osdep.c

   1 /*
   2  * QEMU low level functions
   3  *
   4  * Copyright (c) 2003 Fabrice Bellard
   5  *
   6  * Permission is hereby granted, free of charge, to any person obtaining a copy
   7  * of this software and associated documentation files (the "Software"), to deal
   8  * in the Software without restriction, including without limitation the rights
   9  * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
  10  * copies of the Software, and to permit persons to whom the Software is
  11  * furnished to do so, subject to the following conditions:
  12  *
  13  * The above copyright notice and this permission notice shall be included in
  14  * all copies or substantial portions of the Software.
  15  *
  16  * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
  17  * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
  18  * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
  19  * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
  20  * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
  21  * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
  22  * THE SOFTWARE.
  23  */
  24 #include <stdlib.h>
  25 #include <stdio.h>
  26 #include <stdarg.h>
  27 #include <stdbool.h>
  28 #include <string.h>
  29 #include <errno.h>
  30 #include <unistd.h>
  31 #include <fcntl.h>
  32
  33 /* Needed early for CONFIG_BSD etc. */
  34 #include "config-host.h"
  35
  36 #if defined(CONFIG_MADVISE) || defined(CONFIG_POSIX_MADVISE)
  37 #include <sys/mman.h>
  38 #endif
  39
  40 #ifdef CONFIG_SOLARIS
  41 #include <sys/types.h>
  42 #include <sys/statvfs.h>
  43 /* See MySQL bug #7156 (http://bugs.mysql.com/bug.php?id=7156) for
  44    discussion about Solaris header problems */
  45 extern int madvise(caddr_t, size_t, int);
  46 #endif
  47
  48 #include "qemu-common.h"
  49 #include "trace.h"
  50 #include "qemu_socket.h"
  51 #include "monitor.h"
  52
  53 static bool fips_enabled = false;
  54
  55 static const char *qemu_version = QEMU_VERSION;
  56
  57 int socket_set_cork(int fd, int v)
  58 {
  59 #if defined(SOL_TCP) && defined(TCP_CORK)
  60     return setsockopt(fd, SOL_TCP, TCP_CORK, &v, sizeof(v));
  61 #else
  62     return 0;
  63 #endif
  64 }
  65
  66 int qemu_madvise(void *addr, size_t len, int advice)
  67 {
  68     if (advice == QEMU_MADV_INVALID) {
  69         errno = EINVAL;
  70         return -1;
  71     }
  72 #if defined(CONFIG_MADVISE)
  73     return madvise(addr, len, advice);
  74 #elif defined(CONFIG_POSIX_MADVISE)
  75     return posix_madvise(addr, len, advice);
  76 #else
  77     errno = EINVAL;
  78     return -1;
  79 #endif
  80 }
  81
  82 #ifndef _WIN32
  83 /*
  84  * Dups an fd and sets the flags
  85  */
  86 static int qemu_dup_flags(int fd, int flags)
  87 {
  88     int ret;
  89     int serrno;
  90     int dup_flags;
  91     int setfl_flags;
  92
  93 #ifdef F_DUPFD_CLOEXEC
  94     ret = fcntl(fd, F_DUPFD_CLOEXEC, 0);
  95 #else
  96     ret = dup(fd);
  97     if (ret != -1) {
  98         qemu_set_cloexec(ret);
  99     }
 100 #endif
 101     if (ret == -1) {
 102         goto fail;
 103     }
 104
 105     dup_flags = fcntl(ret, F_GETFL);
 106     if (dup_flags == -1) {
 107         goto fail;
 108     }
 109
 110     if ((flags & O_SYNC) != (dup_flags & O_SYNC)) {
 111         errno = EINVAL;
 112         goto fail;
 113     }
 114
 115     /* Set/unset flags that we can with fcntl */
 116     setfl_flags = O_APPEND | O_ASYNC | O_NONBLOCK;
 117 #ifdef O_NOATIME
 118     setfl_flags |= O_NOATIME;
 119 #endif
 120 #ifdef O_DIRECT
 121     setfl_flags |= O_DIRECT;
 122 #endif
 123     dup_flags &= ~setfl_flags;
 124     dup_flags |= (flags & setfl_flags);
 125     if (fcntl(ret, F_SETFL, dup_flags) == -1) {
 126         goto fail;
 127     }
 128
 129     /* Truncate the file in the cases that open() would truncate it */
 130     if (flags & O_TRUNC ||
 131             ((flags & (O_CREAT | O_EXCL)) == (O_CREAT | O_EXCL))) {
 132         if (ftruncate(ret, 0) == -1) {
 133             goto fail;
 134         }
 135     }
 136
 137     return ret;
 138
 139 fail:
 140     serrno = errno;
 141     if (ret != -1) {
 142         close(ret);
 143     }
 144     errno = serrno;
 145     return -1;
 146 }
 147
 148 static int qemu_parse_fdset(const char *param)
 149 {
 150     return qemu_parse_fd(param);
 151 }
 152 #endif
 153
 154 /*
 155  * Opens a file with FD_CLOEXEC set
 156  */
 157 int qemu_open(const char *name, int flags, ...)
