blob: 0092bb61b953da5a5f374bbc6b4b544a802bd892 [file] [log] [blame]
bellardea888122004-02-16 22:12:40 +00001/*
2 * QEMU low level functions
ths5fafdf22007-09-16 21:08:06 +00003 *
bellardea888122004-02-16 22:12:40 +00004 * Copyright (c) 2003 Fabrice Bellard
ths5fafdf22007-09-16 21:08:06 +00005 *
bellardea888122004-02-16 22:12:40 +00006 * 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>
Paul Moore0f669982012-08-03 14:39:21 -040027#include <stdbool.h>
bellardea888122004-02-16 22:12:40 +000028#include <string.h>
bellardea888122004-02-16 22:12:40 +000029#include <errno.h>
30#include <unistd.h>
thsaa26bb22007-03-25 21:33:06 +000031#include <fcntl.h>
Paolo Bonzinif582af52010-02-02 20:33:11 +010032
33/* Needed early for CONFIG_BSD etc. */
34#include "config-host.h"
35
Andreas Färbere78815a2010-09-25 11:26:05 +000036#if defined(CONFIG_MADVISE) || defined(CONFIG_POSIX_MADVISE)
37#include <sys/mman.h>
38#endif
39
Juan Quinteladfe5fff2009-07-27 16:12:40 +020040#ifdef CONFIG_SOLARIS
ths605686c2007-01-17 23:31:19 +000041#include <sys/types.h>
42#include <sys/statvfs.h>
Andreas Färbere78815a2010-09-25 11:26:05 +000043/* See MySQL bug #7156 (http://bugs.mysql.com/bug.php?id=7156) for
44 discussion about Solaris header problems */
45extern int madvise(caddr_t, size_t, int);
ths605686c2007-01-17 23:31:19 +000046#endif
bellardea888122004-02-16 22:12:40 +000047
blueswir1511d2b12009-03-07 15:32:56 +000048#include "qemu-common.h"
Paolo Bonzini1de7afc2012-12-17 18:20:00 +010049#include "qemu/sockets.h"
Markus Armbrusterd49b6832015-03-17 18:29:20 +010050#include "qemu/error-report.h"
Paolo Bonzini83c90892012-12-17 18:19:49 +010051#include "monitor/monitor.h"
aliguori03ff3ca2008-09-15 15:51:35 +000052
Paul Moore0f669982012-08-03 14:39:21 -040053static bool fips_enabled = false;
54
Crístian Viana93bfef42012-05-30 00:35:51 -030055static const char *qemu_version = QEMU_VERSION;
56
Paolo Bonzini128aa582011-09-21 12:36:48 +020057int socket_set_cork(int fd, int v)
58{
59#if defined(SOL_TCP) && defined(TCP_CORK)
Lei Li4bd1afb2013-03-06 22:29:16 +080060 return qemu_setsockopt(fd, SOL_TCP, TCP_CORK, &v, sizeof(v));
Paolo Bonzini128aa582011-09-21 12:36:48 +020061#else
62 return 0;
63#endif
64}
65
MORITA Kazutakabf1c8522013-02-22 12:39:50 +090066int socket_set_nodelay(int fd)
67{
68 int v = 1;
Lei Li4bd1afb2013-03-06 22:29:16 +080069 return qemu_setsockopt(fd, IPPROTO_TCP, TCP_NODELAY, &v, sizeof(v));
MORITA Kazutakabf1c8522013-02-22 12:39:50 +090070}
71
Andreas Färbere78815a2010-09-25 11:26:05 +000072int qemu_madvise(void *addr, size_t len, int advice)
73{
74 if (advice == QEMU_MADV_INVALID) {
75 errno = EINVAL;
76 return -1;
77 }
78#if defined(CONFIG_MADVISE)
79 return madvise(addr, len, advice);
80#elif defined(CONFIG_POSIX_MADVISE)
81 return posix_madvise(addr, len, advice);
82#else
83 errno = EINVAL;
84 return -1;
85#endif
86}
87
Corey Bryantadb696f2012-08-14 16:43:47 -040088#ifndef _WIN32
89/*
90 * Dups an fd and sets the flags
91 */
92static int qemu_dup_flags(int fd, int flags)
93{
94 int ret;
95 int serrno;
96 int dup_flags;
Corey Bryantadb696f2012-08-14 16:43:47 -040097
98#ifdef F_DUPFD_CLOEXEC
99 ret = fcntl(fd, F_DUPFD_CLOEXEC, 0);
100#else
101 ret = dup(fd);
102 if (ret != -1) {
