optname) {
2855 case IP_TOS:
2856 case IP_TTL:
2857 case IP_HDRINCL:
2858 case IP_ROUTER_ALERT:
2859 case IP_RECVOPTS:
2860 case IP_RETOPTS:
2861 case IP_PKTINFO:
2862 case IP_MTU_DISCOVER:
2863 case IP_RECVERR:
2864 case IP_RECVTOS:
2865 #ifdef IP_FREEBIND
2866 case IP_FREEBIND:
2867 #endif
2868 case IP_MULTICAST_TTL:
2869 case IP_MULTICAST_LOOP:
2870 if (get_user_u32(len, optlen))
2871 return -TARGET_EFAULT;
2872 if (len < 0)
2873 return -TARGET_EINVAL;
2874 lv = sizeof(lv);
2875 ret = get_errno(getsockopt(sockfd, level, optname, &val, &lv));
2876 if (ret < 0)
2877 return ret;
2878 if (len < sizeof(int) && len > 0 && val >= 0 && val < 255) {
2879 len = 1;
2880 if (put_user_u32(len, optlen)
2881 || put_user_u8(val, optval_addr))
2882 return -TARGET_EFAULT;
2883 } else {
2884 if (len > sizeof(int))
2885 len = sizeof(int);
2886 if (put_user_u32(len, optlen)
2887 || put_user_u32(val, optval_addr))
2888 return -TARGET_EFAULT;
2889 }
2890 break;
2891 default:
2892 ret = -TARGET_ENOPROTOOPT;
2893 break;
2894 }
2895 break;
2896 case SOL_IPV6:
2897 switch (optname) {
2898 case IPV6_MTU_DISCOVER:
2899 case IPV6_MTU:
2900 case IPV6_V6ONLY:
2901 case IPV6_RECVPKTINFO:
2902 case IPV6_UNICAST_HOPS:
2903 case IPV6_MULTICAST_HOPS:
2904 case IPV6_MULTICAST_LOOP:
2905 case IPV6_RECVERR:
2906 case IPV6_RECVHOPLIMIT:
2907 case IPV6_2292HOPLIMIT:
2908 case IPV6_CHECKSUM:
2909 case IPV6_ADDRFORM:
2910 case IPV6_2292PKTINFO:
2911 case IPV6_RECVTCLASS:
2912 case IPV6_RECVRTHDR:
2913 case IPV6_2292RTHDR:
2914 case IPV6_RECVHOPOPTS:
2915 case IPV6_2292HOPOPTS:
2916 case IPV6_RECVDSTOPTS:
2917 case IPV6_2292DSTOPTS:
2918 case IPV6_TCLASS:
2919 #ifdef IPV6_RECVPATHMTU
2920 case IPV6_RECVPATHMTU:
2921 #endif
2922 #ifdef IPV6_TRANSPARENT
2923 case IPV6_TRANSPARENT:
2924 #endif
2925 #ifdef IPV6_FREEBIND
2926 case IPV6_FREEBIND:
2927 #endif
2928 #ifdef IPV6_RECVORIGDSTADDR
2929 case IPV6_RECVORIGDSTADDR:
2930 #endif
2931 if (get_user_u32(len, optlen))
2932 return -TARGET_EFAULT;
2933 if (len < 0)
2934 return -TARGET_EINVAL;
2935 lv = sizeof(lv);
2936 ret = get_errno(getsockopt(sockfd, level, optname, &val, &lv));
2937 if (ret < 0)
2938 return ret;
2939 if (len < sizeof(int) && len > 0 && val >= 0 && val < 255) {
2940 len = 1;
2941 if (put_user_u32(len, optlen)
2942 || put_user_u8(val, optval_addr))
2943 return -TARGET_EFAULT;
2944 } else {
2945 if (len > sizeof(int))
2946 len = sizeof(int);
2947 if (put_user_u32(len, optlen)
2948 || put_user_u32(val, optval_addr))
2949 return -TARGET_EFAULT;
2950 }
2951 break;
