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 #endif
 148
 149 /*
 150  * Opens a file with FD_CLOEXEC set
 151  */
 152 int qemu_open(const char *name, int flags, ...)
 153 {
 154     int ret;
 155     int mode = 0;
 156
 157 #ifndef _WIN32
 158     const char *fdset_id_str;
 159
 160     /* Attempt dup of fd from fd set */
 161     if (strstart(name, "/dev/fdset/", &fdset_id_str)) {
 162         int64_t fdset_id;
 163         int fd, dupfd;
 164
 165         fdset_id = qemu_parse_fdset(fdset_id_str);
 166         if (fdset_id == -1) {
 167             errno = EINVAL;
 168             return -1;
 169         }
 170
 171         fd = monitor_fdset_get_fd(fdset_id, flags);
 172         if (fd == -1) {
 173             return -1;
 174         }
 175
 176         dupfd = qemu_dup_flags(fd, flags);
 177         if (dupfd == -1) {
 178             return -1;
 179         }
 180
 181         ret = monitor_fdset_dup_fd_add(fdset_id, dupfd);
 182         if (ret == -1) {
 183             close(dupfd);
 184             errno = EINVAL;
 185             return -1;
 186         }
 187
 188         return dupfd;
 189     }
 190 #endif
 191
 192     if (flags & O_CREAT) {
 193         va_list ap;
 194
 195         va_start(ap, flags);
 196         mode = va_arg(ap, int);
 197         va_end(ap);
 198     }
 199
 200 #ifdef O_CLOEXEC
 201     ret = open(name, flags | O_CLOEXEC, mode);
 202 #else
 203     ret = open(name, flags, mode);
 204     if (ret >= 0) {
 205         qemu_set_cloexec(ret);
 206     }
 207 #endif
 208
 209     return ret;
 210 }
 211
 212 int qemu_close(int fd)
 213 {
 214     int64_t fdset_id;
 215
 216     /* Close fd that was dup'd from an fdset */
 217     fdset_id = monitor_fdset_dup_fd_find(fd);
 218     if (fdset_id != -1) {
 219         int ret;
 220
 221         ret = close(fd);
 222         if (ret == 0) {
 223             monitor_fdset_dup_fd_remove(fd);
 224         }
 225
 226         return ret;
 227     }
 228
 229     return close(fd);
 230 }
 231
 232 /*
 233  * A variant of write(2) which handles partial write.
 234  *
 235  * Return the number of bytes transferred.
 236  * Set errno if fewer than `count' bytes are written.
 237  *
 238  * This function don't work with non-blocking fd's.
 239  * Any of the possibilities with non-bloking fd's is bad:
 240  *   - return a short write (then name is wrong)
 241  *   - busy wait adding (errno == EAGAIN) to the loop
 242  */
 243 ssize_t qemu_write_full(int fd, const void *buf, size_t count)
 244 {
 245     ssize_t ret = 0;
 246     ssize_t total = 0;
 247
 248     while (count) {
 249         ret = write(fd, buf, count);
 250         if (ret < 0) {
 251             if (errno == EINTR)
 252                 continue;
 253             break;
 254         }
 255
 256         count -= ret;
 257         buf += ret;
 258         total += ret;
 259     }
 260
 261     return total;
 262 }
 263
 264 /*
 265  * Opens a socket with FD_CLOEXEC set
 266  */
 267 int qemu_socket(int domain, int type, int protocol)
 268 {
 269     int ret;
 270
 271 #ifdef SOCK_CLOEXEC
 272     ret = socket(domain, type | SOCK_CLOEXEC, protocol);
 273     if (ret != -1 || errno != EINVAL) {
 274         return ret;
 275     }
 276 #endif
 277     ret = socket(domain, type, protocol);
 278     if (ret >= 0) {
 279         qemu_set_cloexec(ret);
 280     }
 281
 282     return ret;
 283 }
 284
 285 /*
 286  * Accept a connection and set FD_CLOEXEC
 287  */
 288 int qemu_accept(int s, struct sockaddr *addr, socklen_t *addrlen)
 289 {
 290     int ret;
 291
 292 #ifdef CONFIG_ACCEPT4
 293     ret = accept4(s, addr, addrlen, SOCK_CLOEXEC);
 294     if (ret != -1 || errno != ENOSYS) {
 295         return ret;
 296     }
 297 #endif
 298     ret = accept(s, addr, addrlen);
 299     if (ret >= 0) {
 300         qemu_set_cloexec(ret);
 301     }
 302
 303     return ret;
 304 }
 305
 306 /*
 307  * A variant of send(2) which handles partial write.
 308  *
 309  * Return the number of bytes transferred, which is only
 310  * smaller than `count' if there is an error.
 311  *
 312  * This function won't work with non-blocking fd's.
 313  * Any of the possibilities with non-bloking fd's is bad:
 314  *   - return a short write (then name is wrong)
 315  *   - busy wait adding (errno == EAGAIN) to the loop
 316  */
 317 ssize_t qemu_send_full(int fd, const void *buf, size_t count, int flags)
 318 {
 319     ssize_t ret = 0;
 320     ssize_t total = 0;
 321
 322     while (count) {
 323         ret = send(fd, buf, count, flags);
 324         if (ret < 0) {
 325             if (errno == EINTR) {
 326                 continue;
 327             }
 328             break;
 329         }
 330
 331         count -= ret;
 332         buf += ret;
 333         total += ret;
 334     }
 335
 336     return total;
 337 }
 338
 339 /*
 340  * A variant of recv(2) which handles partial write.
 341  *
 342  * Return the number of bytes transferred, which is only
 343  * smaller than `count' if there is an error.
 344  *
 345  * This function won't work with non-blocking fd's.
 346  * Any of the possibilities with non-bloking fd's is bad:
 347  *   - return a short write (then name is wrong)
 348  *   - busy wait adding (errno == EAGAIN) to the loop
 349  */
 350 ssize_t qemu_recv_full(int fd, void *buf, size_t count, int flags)
 351 {
 352     ssize_t ret = 0;
 353     ssize_t total = 0;
 354
 355     while (count) {
 356         ret = qemu_recv(fd, buf, count, flags);
 357         if (ret <= 0) {
 358             if (ret < 0 && errno == EINTR) {
 359                 continue;
 360             }
 361             break;
 362         }
 363
 364         count -= ret;
 365         buf += ret;
 366         total += ret;
 367     }
 368
 369     return total;
 370 }
 371
 372 void qemu_set_version(const char *version)
 373 {
 374     qemu_version = version;
 375 }
 376
 377 const char *qemu_get_version(void)
 378 {
 379     return qemu_version;
 380 }
 381
 382 void fips_set_state(bool requested)
 383 {
 384 #ifdef __linux__
 385     if (requested) {
 386         FILE *fds = fopen("/proc/sys/crypto/fips_enabled", "r");
 387         if (fds != NULL) {
 388             fips_enabled = (fgetc(fds) == '1');
 389             fclose(fds);
 390         }
 391     }
 392 #else
 393     fips_enabled = false;
 394 #endif /* __linux__ */
 395
 396 #ifdef _FIPS_DEBUG
 397     fprintf(stderr, "FIPS mode %s (requested %s)\n",
 398             (fips_enabled ? "enabled" : "disabled"),
 399             (requested ? "enabled" : "disabled"));
 400 #endif
 401 }
 402
 403 bool fips_get_state(void)
 404 {
 405     return fips_enabled;
 406 }