103 qemu_set_cloexec(ret);
104 }
105#endif
106 if (ret == -1) {
107 goto fail;
108 }
109
110 dup_flags = fcntl(ret, F_GETFL);
111 if (dup_flags == -1) {
112 goto fail;
113 }
114
115 if ((flags & O_SYNC) != (dup_flags & O_SYNC)) {
116 errno = EINVAL;
117 goto fail;
118 }
119
120 /* Set/unset flags that we can with fcntl */
Corey Bryant3b6eda22012-10-18 16:41:04 -0400121 if (fcntl(ret, F_SETFL, flags) == -1) {
Corey Bryantadb696f2012-08-14 16:43:47 -0400122 goto fail;
123 }
124
125 /* Truncate the file in the cases that open() would truncate it */
126 if (flags & O_TRUNC ||
127 ((flags & (O_CREAT | O_EXCL)) == (O_CREAT | O_EXCL))) {
128 if (ftruncate(ret, 0) == -1) {
129 goto fail;
130 }
131 }
132
133 return ret;
134
135fail:
136 serrno = errno;
137 if (ret != -1) {
138 close(ret);
139 }
140 errno = serrno;
141 return -1;
142}
Paolo Bonzini0100fbb2012-10-29 15:19:18 +0100143
144static int qemu_parse_fdset(const char *param)
145{
146 return qemu_parse_fd(param);
147}
Corey Bryantadb696f2012-08-14 16:43:47 -0400148#endif
aliguori03ff3ca2008-09-15 15:51:35 +0000149
Kevin Wolf40ff6d72009-12-02 12:24:42 +0100150/*
151 * Opens a file with FD_CLOEXEC set
152 */
153int qemu_open(const char *name, int flags, ...)
154{
155 int ret;
156 int mode = 0;
157
Corey Bryantadb696f2012-08-14 16:43:47 -0400158#ifndef _WIN32
159 const char *fdset_id_str;
160
161 /* Attempt dup of fd from fd set */
162 if (strstart(name, "/dev/fdset/", &fdset_id_str)) {
163 int64_t fdset_id;
164 int fd, dupfd;
165
166 fdset_id = qemu_parse_fdset(fdset_id_str);
167 if (fdset_id == -1) {
168 errno = EINVAL;
169 return -1;
170 }
171
172 fd = monitor_fdset_get_fd(fdset_id, flags);
173 if (fd == -1) {
174 return -1;
175 }
176
177 dupfd = qemu_dup_flags(fd, flags);
178 if (dupfd == -1) {
179 return -1;
180 }
181
182 ret = monitor_fdset_dup_fd_add(fdset_id, dupfd);
183 if (ret == -1) {
184 close(dupfd);
185 errno = EINVAL;
186 return -1;
187 }
188
189 return dupfd;
190 }
191#endif
192
Kevin Wolf40ff6d72009-12-02 12:24:42 +0100193 if (flags & O_CREAT) {
194 va_list ap;
195
196 va_start(ap, flags);
197 mode = va_arg(ap, int);
198 va_end(ap);
199 }
200
201#ifdef O_CLOEXEC
202 ret = open(name, flags | O_CLOEXEC, mode);
203#else
204 ret = open(name, flags, mode);
205 if (ret >= 0) {
206 qemu_set_cloexec(ret);
207 }
208#endif
209
Stefan Hajnoczia5813072013-08-22 11:29:03 +0200210#ifdef O_DIRECT
211 if (ret == -1 && errno == EINVAL && (flags & O_DIRECT)) {
212 error_report("file system may not support O_DIRECT");
213 errno = EINVAL; /* in case it was clobbered */
214 }
215#endif /* O_DIRECT */
216
Kevin Wolf40ff6d72009-12-02 12:24:42 +0100217 return ret;
218}
219
Corey Bryant2e1e79d2012-08-14 16:43:46 -0400220int qemu_close(int fd)
221{
Corey Bryantadb696f2012-08-14 16:43:47 -0400222 int64_t fdset_id;
223
224 /* Close fd that was dup'd from an fdset */
225 fdset_id = monitor_fdset_dup_fd_find(fd);
226 if (fdset_id != -1) {
227 int ret;
228
229 ret = close(fd);
230 if (ret == 0) {
231 monitor_fdset_dup_fd_remove(fd);
232 }
233
234 return ret;
235 }
236
Corey Bryant2e1e79d2012-08-14 16:43:46 -0400237 return close(fd);
238}
239
Kirill A. Shutemov7b5f6992010-01-20 00:56:08 +0100240/*
241 * A variant of write(2) which handles partial write.
