oslib-posix.c

   1 /*
   2  * os-posix-lib.c
   3  *
   4  * Copyright (c) 2003-2008 Fabrice Bellard
   5  * Copyright (c) 2010 Red Hat, Inc.
   6  *
   7  * QEMU library functions on POSIX which are shared between QEMU and
   8  * the QEMU tools.
   9  *
  10  * Permission is hereby granted, free of charge, to any person obtaining a copy
  11  * of this software and associated documentation files (the "Software"), to deal
  12  * in the Software without restriction, including without limitation the rights
  13  * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
  14  * copies of the Software, and to permit persons to whom the Software is
  15  * furnished to do so, subject to the following conditions:
  16  *
  17  * The above copyright notice and this permission notice shall be included in
  18  * all copies or substantial portions of the Software.
  19  *
  20  * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
  21  * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
  22  * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
  23  * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
  24  * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
  25  * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
  26  * THE SOFTWARE.
  27  */
  28
  29 /* The following block of code temporarily renames the daemon() function so the
  30    compiler does not see the warning associated with it in stdlib.h on OSX */
  31 #ifdef __APPLE__
  32 #define daemon qemu_fake_daemon_function
  33 #include <stdlib.h>
  34 #undef daemon
  35 extern int daemon(int, int);
  36 #endif
  37
  38 #if defined(__linux__) && defined(__x86_64__)
  39    /* Use 2 MiB alignment so transparent hugepages can be used by KVM.
  40       Valgrind does not support alignments larger than 1 MiB,
  41       therefore we need special code which handles running on Valgrind. */
  42 #  define QEMU_VMALLOC_ALIGN (512 * 4096)
  43 #  define CONFIG_VALGRIND
  44 #elif defined(__linux__) && defined(__s390x__)
  45    /* Use 1 MiB (segment size) alignment so gmap can be used by KVM. */
  46 #  define QEMU_VMALLOC_ALIGN (256 * 4096)
  47 #else
  48 #  define QEMU_VMALLOC_ALIGN getpagesize()
  49 #endif
  50
  51 #include "config-host.h"
  52 #include "sysemu.h"
  53 #include "trace.h"
  54 #include "qemu_socket.h"
  55
  56 #if defined(CONFIG_VALGRIND)
  57 static int running_on_valgrind = -1;
  58 #else
  59 #  define running_on_valgrind 0
  60 #endif
  61 #ifdef CONFIG_LINUX
  62 #include <sys/syscall.h>
  63 #endif
  64 #ifdef CONFIG_EVENTFD
  65 #include <sys/eventfd.h>
  66 #endif
  67
  68 int qemu_get_thread_id(void)
  69 {
  70 #if defined(__linux__)
  71     return syscall(SYS_gettid);
  72 #else
  73     return getpid();
  74 #endif
  75 }
  76
  77 int qemu_daemon(int nochdir, int noclose)
  78 {
  79     return daemon(nochdir, noclose);
  80 }
  81
  82 void *qemu_oom_check(void *ptr)
  83 {
  84     if (ptr == NULL) {
  85         fprintf(stderr, "Failed to allocate memory: %s\n", strerror(errno));
  86         abort();
  87     }
  88     return ptr;
  89 }
  90
  91 void *qemu_memalign(size_t alignment, size_t size)
  92 {
  93     void *ptr;
  94 #if defined(_POSIX_C_SOURCE) && !defined(__sun__)
  95     int ret;
  96     ret = posix_memalign(&ptr, alignment, size);
  97     if (ret != 0) {
  98         fprintf(stderr, "Failed to allocate %zu B: %s\n",
  99                 size, strerror(ret));
 100         abort();
 101     }
 102 #elif defined(CONFIG_BSD)
 103     ptr = qemu_oom_check(valloc(size));
 104 #else
 105     ptr = qemu_oom_check(memalign(alignment, size));
 106 #endif
 107     trace_qemu_memalign(alignment, size, ptr);
 108     return ptr;
 109 }
 110
 111 /* conflicts with qemu_vmalloc in bsd-user/mmap.c */
 112 #if !defined(CONFIG_BSD_USER)
 113 /* alloc shared memory pages */
 114 void *qemu_vmalloc(size_t size)
 115 {
 116     void *ptr;
 117     size_t align = QEMU_VMALLOC_ALIGN;
 118
 119 #if defined(CONFIG_VALGRIND)
 120     if (running_on_valgrind < 0) {
 121         /* First call, test whether we are running on Valgrind.
