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 #else
  45 #  define QEMU_VMALLOC_ALIGN getpagesize()
  46 #endif
  47
  48 #include "config-host.h"
  49 #include "sysemu.h"
  50 #include "trace.h"
  51 #include "qemu_socket.h"
  52
  53 #if defined(CONFIG_VALGRIND)
  54 static int running_on_valgrind = -1;
  55 #else
  56 #  define running_on_valgrind 0
  57 #endif
  58 #ifdef CONFIG_LINUX
  59 #include <sys/syscall.h>
  60 #endif
  61 #ifdef CONFIG_EVENTFD
  62 #include <sys/eventfd.h>
  63 #endif
  64
  65 int qemu_get_thread_id(void)
  66 {
  67 #if defined(__linux__)
  68     return syscall(SYS_gettid);
  69 #else
  70     return getpid();
  71 #endif
  72 }
  73
  74 int qemu_daemon(int nochdir, int noclose)
  75 {
  76     return daemon(nochdir, noclose);
  77 }
  78
  79 void *qemu_oom_check(void *ptr)
  80 {
  81     if (ptr == NULL) {
  82         fprintf(stderr, "Failed to allocate memory: %s\n", strerror(errno));
  83         abort();
  84     }
  85     return ptr;
  86 }
  87
  88 void *qemu_memalign(size_t alignment, size_t size)
  89 {
  90     void *ptr;
  91 #if defined(_POSIX_C_SOURCE) && !defined(__sun__)
  92     int ret;
  93     ret = posix_memalign(&ptr, alignment, size);
  94     if (ret != 0) {
  95         fprintf(stderr, "Failed to allocate %zu B: %s\n",
  96                 size, strerror(ret));
  97         abort();
  98     }
  99 #elif defined(CONFIG_BSD)
 100     ptr = qemu_oom_check(valloc(size));
 101 #else
 102     ptr = qemu_oom_check(memalign(alignment, size));
 103 #endif
 104     trace_qemu_memalign(alignment, size, ptr);
 105     return ptr;
 106 }
 107
 108 /* alloc shared memory pages */
 109 void *qemu_vmalloc(size_t size)
 110 {
 111     void *ptr;
 112     size_t align = QEMU_VMALLOC_ALIGN;
 113
 114 #if defined(CONFIG_VALGRIND)
 115     if (running_on_valgrind < 0) {
 116         /* First call, test whether we are running on Valgrind.
 117            This is a substitute for RUNNING_ON_VALGRIND from valgrind.h. */
 118         const char *ld = getenv("LD_PRELOAD");
 119         running_on_valgrind = (ld != NULL && strstr(ld, "vgpreload"));
 120     }
 121 #endif
 122
 123     if (size < align || running_on_valgrind) {
 124         align = getpagesize();
 125     }
 126     ptr = qemu_memalign(align, size);
 127     trace_qemu_vmalloc(size, ptr);
 128     return ptr;
 129 }
 130
 131 void qemu_vfree(void *ptr)
 132 {
 133     trace_qemu_vfree(ptr);
 134     free(ptr);
 135 }
 136
 137 void socket_set_block(int fd)
 138 {
 139     int f;
 140     f = fcntl(fd, F_GETFL);
 141     fcntl(fd, F_SETFL, f & ~O_NONBLOCK);
 142 }
 143
 144 void socket_set_nonblock(int fd)
 145 {
 146     int f;
 147     f = fcntl(fd, F_GETFL);
 148     fcntl(fd, F_SETFL, f | O_NONBLOCK);
 149 }
 150
 151 void qemu_set_cloexec(int fd)
 152 {
 153     int f;
 154     f = fcntl(fd, F_GETFD);
 155     fcntl(fd, F_SETFD, f | FD_CLOEXEC);
 156 }
 157
 158 /*
 159  * Creates a pipe with FD_CLOEXEC set on both file descriptors
 160  */
 161 int qemu_pipe(int pipefd[2])
 162 {
 163     int ret;
 164
 165 #ifdef CONFIG_PIPE2
 166     ret = pipe2(pipefd, O_CLOEXEC);
 167     if (ret != -1 || errno != ENOSYS) {
 168         return ret;
 169     }
 170 #endif
 171     ret = pipe(pipefd);
 172     if (ret == 0) {
 173         qemu_set_cloexec(pipefd[0]);
 174         qemu_set_cloexec(pipefd[1]);
 175     }
 176
 177     return ret;
 178 }
 179
 180 /*
 181  * Creates an eventfd that looks like a pipe and has EFD_CLOEXEC set.
 182  */
 183 int qemu_eventfd(int fds[2])
 184 {
 185 #ifdef CONFIG_EVENTFD
 186     int ret;
 187
 188     ret = eventfd(0, 0);
 189     if (ret >= 0) {
 190         fds[0] = ret;
 191         fds[1] = dup(ret);
 192         if (fds[1] == -1) {
 193             close(ret);
 194             return -1;
 195         }
 196         qemu_set_cloexec(ret);
 197         qemu_set_cloexec(fds[1]);
 198         return 0;
 199     }
 200     if (errno != ENOSYS) {
 201         return -1;
 202     }
 203 #endif
 204
 205     return qemu_pipe(fds);
 206 }
 207
 208 int qemu_utimens(const char *path, const struct timespec *times)
 209 {
 210     struct timeval tv[2], tv_now;
 211     struct stat st;
 212     int i;
 213 #ifdef CONFIG_UTIMENSAT
 214     int ret;
 215
 216     ret = utimensat(AT_FDCWD, path, times, AT_SYMLINK_NOFOLLOW);
 217     if (ret != -1 || errno != ENOSYS) {
 218         return ret;
 219     }
 220 #endif
 221     /* Fallback: use utimes() instead of utimensat() */
 222
 223     /* happy if special cases */
 224     if (times[0].tv_nsec == UTIME_OMIT && times[1].tv_nsec == UTIME_OMIT) {
 225         return 0;
 226     }
 227     if (times[0].tv_nsec == UTIME_NOW && times[1].tv_nsec == UTIME_NOW) {
 228         return utimes(path, NULL);
 229     }
 230
 231     /* prepare for hard cases */
 232     if (times[0].tv_nsec == UTIME_NOW || times[1].tv_nsec == UTIME_NOW) {
 233         gettimeofday(&tv_now, NULL);
 234     }
 235     if (times[0].tv_nsec == UTIME_OMIT || times[1].tv_nsec == UTIME_OMIT) {
 236         stat(path, &st);
 237     }
 238
 239     for (i = 0; i < 2; i++) {
 240         if (times[i].tv_nsec == UTIME_NOW) {
 241             tv[i].tv_sec = tv_now.tv_sec;
 242             tv[i].tv_usec = tv_now.tv_usec;
 243         } else if (times[i].tv_nsec == UTIME_OMIT) {
 244             tv[i].tv_sec = (i == 0) ? st.st_atime : st.st_mtime;
 245             tv[i].tv_usec = 0;
 246         } else {
 247             tv[i].tv_sec = times[i].tv_sec;
 248             tv[i].tv_usec = times[i].tv_nsec / 1000;
 249         }
 250     }
 251
 252     return utimes(path, &tv[0]);
 253 }