2952 default:
2953 ret = -TARGET_ENOPROTOOPT;
2954 break;
2955 }
2956 break;
2957 #ifdef SOL_NETLINK
2958 case SOL_NETLINK:
2959 switch (optname) {
2960 case NETLINK_PKTINFO:
2961 case NETLINK_BROADCAST_ERROR:
2962 case NETLINK_NO_ENOBUFS:
2963 #if LINUX_VERSION_CODE >= KERNEL_VERSION(4, 2, 0)
2964 case NETLINK_LISTEN_ALL_NSID:
2965 case NETLINK_CAP_ACK:
2966 #endif /* LINUX_VERSION_CODE >= KERNEL_VERSION(4, 2, 0) */
2967 #if LINUX_VERSION_CODE >= KERNEL_VERSION(4, 12, 0)
2968 case NETLINK_EXT_ACK:
2969 #endif /* LINUX_VERSION_CODE >= KERNEL_VERSION(4, 12, 0) */
2970 #if LINUX_VERSION_CODE >= KERNEL_VERSION(4, 20, 0)
2971 case NETLINK_GET_STRICT_CHK:
2972 #endif /* LINUX_VERSION_CODE >= KERNEL_VERSION(4, 12, 0) */
2973 if (get_user_u32(len, optlen)) {
2974 return -TARGET_EFAULT;
2975 }
2976 if (len != sizeof(val)) {
2977 return -TARGET_EINVAL;
2978 }
2979 lv = len;
2980 ret = get_errno(getsockopt(sockfd, level, optname, &val, &lv));
2981 if (ret < 0) {
2982 return ret;
2983 }
2984 if (put_user_u32(lv, optlen)
2985 || put_user_u32(val, optval_addr)) {
2986 return -TARGET_EFAULT;
2987 }
2988 break;
2989 #if LINUX_VERSION_CODE >= KERNEL_VERSION(4, 2, 0)
2990 case NETLINK_LIST_MEMBERSHIPS:
2991 {
2992 uint32_t *results;
2993 int i;
2994 if (get_user_u32(len, optlen)) {
2995 return -TARGET_EFAULT;
2996 }
2997 if (len < 0) {
2998 return -TARGET_EINVAL;
2999 }
3000 results = lock_user(VERIFY_WRITE, optval_addr, len, 1);
3001 if (!results) {
3002 return -TARGET_EFAULT;
3003 }
3004 lv = len;
3005 ret = get_errno(getsockopt(sockfd, level, optname, results, &lv));
3006 if (ret < 0) {
3007 unlock_user(results, optval_addr, 0);
3008 return ret;
3009 }
3010 /* swap host endianess to target endianess. */
3011 for (i = 0; i < (len / sizeof(uint32_t)); i++) {
3012 results[i] = tswap32(results[i]);
3013 }
3014 if (put_user_u32(lv, optlen)) {
3015 return -TARGET_EFAULT;
3016 }
3017 unlock_user(results, optval_addr, 0);
3018 break;
3019 }
3020 #endif /* LINUX_VERSION_CODE >= KERNEL_VERSION(4, 2, 0) */
3021 default:
3022 goto unimplemented;
3023 }
3024 break;
3025 #endif /* SOL_NETLINK */
3026 default:
3027 unimplemented:
3028 qemu_log_mask(LOG_UNIMP,
3029 "getsockopt level=%d optname=%d not yet supported\n",
3030 level, optname);
3031 ret = -TARGET_EOPNOTSUPP;
3032 break;
3033 }
3034 return ret;
3035 }
3036
3037 /* Convert target low/high pair representing file offset into the host
3038 * low/high pair. This function doesn't handle offsets bigger than 64 bits
3039 * as the kernel doesn't handle them either.