 158 {
 159     int ret;
 160     int mode = 0;
 161
 162 #ifndef _WIN32
 163     const char *fdset_id_str;
 164
 165     /* Attempt dup of fd from fd set */
 166     if (strstart(name, "/dev/fdset/", &fdset_id_str)) {
 167         int64_t fdset_id;
 168         int fd, dupfd;
 169
 170         fdset_id = qemu_parse_fdset(fdset_id_str);
 171         if (fdset_id == -1) {
 172             errno = EINVAL;
 173             return -1;
 174         }
 175
 176         fd = monitor_fdset_get_fd(fdset_id, flags);
 177         if (fd == -1) {
 178             return -1;
 179         }
 180
 181         dupfd = qemu_dup_flags(fd, flags);
 182         if (dupfd == -1) {
 183             return -1;
 184         }
 185
 186         ret = monitor_fdset_dup_fd_add(fdset_id, dupfd);
 187         if (ret == -1) {
 188             close(dupfd);
 189             errno = EINVAL;
 190             return -1;
 191         }
 192
 193         return dupfd;
 194     }
 195 #endif
 196
 197     if (flags & O_CREAT) {
 198         va_list ap;
 199
 200         va_start(ap, flags);
 201         mode = va_arg(ap, int);
 202         va_end(ap);
 203     }
 204
 205 #ifdef O_CLOEXEC
 206     ret = open(name, flags | O_CLOEXEC, mode);
 207 #else
 208     ret = open(name, flags, mode);
 209     if (ret >= 0) {
 210         qemu_set_cloexec(ret);
 211     }
 212 #endif
 213
 214     return ret;
 215 }
 216
 217 int qemu_close(int fd)
 218 {
 219     int64_t fdset_id;
 220
 221     /* Close fd that was dup'd from an fdset */
 222     fdset_id = monitor_fdset_dup_fd_find(fd);
 223     if (fdset_id != -1) {
 224         int ret;
 225
 226         ret = close(fd);
 227         if (ret == 0) {
 228             monitor_fdset_dup_fd_remove(fd);
 229         }
 230
 231         return ret;
 232     }
 233
 234     return close(fd);
 235 }
 236
 237 /*
 238  * A variant of write(2) which handles partial write.
 239  *
 240  * Return the number of bytes transferred.
 241  * Set errno if fewer than `count' bytes are written.
 242  *
 243  * This function don't work with non-blocking fd's.
 244  * Any of the possibilities with non-bloking fd's is bad:
 245  *   - return a short write (then name is wrong)
 246  *   - busy wait adding (errno == EAGAIN) to the loop
 247  */
 248 ssize_t qemu_write_full(int fd, const void *buf, size_t count)
 249 {
 250     ssize_t ret = 0;
 251     ssize_t total = 0;
 252
 253     while (count) {
 254         ret = write(fd, buf, count);
 255         if (ret < 0) {
 256             if (errno == EINTR)
 257                 continue;
 258             break;
 259         }
 260
 261         count -= ret;
 262         buf += ret;
 263         total += ret;
 264     }
 265
 266     return total;
 267 }
 268
 269 /*
 270  * Opens a socket with FD_CLOEXEC set
 271  */
 272 int qemu_socket(int domain, int type, int protocol)
 273 {
 274     int ret;
 275
 276 #ifdef SOCK_CLOEXEC
 277     ret = socket(domain, type | SOCK_CLOEXEC, protocol);
 278     if (ret != -1 || errno != EINVAL) {
 279         return ret;
 280     }
 281 #endif
 282     ret = socket(domain, type, protocol);
 283     if (ret >= 0) {
 284         qemu_set_cloexec(ret);
 285     }
 286
 287     return ret;
 288 }
 289
 290 /*
 291  * Accept a connection and set FD_CLOEXEC
 292  */
 293 int qemu_accept(int s, struct sockaddr *addr, socklen_t *addrlen)
 294 {
 295     int ret;
 296
 297 #ifdef CONFIG_ACCEPT4
 298     ret = accept4(s, addr, addrlen, SOCK_CLOEXEC);
 299     if (ret != -1 || errno != ENOSYS) {
 300         return ret;
 301     }
 302 #endif
 303     ret = accept(s, addr, addrlen);
 304     if (ret >= 0) {
 305         qemu_set_cloexec(ret);
 306     }
 307
 308     return ret;
 309 }
 310
 311 /*
 312  * A variant of send(2) which handles partial write.
