2 * QEMU low level functions
4 * Copyright (c) 2003 Fabrice Bellard
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:
13 * The above copyright notice and this permission notice shall be included in
14 * all copies or substantial portions of the Software.
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
24 #include "qemu/osdep.h"
26 /* Needed early for CONFIG_BSD etc. */
29 #include <sys/statvfs.h>
30 /* See MySQL bug #7156 (http://bugs.mysql.com/bug.php?id=7156) for
31 discussion about Solaris header problems */
32 extern int madvise(caddr_t, size_t, int);
35 #include "qemu-common.h"
36 #include "qemu/cutils.h"
37 #include "qemu/sockets.h"
38 #include "qemu/error-report.h"
39 #include "monitor/monitor.h"
41 static bool fips_enabled = false;
43 static const char *hw_version = QEMU_HW_VERSION;
45 int socket_set_cork(int fd, int v)
47 #if defined(SOL_TCP) && defined(TCP_CORK)
48 return qemu_setsockopt(fd, SOL_TCP, TCP_CORK, &v, sizeof(v));
54 int socket_set_nodelay(int fd)
57 return qemu_setsockopt(fd, IPPROTO_TCP, TCP_NODELAY, &v, sizeof(v));
60 int qemu_madvise(void *addr, size_t len, int advice)
62 if (advice == QEMU_MADV_INVALID) {
66 #if defined(CONFIG_MADVISE)
67 return madvise(addr, len, advice);
68 #elif defined(CONFIG_POSIX_MADVISE)
69 return posix_madvise(addr, len, advice);
76 static int qemu_mprotect__osdep(void *addr, size_t size, int prot)
78 g_assert(!((uintptr_t)addr & ~qemu_real_host_page_mask));
79 g_assert(!(size & ~qemu_real_host_page_mask));
84 if (!VirtualProtect(addr, size, prot, &old_protect)) {
85 error_report("%s: VirtualProtect failed with error code %ld",
86 __func__, GetLastError());
91 if (mprotect(addr, size, prot)) {
92 error_report("%s: mprotect failed: %s", __func__, strerror(errno));
99 int qemu_mprotect_rwx(void *addr, size_t size)
102 return qemu_mprotect__osdep(addr, size, PAGE_EXECUTE_READWRITE);
104 return qemu_mprotect__osdep(addr, size, PROT_READ | PROT_WRITE | PROT_EXEC);
108 int qemu_mprotect_none(void *addr, size_t size)
111 return qemu_mprotect__osdep(addr, size, PAGE_NOACCESS);
113 return qemu_mprotect__osdep(addr, size, PROT_NONE);
119 static int fcntl_op_setlk = -1;
120 static int fcntl_op_getlk = -1;
123 * Dups an fd and sets the flags
125 static int qemu_dup_flags(int fd, int flags)
136 dup_flags = fcntl(ret, F_GETFL);
137 if (dup_flags == -1) {
141 if ((flags & O_SYNC) != (dup_flags & O_SYNC)) {
146 /* Set/unset flags that we can with fcntl */
147 if (fcntl(ret, F_SETFL, flags) == -1) {
151 /* Truncate the file in the cases that open() would truncate it */
152 if (flags & O_TRUNC ||
153 ((flags & (O_CREAT | O_EXCL)) == (O_CREAT | O_EXCL))) {
154 if (ftruncate(ret, 0) == -1) {
173 #ifdef F_DUPFD_CLOEXEC
174 ret = fcntl(fd, F_DUPFD_CLOEXEC, 0);
178 qemu_set_cloexec(ret);
184 static int qemu_parse_fdset(const char *param)
186 return qemu_parse_fd(param);
189 static void qemu_probe_lock_ops(void)
191 if (fcntl_op_setlk == -1) {
196 .l_whence = SEEK_SET,
202 fd = open("/dev/null", O_RDWR);
205 "Failed to open /dev/null for OFD lock probing: %s\n",
207 fcntl_op_setlk = F_SETLK;
208 fcntl_op_getlk = F_GETLK;
211 ret = fcntl(fd, F_OFD_GETLK, &fl);
214 fcntl_op_setlk = F_OFD_SETLK;
215 fcntl_op_getlk = F_OFD_GETLK;
217 fcntl_op_setlk = F_SETLK;
218 fcntl_op_getlk = F_GETLK;
221 fcntl_op_setlk = F_SETLK;
222 fcntl_op_getlk = F_GETLK;
227 bool qemu_has_ofd_lock(void)
229 qemu_probe_lock_ops();
231 return fcntl_op_setlk == F_OFD_SETLK;
237 static int qemu_lock_fcntl(int fd, int64_t start, int64_t len, int fl_type)
241 .l_whence = SEEK_SET,
246 qemu_probe_lock_ops();
248 ret = fcntl(fd, fcntl_op_setlk, &fl);
249 } while (ret == -1 && errno == EINTR);
250 return ret == -1 ? -errno : 0;
253 int qemu_lock_fd(int fd, int64_t start, int64_t len, bool exclusive)
255 return qemu_lock_fcntl(fd, start, len, exclusive ? F_WRLCK : F_RDLCK);
258 int qemu_unlock_fd(int fd, int64_t start, int64_t len)
260 return qemu_lock_fcntl(fd, start, len, F_UNLCK);
263 int qemu_lock_fd_test(int fd, int64_t start, int64_t len, bool exclusive)
267 .l_whence = SEEK_SET,
270 .l_type = exclusive ? F_WRLCK : F_RDLCK,
272 qemu_probe_lock_ops();
273 ret = fcntl(fd, fcntl_op_getlk, &fl);
277 return fl.l_type == F_UNLCK ? 0 : -EAGAIN;
283 * Opens a file with FD_CLOEXEC set
285 int qemu_open(const char *name, int flags, ...)