3040 */
3041 static void target_to_host_low_high(abi_ulong tlow,
3042 abi_ulong thigh,
3043 unsigned long *hlow,
3044 unsigned long *hhigh)
3045 {
3046 uint64_t off = tlow |
3047 ((unsigned long long)thigh << TARGET_LONG_BITS / 2) <<
3048 TARGET_LONG_BITS / 2;
3049
3050 *hlow = off;
3051 *hhigh = (off >> HOST_LONG_BITS / 2) >> HOST_LONG_BITS / 2;
3052 }
3053
3054 static struct iovec *lock_iovec(int type, abi_ulong target_addr,
3055 abi_ulong count, int copy)
3056 {
3057 struct target_iovec *target_vec;
3058 struct iovec *vec;
3059 abi_ulong total_len, max_len;
3060 int i;
3061 int err = 0;
3062 bool bad_address = false;
3063
3064 if (count == 0) {
3065 errno = 0;
3066 return NULL;
3067 }
3068 if (count > IOV_MAX) {
3069 errno = EINVAL;
3070 return NULL;
3071 }
3072
3073 vec = g_try_new0(struct iovec, count);
3074 if (vec == NULL) {
3075 errno = ENOMEM;
3076 return NULL;
3077 }
3078
3079 target_vec = lock_user(VERIFY_READ, target_addr,
3080 count * sizeof(struct target_iovec), 1);
3081 if (target_vec == NULL) {
3082 err = EFAULT;
3083 goto fail2;
3084 }
3085
3086 /* ??? If host page size > target page size, this will result in a
3087 value larger than what we can actually support. */
3088 max_len = 0x7fffffff & TARGET_PAGE_MASK;
3089 total_len = 0;
3090
3091 for (i = 0; i < count; i++) {
3092 abi_ulong base = tswapal(target_vec[i].iov_base);
3093 abi_long len = tswapal(target_vec[i].iov_len);
3094
3095 if (len < 0) {
3096 err = EINVAL;
3097 goto fail;
3098 } else if (len == 0) {
3099 /* Zero length pointer is ignored. */
3100 vec[i].iov_base = 0;
3101 } else {
3102 vec[i].iov_base = lock_user(type, base, len, copy);
3103 /* If the first buffer pointer is bad, this is a fault. But
3104 * subsequent bad buffers will result in a partial write; this
3105 * is realized by filling the vector with null pointers and
3106 * zero lengths. */
3107 if (!vec[i].iov_base) {
3108 if (i == 0) {
3109 err = EFAULT;
3110 goto fail;
3111 } else {
3112 bad_address = true;
3113 }
3114 }
3115 if (bad_address) {
3116 len = 0;
3117 }
3118 if (len > max_len - total_len) {
3119 len = max_len - total_len;
3120 }
3121 }
3122 vec[i].iov_len = len;
3123 total_len += len;
3124 }
3125
3126 unlock_user(target_vec, target_addr, 0);
3127 return vec;
3128
3129 fail:
3130 while (--i >= 0) {
3131 if (tswapal(target_vec[i].iov_len) > 0) {
3132 unlock_user(vec[i].iov_base, tswapal(target_vec[i].iov_base), 0);
3133 }
3134 }
3135 unlock_user(target_vec, target_addr, 0);
3136 fail2:
3137 g_free(vec);
3138 errno = err;
3139 return NULL;
3140 }
3141
3142 static void unlock_iovec(struct iovec *vec, abi_ulong target_addr,
3143 abi_ulong count, int copy)
3144 {
3145 struct target_iovec *target_vec;
3146 int i;
3147
3148 target_vec = lock_user(VERIFY_READ, target_addr,
3149 count * sizeof(struct target_iovec), 1);
3150 if (target_vec) {
3151 for (i = 0; i < count; i++) {
3152 abi_ulong base = tswapal(target_vec[i].iov_base);
3153 abi_long len = tswapal(target_vec[i].iov_len);
3154 if (len < 0) {
3155 break;
3156 }
3157 unlock_user(vec[i].iov_base, base, copy ? vec[i].iov_len : 0);
3158 }
3159 unlock_user(target_vec, target_addr, 0);
3160 }
3161
3162 g_free(vec);
3163 }
3164
3165 static inline int target_to_host_sock_type(int *type)
3166 {
3167 int host_type = 0;
3168 int target_type = *type;
3169
3170 switch (target_type & TARGET_SOCK_TYPE_MASK) {
3171 case TARGET_SOCK_DGRAM:
3172 host_type = SOCK_DGRAM;
3173 break;
3174 case TARGET_SOCK_STREAM:
3175 host_type = SOCK_STREAM;
3176 break;
3177 default:
3178 host_type = target_type & TARGET_SOCK_TYPE_MASK;