 122            This is a substitute for RUNNING_ON_VALGRIND from valgrind.h. */
 123         const char *ld = getenv("LD_PRELOAD");
 124         running_on_valgrind = (ld != NULL && strstr(ld, "vgpreload"));
 125     }
 126 #endif
 127
 128     if (size < align || running_on_valgrind) {
 129         align = getpagesize();
 130     }
 131     ptr = qemu_memalign(align, size);
 132     trace_qemu_vmalloc(size, ptr);
 133     return ptr;
 134 }
 135 #endif
 136
 137 void qemu_vfree(void *ptr)
 138 {
 139     trace_qemu_vfree(ptr);
 140     free(ptr);
 141 }
 142
 143 void socket_set_block(int fd)
 144 {
 145     int f;
 146     f = fcntl(fd, F_GETFL);
 147     fcntl(fd, F_SETFL, f & ~O_NONBLOCK);
 148 }
 149
 150 void socket_set_nonblock(int fd)
 151 {
 152     int f;
 153     f = fcntl(fd, F_GETFL);
 154     fcntl(fd, F_SETFL, f | O_NONBLOCK);
 155 }
 156
 157 void qemu_set_cloexec(int fd)
 158 {
 159     int f;
 160     f = fcntl(fd, F_GETFD);
 161     fcntl(fd, F_SETFD, f | FD_CLOEXEC);
 162 }
 163
 164 /*
 165  * Creates a pipe with FD_CLOEXEC set on both file descriptors
 166  */
 167 int qemu_pipe(int pipefd[2])
 168 {
 169     int ret;
 170
 171 #ifdef CONFIG_PIPE2
 172     ret = pipe2(pipefd, O_CLOEXEC);
 173     if (ret != -1 || errno != ENOSYS) {
 174         return ret;
 175     }
 176 #endif
 177     ret = pipe(pipefd);
 178     if (ret == 0) {
 179         qemu_set_cloexec(pipefd[0]);
 180         qemu_set_cloexec(pipefd[1]);
 181     }
 182
 183     return ret;
 184 }
 185
 186 /*
 187  * Creates an eventfd that looks like a pipe and has EFD_CLOEXEC set.
 188  */
 189 int qemu_eventfd(int fds[2])
 190 {
 191 #ifdef CONFIG_EVENTFD
 192     int ret;
 193
 194     ret = eventfd(0, 0);
 195     if (ret >= 0) {
 196         fds[0] = ret;
 197         fds[1] = dup(ret);
 198         if (fds[1] == -1) {
 199             close(ret);
 200             return -1;
 201         }
 202         qemu_set_cloexec(ret);
 203         qemu_set_cloexec(fds[1]);
 204         return 0;
 205     }
 206     if (errno != ENOSYS) {
 207         return -1;
 208     }
 209 #endif
 210
 211     return qemu_pipe(fds);
 212 }
 213
 214 int qemu_utimens(const char *path, const struct timespec *times)
 215 {
 216     struct timeval tv[2], tv_now;
 217     struct stat st;
 218     int i;
 219 #ifdef CONFIG_UTIMENSAT
 220     int ret;
 221
 222     ret = utimensat(AT_FDCWD, path, times, AT_SYMLINK_NOFOLLOW);
 223     if (ret != -1 || errno != ENOSYS) {
 224         return ret;
 225     }
 226 #endif
 227     /* Fallback: use utimes() instead of utimensat() */
 228
 229     /* happy if special cases */
 230     if (times[0].tv_nsec == UTIME_OMIT && times[1].tv_nsec == UTIME_OMIT) {
 231         return 0;
 232     }
 233     if (times[0].tv_nsec == UTIME_NOW && times[1].tv_nsec == UTIME_NOW) {
 234         return utimes(path, NULL);
 235     }
 236
 237     /* prepare for hard cases */
 238     if (times[0].tv_nsec == UTIME_NOW || times[1].tv_nsec == UTIME_NOW) {
 239         gettimeofday(&tv_now, NULL);
 240     }
 241     if (times[0].tv_nsec == UTIME_OMIT || times[1].tv_nsec == UTIME_OMIT) {
 242         stat(path, &st);
 243     }
 244
 245     for (i = 0; i < 2; i++) {
 246         if (times[i].tv_nsec == UTIME_NOW) {
 247             tv[i].tv_sec = tv_now.tv_sec;
 248             tv[i].tv_usec = tv_now.tv_usec;
 249         } else if (times[i].tv_nsec == UTIME_OMIT) {
 250             tv[i].tv_sec = (i == 0) ? st.st_atime : st.st_mtime;
 251             tv[i].tv_usec = 0;
 252         } else {
 253             tv[i].tv_sec = times[i].tv_sec;
 254             tv[i].tv_usec = times[i].tv_nsec / 1000;
 255         }
 256     }
 257
 258     return utimes(path, &tv[0]);
 259 }