242 *
243 * Return the number of bytes transferred.
244 * Set errno if fewer than `count' bytes are written.
Juan Quintela1298cb62010-03-04 10:00:39 +0100245 *
246 * This function don't work with non-blocking fd's.
247 * Any of the possibilities with non-bloking fd's is bad:
248 * - return a short write (then name is wrong)
249 * - busy wait adding (errno == EAGAIN) to the loop
Kirill A. Shutemov7b5f6992010-01-20 00:56:08 +0100250 */
251ssize_t qemu_write_full(int fd, const void *buf, size_t count)
252{
253 ssize_t ret = 0;
254 ssize_t total = 0;
255
256 while (count) {
257 ret = write(fd, buf, count);
258 if (ret < 0) {
259 if (errno == EINTR)
260 continue;
261 break;
262 }
263
264 count -= ret;
265 buf += ret;
266 total += ret;
267 }
268
269 return total;
270}
271
Kevin Wolf40ff6d72009-12-02 12:24:42 +0100272/*
273 * Opens a socket with FD_CLOEXEC set
274 */
275int qemu_socket(int domain, int type, int protocol)
276{
277 int ret;
278
279#ifdef SOCK_CLOEXEC
280 ret = socket(domain, type | SOCK_CLOEXEC, protocol);
Andre Przywara3a03bfa2009-12-18 10:45:07 +0100281 if (ret != -1 || errno != EINVAL) {
282 return ret;
283 }
284#endif
Kevin Wolf40ff6d72009-12-02 12:24:42 +0100285 ret = socket(domain, type, protocol);
286 if (ret >= 0) {
287 qemu_set_cloexec(ret);
288 }
Kevin Wolf40ff6d72009-12-02 12:24:42 +0100289
290 return ret;
291}
292
293/*
294 * Accept a connection and set FD_CLOEXEC
295 */
296int qemu_accept(int s, struct sockaddr *addr, socklen_t *addrlen)
297{
298 int ret;
299
300#ifdef CONFIG_ACCEPT4
301 ret = accept4(s, addr, addrlen, SOCK_CLOEXEC);
Kevin Wolf347ed552010-01-13 16:20:56 +0100302 if (ret != -1 || errno != ENOSYS) {
Andre Przywara3a03bfa2009-12-18 10:45:07 +0100303 return ret;
304 }
305#endif
Kevin Wolf40ff6d72009-12-02 12:24:42 +0100306 ret = accept(s, addr, addrlen);
307 if (ret >= 0) {
308 qemu_set_cloexec(ret);
309 }
Kevin Wolf40ff6d72009-12-02 12:24:42 +0100310
311 return ret;
312}
Paolo Bonzini993295f2011-09-17 16:27:59 +0200313
Crístian Viana93bfef42012-05-30 00:35:51 -0300314void qemu_set_version(const char *version)
315{
316 qemu_version = version;
317}
318
319const char *qemu_get_version(void)
320{
321 return qemu_version;
322}
Paul Moore0f669982012-08-03 14:39:21 -0400323
324void fips_set_state(bool requested)
325{
326#ifdef __linux__
327 if (requested) {
328 FILE *fds = fopen("/proc/sys/crypto/fips_enabled", "r");
329 if (fds != NULL) {
330 fips_enabled = (fgetc(fds) == '1');
331 fclose(fds);
332 }
333 }
334#else
335 fips_enabled = false;
336#endif /* __linux__ */
337
