/***************************************************************************** * RRDtool 1.7.2 Copyright by Tobi Oetiker ***************************************************************************** * rrd_open.c Open an RRD File ***************************************************************************** * $Id$ *****************************************************************************/ #ifdef _WIN32 #include #include /* Defines _WIN32_MAXVER */ #if _WIN32_MAXVER >= 0x0602 /* _WIN32_WINNT_WIN8 */ #include #endif #include #include #include #include #include #endif /* WIN32 */ #include "rrd_tool.h" #include "unused.h" #ifdef HAVE_BROKEN_MS_ASYNC #include #include #endif #ifdef HAVE_LIBRADOS #include "rrd_rados.h" #endif #define MEMBLK 8192 #ifdef _WIN32 #define _LK_UNLCK 0 /* Unlock */ #define _LK_LOCK 1 /* Lock */ #define _LK_NBLCK 2 /* Non-blocking lock */ #define _LK_RLCK 3 /* "Same as _LK_NBLCK" */ #define _LK_NBRLCK 4 /* "Same as _LK_LOCK" */ #define LK_UNLCK _LK_UNLCK #define LK_LOCK _LK_LOCK #define LK_NBLCK _LK_NBLCK #define LK_RLCK _LK_RLCK #define LK_NBRLCK _LK_NBRLCK /* Variables for CreateFileA(). Names of variables are according to * https://docs.microsoft.com/en-us/windows/desktop/api/fileapi/nf-fileapi-createfilea */ DWORD dwDesiredAccess = 0; DWORD dwCreationDisposition = 0; #endif /* DEBUG 2 prints information obtained via mincore(2) */ // #define DEBUG 1 /* do not calculate exact madvise hints but assume 1 page for headers and * set DONTNEED for the rest, which is assumed to be data */ /* Avoid calling madvise on areas that were already hinted. May be beneficial if * your syscalls are very slow */ #ifdef HAVE_MMAP /* the cast to void* is there to avoid this warning seen on ia64 with certain versions of gcc: 'cast increases required alignment of target type' */ #define __rrd_read_mmap(dst, dst_t, cnt) { \ size_t wanted = sizeof(dst_t)*(cnt); \ if (offset + wanted > rrd_file->file_len) { \ rrd_set_error("reached EOF while loading header " #dst); \ goto out_close; \ } \ (dst) = (dst_t*)(void*) (data + offset); \ offset += wanted; \ } #else #define __rrd_read_seq(dst, dst_t, cnt) { \ size_t wanted = sizeof(dst_t)*(cnt); \ size_t got; \ if ((dst = (dst_t*)malloc(wanted)) == NULL) { \ rrd_set_error(#dst " malloc"); \ goto out_close; \ } \ got = read (rrd_simple_file->fd, dst, wanted); \ if (got != wanted) { \ rrd_set_error("short read while reading header " #dst); \ goto out_close; \ } \ offset += got; \ } #endif #ifdef HAVE_LIBRADOS #define __rrd_read_rados(dst, dst_t, cnt) { \ size_t wanted = sizeof(dst_t)*(cnt); \ size_t got; \ if ((dst = (dst_t*)malloc(wanted)) == NULL) { \ rrd_set_error(#dst " malloc"); \ goto out_close; \ } \ got = rrd_rados_read(rrd_file->rados, dst, wanted, offset); \ if (got != wanted) { \ rrd_set_error("short read while reading header " #dst); \ goto out_close; \ } \ offset += got; \ } #endif #if defined(HAVE_LIBRADOS) && defined(HAVE_MMAP) #define __rrd_read(dst, dst_t, cnt) { \ if (rrd_file->rados) \ __rrd_read_rados(dst, dst_t, cnt) \ else \ __rrd_read_mmap(dst, dst_t, cnt) \ } #elif defined(HAVE_LIBRADOS) && !defined(HAVE_MMAP) #define __rrd_read(dst, dst_t, cnt) { \ if (rrd_file->rados) \ __rrd_read_rados(dst, dst_t, cnt) \ else \ __rrd_read_seq(dst, dst_t, cnt) \ } #elif defined(HAVE_MMAP) #define __rrd_read(dst, dst_t, cnt) \ __rrd_read_mmap(dst, dst_t, cnt) #else #define __rrd_read(dst, dst_t, cnt) \ __rrd_read_seq(dst, dst_t, cnt) #endif /* get the address of the start of this page */ #if defined USE_MADVISE || defined HAVE_POSIX_FADVISE #ifndef PAGE_START #define PAGE_START(addr) ((addr)&(~(_page_size-1))) #endif #endif static int rrd_rwlock( rrd_file_t *rrd_file, int writelock); static int