291 const char *fdset_id_str;
293 /* Attempt dup of fd from fd set */
294 if (strstart(name, "/dev/fdset/", &fdset_id_str)) {
298 fdset_id = qemu_parse_fdset(fdset_id_str);
299 if (fdset_id == -1) {
304 fd = monitor_fdset_get_fd(fdset_id, flags);
310 dupfd = qemu_dup_flags(fd, flags);
315 ret = monitor_fdset_dup_fd_add(fdset_id, dupfd);
326 if (flags & O_CREAT) {
330 mode = va_arg(ap, int);
335 ret = open(name, flags | O_CLOEXEC, mode);
337 ret = open(name, flags, mode);
339 qemu_set_cloexec(ret);
344 if (ret == -1 && errno == EINVAL && (flags & O_DIRECT)) {
345 error_report("file system may not support O_DIRECT");
346 errno = EINVAL; /* in case it was clobbered */
348 #endif /* O_DIRECT */
353 int qemu_close(int fd)
357 /* Close fd that was dup'd from an fdset */
358 fdset_id = monitor_fdset_dup_fd_find(fd);
359 if (fdset_id != -1) {
364 monitor_fdset_dup_fd_remove(fd);
374 * A variant of write(2) which handles partial write.
376 * Return the number of bytes transferred.
377 * Set errno if fewer than `count' bytes are written.
379 * This function don't work with non-blocking fd's.
380 * Any of the possibilities with non-bloking fd's is bad:
381 * - return a short write (then name is wrong)
382 * - busy wait adding (errno == EAGAIN) to the loop
384 ssize_t qemu_write_full(int fd, const void *buf, size_t count)
390 ret = write(fd, buf, count);
406 * Opens a socket with FD_CLOEXEC set
408 int qemu_socket(int domain, int type, int protocol)
413 ret = socket(domain, type | SOCK_CLOEXEC, protocol);
414 if (ret != -1 || errno != EINVAL) {
418 ret = socket(domain, type, protocol);
420 qemu_set_cloexec(ret);
427 * Accept a connection and set FD_CLOEXEC
429 int qemu_accept(int s, struct sockaddr *addr, socklen_t *addrlen)
433 #ifdef CONFIG_ACCEPT4
434 ret = accept4(s, addr, addrlen, SOCK_CLOEXEC);
435 if (ret != -1 || errno != ENOSYS) {
439 ret = accept(s, addr, addrlen);
441 qemu_set_cloexec(ret);
447 void qemu_set_hw_version(const char *version)
449 hw_version = version;
452 const char *qemu_hw_version(void)
457 void fips_set_state(bool requested)
461 FILE *fds = fopen("/proc/sys/crypto/fips_enabled", "r");
463 fips_enabled = (fgetc(fds) == '1');
468 fips_enabled = false;
469 #endif /* __linux__ */
472 fprintf(stderr, "FIPS mode %s (requested %s)\n",
473 (fips_enabled ? "enabled" : "disabled"),
474 (requested ? "enabled" : "disabled"));
478 bool fips_get_state(void)
484 static void socket_cleanup(void)
490 int socket_init(void)
496 ret = WSAStartup(MAKEWORD(2, 2), &Data);
498 err = WSAGetLastError();
499 fprintf(stderr, "WSAStartup: %d\n", err);
502 atexit(socket_cleanup);
509 /* helper function for iov_send_recv() */
511 readv_writev(int fd, const struct iovec *iov, int iov_cnt, bool do_write)
515 while (i < iov_cnt) {
517 ? write(fd, iov[i].iov_base, iov[i].iov_len)
518 : read(fd, iov[i].iov_base, iov[i].iov_len);
523 } else if (errno == EINTR) {
526 /* else it is some "other" error,
527 * only return if there was no data processed. */
539 readv(int fd, const struct iovec *iov, int iov_cnt)
541 return readv_writev(fd, iov, iov_cnt, false);
545 writev(int fd, const struct iovec *iov, int iov_cnt)
547 return readv_writev(fd, iov, iov_cnt, true);