338#ifdef _FIPS_DEBUG
339 fprintf(stderr, "FIPS mode %s (requested %s)\n",
340 (fips_enabled ? "enabled" : "disabled"),
341 (requested ? "enabled" : "disabled"));
342#endif
343}
344
345bool fips_get_state(void)
346{
347 return fips_enabled;
348}
Paolo Bonzini0100fbb2012-10-29 15:19:18 +0100349
Marc-André Lureaud3bf8252013-02-25 23:31:11 +0100350#ifdef _WIN32
351static void socket_cleanup(void)
352{
353 WSACleanup();
354}
355#endif
356
357int socket_init(void)
358{
359#ifdef _WIN32
360 WSADATA Data;
361 int ret, err;
362
363 ret = WSAStartup(MAKEWORD(2, 2), &Data);
364 if (ret != 0) {
365 err = WSAGetLastError();
366 fprintf(stderr, "WSAStartup: %d\n", err);
367 return -1;
368 }
369 atexit(socket_cleanup);
370#endif
371 return 0;
372}
Paolo Bonzini9adea5f2013-04-21 12:01:06 +0200373
Michael Tokarevf33cc842014-05-02 18:35:55 +0400374#if !GLIB_CHECK_VERSION(2, 31, 0)
375/* Ensure that glib is running in multi-threaded mode
376 * Old versions of glib require explicit initialization. Failure to do
377 * this results in the single-threaded code paths being taken inside
378 * glib. For example, the g_slice allocator will not be thread-safe
379 * and cause crashes.
380 */
Stefan Hajnocziae2990c2013-10-08 11:58:31 +0200381static void __attribute__((constructor)) thread_init(void)
382{
383 if (!g_thread_supported()) {
Michael Tokarevf33cc842014-05-02 18:35:55 +0400384 g_thread_init(NULL);
Stefan Hajnocziae2990c2013-10-08 11:58:31 +0200385 }
386}
Michael Tokarevf33cc842014-05-02 18:35:55 +0400387#endif
Stefan Hajnocziae2990c2013-10-08 11:58:31 +0200388
Paolo Bonzini9adea5f2013-04-21 12:01:06 +0200389#ifndef CONFIG_IOVEC
390/* helper function for iov_send_recv() */
391static ssize_t
392readv_writev(int fd, const struct iovec *iov, int iov_cnt, bool do_write)
393{
394 unsigned i = 0;
395 ssize_t ret = 0;
396 while (i < iov_cnt) {
397 ssize_t r = do_write
398 ? write(fd, iov[i].iov_base, iov[i].iov_len)
399 : read(fd, iov[i].iov_base, iov[i].iov_len);
400 if (r > 0) {
401 ret += r;
402 } else if (!r) {
403 break;
404 } else if (errno == EINTR) {
405 continue;
406 } else {
407 /* else it is some "other" error,
408 * only return if there was no data processed. */
409 if (ret == 0) {
410 ret = -1;
411 }
412 break;
413 }
414 i++;
415 }
416 return ret;
417}
418
419ssize_t
420readv(int fd, const struct iovec *iov, int iov_cnt)
421{
422 return readv_writev(fd, iov, iov_cnt, false);
423}
424
425ssize_t
426writev(int fd, const struct iovec *iov, int iov_cnt)
427{
428 return readv_writev(fd, iov, iov_cnt, true);
429}
430#endif