3179 break;
3180 }
3181 if (target_type & TARGET_SOCK_CLOEXEC) {
3182 #if defined(SOCK_CLOEXEC)
3183 host_type |= SOCK_CLOEXEC;
3184 #else
3185 return -TARGET_EINVAL;
3186 #endif
3187 }
3188 if (target_type & TARGET_SOCK_NONBLOCK) {
3189 #if defined(SOCK_NONBLOCK)
3190 host_type |= SOCK_NONBLOCK;
3191 #elif !defined(O_NONBLOCK)
3192 return -TARGET_EINVAL;
3193 #endif
3194 }
3195 *type = host_type;
3196 return 0;
3197 }
3198
3199 /* Try to emulate socket type flags after socket creation. */
3200 static int sock_flags_fixup(int fd, int target_type)
3201 {
3202 #if !defined(SOCK_NONBLOCK) && defined(O_NONBLOCK)
3203 if (target_type & TARGET_SOCK_NONBLOCK) {
3204 int flags = fcntl(fd, F_GETFL);
3205 if (fcntl(fd, F_SETFL, O_NONBLOCK | flags) == -1) {
3206 close(fd);
3207 return -TARGET_EINVAL;
3208 }
3209 }
3210 #endif
3211 return fd;
3212 }
3213
3214 /* do_socket() Must return target values and target errnos. */
3215 static abi_long do_socket(int domain, int type, int protocol)
3216 {
3217 int target_type = type;
3218 int ret;
3219
3220 ret = target_to_host_sock_type(&type);
3221 if (ret) {
3222 return ret;
3223 }
3224
3225 if (domain == PF_NETLINK && !(
3226 #ifdef CONFIG_RTNETLINK
3227 protocol == NETLINK_ROUTE ||
3228 #endif
3229 protocol == NETLINK_KOBJECT_UEVENT ||
3230 protocol == NETLINK_AUDIT)) {
3231 return -TARGET_EPROTONOSUPPORT;
3232 }
3233
3234 if (domain == AF_PACKET ||
3235 (domain == AF_INET && type == SOCK_PACKET)) {
3236 protocol = tswap16(protocol);
3237 }
3238
3239 ret = get_errno(socket(domain, type, protocol));
3240 if (ret >= 0) {
3241 ret = sock_flags_fixup(ret, target_type);
3242 if (type == SOCK_PACKET) {
3243 /* Manage an obsolete case :
3244 * if socket type is SOCK_PACKET, bind by name
3245 */
3246 fd_trans_register(ret, &target_packet_trans);
3247 } else if (domain == PF_NETLINK) {
3248 switch (protocol) {
3249 #ifdef CONFIG_RTNETLINK
3250 case NETLINK_ROUTE:
3251 fd_trans_register(ret, &target_netlink_route_trans);
3252 break;
3253 #endif
3254 case NETLINK_KOBJECT_UEVENT:
3255 /* nothing to do: messages are strings */
3256 break;
3257 case NETLINK_AUDIT:
3258 fd_trans_register(ret, &target_netlink_audit_trans);
3259 break;
3260 default:
3261 g_assert_not_reached();
3262 }
3263 }
3264 }
3265 return ret;
3266 }
3267
3268 /* do_bind() Must return target values and target errnos. */
3269 static abi_long do_bind(int sockfd, abi_ulong target_addr,
3270 socklen_t addrlen)
3271 {
3272 void *addr;
3273 abi_long ret;
3274
3275 if ((int)addrlen < 0) {
3276 return -TARGET_EINVAL;
3277 }
3278
3279 addr = alloca(addrlen+1);
3280
3281 ret = target_to_host_sockaddr(sockfd, addr, target_addr, addrlen);
3282 if (ret)
3283 return ret;
3284
3285 return get_errno(bind(sockfd, addr, addrlen));
3286 }
3287
3288 /* do_connect() Must return target values and target errnos. */
3289 static abi_long do_connect(int sockfd, abi_ulong target_addr,
3290 socklen_t addrlen)
3291 {
3292 void *addr;
3293 abi_long ret;
3294
3295 if ((int)addrlen < 0) {
3296 return -TARGET_EINVAL;
3297 }
3298
3299 addr = alloca(addrlen+1);
3300
3301 ret = target_to_host_sockaddr(sockfd, addr, target_addr, addrlen);
3302 if (ret)
3303 return ret;
3304
3305 return get_errno(safe_connect(sockfd, addr, addrlen));
3306 }
3307
3308 /* do_sendrecvmsg_locked() Must return target values and target errnos. */
3309 static abi_long do_sendrecvmsg_locked(int fd, struct target_msghdr *msgp,
3310 int flags, int send)
3311 {
3312 abi_long ret, len;
3313 struct msghdr msg;
3314 abi_ulong count;
3315