 313  *
 314  * Return the number of bytes transferred, which is only
 315  * smaller than `count' if there is an error.
 316  *
 317  * This function won't work with non-blocking fd's.
 318  * Any of the possibilities with non-bloking fd's is bad:
 319  *   - return a short write (then name is wrong)
 320  *   - busy wait adding (errno == EAGAIN) to the loop
 321  */
 322 ssize_t qemu_send_full(int fd, const void *buf, size_t count, int flags)
 323 {
 324     ssize_t ret = 0;
 325     ssize_t total = 0;
 326
 327     while (count) {
 328         ret = send(fd, buf, count, flags);
 329         if (ret < 0) {
 330             if (errno == EINTR) {
 331                 continue;
 332             }
 333             break;
 334         }
 335
 336         count -= ret;
 337         buf += ret;
 338         total += ret;
 339     }
 340
 341     return total;
 342 }
 343
 344 /*
 345  * A variant of recv(2) which handles partial write.
 346  *
 347  * Return the number of bytes transferred, which is only
 348  * smaller than `count' if there is an error.
 349  *
 350  * This function won't work with non-blocking fd's.
 351  * Any of the possibilities with non-bloking fd's is bad:
 352  *   - return a short write (then name is wrong)
 353  *   - busy wait adding (errno == EAGAIN) to the loop
 354  */
 355 ssize_t qemu_recv_full(int fd, void *buf, size_t count, int flags)
 356 {
 357     ssize_t ret = 0;
 358     ssize_t total = 0;
 359
 360     while (count) {
 361         ret = qemu_recv(fd, buf, count, flags);
 362         if (ret <= 0) {
 363             if (ret < 0 && errno == EINTR) {
 364                 continue;
 365             }
 366             break;
 367         }
 368
 369         count -= ret;
 370         buf += ret;
 371         total += ret;
 372     }
 373
 374     return total;
 375 }
 376
 377 void qemu_set_version(const char *version)
 378 {
 379     qemu_version = version;
 380 }
 381
 382 const char *qemu_get_version(void)
 383 {
 384     return qemu_version;
 385 }
 386
 387 void fips_set_state(bool requested)
 388 {
 389 #ifdef __linux__
 390     if (requested) {
 391         FILE *fds = fopen("/proc/sys/crypto/fips_enabled", "r");
 392         if (fds != NULL) {
 393             fips_enabled = (fgetc(fds) == '1');
 394             fclose(fds);
 395         }
 396     }
 397 #else
 398     fips_enabled = false;
 399 #endif /* __linux__ */
 400
 401 #ifdef _FIPS_DEBUG
 402     fprintf(stderr, "FIPS mode %s (requested %s)\n",
 403             (fips_enabled ? "enabled" : "disabled"),
 404             (requested ? "enabled" : "disabled"));
 405 #endif
 406 }
 407
 408 bool fips_get_state(void)
 409 {
 410     return fips_enabled;
 411 }
 412
 413
 414 static int default_fdset_get_fd(int64_t fdset_id, int flags)
 415 {
 416     return -1;
 417 }
 418 QEMU_WEAK_ALIAS(monitor_fdset_get_fd, default_fdset_get_fd);
 419
 420 static int default_fdset_dup_fd_add(int64_t fdset_id, int dup_fd)
 421 {
 422     return -1;
 423 }
 424 QEMU_WEAK_ALIAS(monitor_fdset_dup_fd_add, default_fdset_dup_fd_add);
 425
 426 static int default_fdset_dup_fd_remove(int dup_fd)
 427 {
 428     return -1;
 429 }
 430 QEMU_WEAK_ALIAS(monitor_fdset_dup_fd_remove, default_fdset_dup_fd_remove);
 431
 432 static int default_fdset_dup_fd_find(int dup_fd)
 433 {
 434     return -1;
 435 }
 436 QEMU_WEAK_ALIAS(monitor_fdset_dup_fd_find, default_fdset_dup_fd_find);