/* Copyright 2016, Google Inc. * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions are * met: * * * Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * * Redistributions in binary form must reproduce the above * copyright notice, this list of conditions and the following * disclaimer in the documentation and/or other materials provided * with the distribution. * * Neither the name of Google Inc. nor the names of its * contributors may be used to endorse or promote products derived * from this software without specific prior written permission. * * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. * * Alternatively, this software may be distributed under the terms of the * GNU General Public License ("GPL") version 2 as published by the Free * Software Foundation. * * file_lock.c: Implementation for a binary semaphore using a file lock. * * Warning: This relies on flock() which is known to be broken on NFS. * * The file will remain persistent once the lock has been used. Unfortunately, * unlinking the file can introduce a race condition so we leave the file * in place. * * The current process's PID will be written to the file for debug purposes. */ #include #include #include #include #include #include #include #include #include #include #include #include #include "flash.h" #include "ipc_lock.h" #define SLEEP_INTERVAL_MS 50 static void msecs_to_timespec(int msecs, struct timespec *tmspec) { tmspec->tv_sec = msecs / 1000; tmspec->tv_nsec = (msecs % 1000) * 1000 * 1000; } static int lock_is_held(struct ipc_lock *lock) { return lock->is_held; } static int test_dir(const char *path) { struct stat s; if (lstat(path, &s) < 0) { msg_gerr("Cannot stat %s.\n", path); return -1; } if (!S_ISDIR(s.st_mode)) { msg_gerr("%s is not a directory.\n", path); return -1; } return 0; } static int file_lock_try_open_or_create(const char *dir, struct ipc_lock *lock) { char path[PATH_MAX]; if (test_dir(dir)) return -1; if (snprintf(path, sizeof(path), "%s/%s", dir, lock->filename) < 0) return -1; /* 0666: User/Group/Others: read, write. */ mode_t permissions = S_IRUSR | S_IWUSR | S_IRGRP | S_IWGRP | S_IROTH | S_IWOTH; lock->fd = open(path, O_RDWR | O_CREAT, permissions); if (lock->fd < 0) { msg_gerr("Cannot open lockfile %s\n", path); return -1; } /* Ignore potential fchmod() failures since we might not be the owner. */ if (fchmod(lock->fd, permissions) == -1) msg_gdbg("Cannot change the permissions of lockfile %s\n", path); msg_gdbg("Opened file lock \"%s\"\n", path); return 0; } static int file_lock_open_or_create(struct ipc_lock *lock) { const char *dirs[] = { #ifndef __ANDROID__ // Default modern Linux lock path. "/run/lock", // Fallback to temporary directory. "/tmp", #else // flashrom called as a subprocess with its own SELinux context. "/data/vendor/flashrom/tmp", // Same as above but for case when there is no tmpfs. "/data/vendor/flashrom", // flashrom called from the console/shell. Comes last as a fallback. "/data/local/tmp", #endif }; if (file_lock_try_open_or_create(dirs[0], lock) == 0) return 0; for (size_t i = 1; i < ARRAY_SIZE(dirs); ++i) { msg_gwarn("Trying fallback directory: %s\n", dirs[i]); if (file_lock_try_open_or_create(dirs[i], lock) == 0) return 0; } msg_gerr("Failed to find usable directory for file lock\n"); return -1; } static int file_lock_get(struct ipc_lock *lock, int timeout_msecs) { int remaining_msecs; struct timespec sleep_interval; int ret; if (timeout_msecs == 0) return flock(lock->fd, LOCK_EX | LOCK_NB); if (timeout_msecs < 0) remaining_msecs = SLEEP_INTERVAL_MS; else remaining_msecs = timeout_msecs; while ((ret = flock(lock->fd, LOCK_EX | LOCK_NB)) != 0) { struct timespec rem; if (errno != EWOULDBLOCK) { msg_gerr("Error obtaining lock\n"); return -1; } msecs_to_timespec(MIN(remaining_msecs, SLEEP_INTERVAL_MS), &sleep_interval); while (nanosleep(&sleep_interval, &rem) != 0) { if (errno == EINTR) { sleep_interval = rem; continue; } msg_gerr("nanosleep() failed\n"); return -1; } if (timeout_msecs < 0) continue; remaining_msecs -= SLEEP_INTERVAL_MS; if (remaining_msecs < 0) break; } if (ret != 0) { msg_gerr("Timed out waiting for file lock.\n"); return -1; } return 0; } static int file_lock_write_pid(struct ipc_lock *lock) { ssize_t len; /* PIDs are usually 5 digits, but we'll reserve enough room for a value of 2^32 (10 digits) out of paranoia. */ char pid_str[11]; if (ftruncate(lock->fd, 0) < 0) { msg_gerr("Cannot truncate lockfile\n"); return -1; } snprintf(pid_str, sizeof(pid_str), "%lu", (unsigned long)getpid()); len = write(lock->fd, pid_str, strlen(pid_str)); if (len < 0) { msg_gerr("Cannot write PID to lockfile\n"); return -1; } return 0; } static void file_lock_release(struct ipc_lock *lock) { if (flock(lock->fd, LOCK_UN) < 0) msg_gerr("Cannot release lock\n"); if (close(lock->fd) < 0) msg_gerr("Cannot close lockfile\n"); } /* * timeout <0 = no timeout (try forever) * timeout 0 = do not wait (return immediately) * timeout >0 = wait up to $timeout milliseconds * * returns 0 to indicate lock acquired * returns >0 to indicate lock was already held * returns <0 to indicate failed to acquire lock */ int acquire_lock(struct ipc_lock *lock, int timeout_msecs) { /* check if it is already held */ if (lock_is_held(lock)) return 1; if (file_lock_open_or_create(lock)) return -1; if (file_lock_get(lock, timeout_msecs)) { lock->is_held = 0; close(lock->fd); return -1; } else { lock->is_held = 1; } /* * Write PID to lockfile for debug purposes. Failure to write to * the file should not be considered fatal. There might be something * bad happening with the filesystem, but the lock has already been * obtained and we may need our tools for diagnostics and repairs * so we should continue anyway. */ file_lock_write_pid(lock); return 0; } /* * returns 0 if lock was released successfully * returns -1 if lock had not been held before the call */ int release_lock(struct ipc_lock *lock) { if (lock_is_held(lock)) { file_lock_release(lock); lock->is_held = 0; return 0; } msg_ginfo("%s called but lock was not held on %s.\n", __func__, lock->filename); return -1; }