close_and_unlock( int fd); /* Open a database file, return its header and an open filehandle, * positioned to the first cdp in the first rra. * In the error path of rrd_open, only rrd_free(&rrd) has to be called * before returning an error. Do not call rrd_close upon failure of rrd_open. * If creating a new file, the parameter rrd must be initialized with * details of the file content. * If opening an existing file, then use rrd must be initialized by * rrd_init(rrd) prior to invoking rrd_open */ rrd_file_t *rrd_open( const char *const file_name, rrd_t *rrd, unsigned rdwr) { unsigned long ui; int flags = 0; int version; #ifdef HAVE_MMAP char *data = MAP_FAILED; #endif off_t offset = 0; struct stat statb; rrd_file_t *rrd_file = NULL; rrd_simple_file_t *rrd_simple_file = NULL; size_t newfile_size = 0; /* Are we creating a new file? */ if (rdwr & RRD_CREAT) { size_t header_len, value_cnt, data_len; header_len = rrd_get_header_size(rrd); value_cnt = 0; for (ui = 0; ui < rrd->stat_head->rra_cnt; ui++) value_cnt += rrd->stat_head->ds_cnt * rrd->rra_def[ui].row_cnt; data_len = sizeof(rrd_value_t) * value_cnt; newfile_size = header_len + data_len; } rrd_file = (rrd_file_t *) malloc(sizeof(rrd_file_t)); if (rrd_file == NULL) { rrd_set_error("allocating rrd_file descriptor for '%s'", file_name); return NULL; } memset(rrd_file, 0, sizeof(rrd_file_t)); rrd_file->rrd = rrd; rrd_file->pvt = malloc(sizeof(rrd_simple_file_t)); if (rrd_file->pvt == NULL) { rrd_set_error("allocating rrd_simple_file for '%s'", file_name); free(rrd_file); return NULL; } memset(rrd_file->pvt, 0, sizeof(rrd_simple_file_t)); rrd_simple_file = (rrd_simple_file_t *) rrd_file->pvt; rrd_simple_file->fd = -1; #ifdef DEBUG if ((rdwr & (RRD_READONLY | RRD_READWRITE)) == (RRD_READONLY | RRD_READWRITE)) { /* Both READONLY and READWRITE were given, which is invalid. */ rrd_set_error("in read/write request mask"); free(rrd_file); return NULL; } #endif #ifdef HAVE_LIBRADOS if (strncmp("ceph//", file_name, 6) == 0) { rrd_file->rados = rrd_rados_open(file_name + 6); if (rrd_file->rados == NULL) goto out_free; if (rdwr & RRD_LOCK) { /* Note: rados read lock is not implemented. See rrd_lock(). */ if (rrd_rwlock(rrd_file, rdwr & RRD_READWRITE) != 0) { rrd_set_error("could not lock RRD"); goto out_close; } } if (rdwr & RRD_CREAT) goto out_done; goto read_check; } #endif #ifdef HAVE_MMAP rrd_simple_file->mm_prot = PROT_READ; rrd_simple_file->mm_flags = 0; #endif if (rdwr & RRD_READONLY) { flags |= O_RDONLY; #ifdef _WIN32 dwDesiredAccess = GENERIC_READ; dwCreationDisposition = OPEN_EXISTING; #endif #ifdef HAVE_MMAP # if !defined(AIX) rrd_simple_file->mm_flags = MAP_PRIVATE; # endif # ifdef MAP_NORESERVE rrd_simple_file->mm_flags |= MAP_NORESERVE; /* readonly, so no swap backing needed */ # endif #endif } else { if (rdwr & RRD_READWRITE) { flags |= O_RDWR; #ifdef _WIN32 dwDesiredAccess = GENERIC_READ | GENERIC_WRITE; dwCreationDisposition = OPEN_EXISTING; #endif #ifdef HAVE_MMAP rrd_simple_file->mm_flags = MAP_SHARED; rrd_simple_file->mm_prot |= PROT_WRITE; #endif } if (rdwr & RRD_CREAT) { flags |= (O_CREAT | O_TRUNC); #ifdef _WIN32 dwDesiredAccess = GENERIC_READ | GENERIC_WRITE; dwCreationDisposition = CREATE_ALWAYS; #endif } if (rdwr & RRD_EXCL) { flags |= O_EXCL; #ifdef _WIN32 dwDesiredAccess = GENERIC_READ | GENERIC_WRITE; dwCreationDisposition = CREATE_NEW; #endif } } if (rdwr & RRD_READAHEAD) { #ifdef MAP_POPULATE rrd_simple_file->mm_flags |= MAP_POPULATE; /* populate ptes and data */ #endif #if defined MAP_NONBLOCK rrd_simple_file->mm_flags |= MAP_NONBLOCK; /* just populate ptes */ #endif } #if defined(_WIN32) && !defined(__CYGWIN__) && !defined(__CYGWIN32__) flags |= O_BINARY; #endif #if defined(_WIN32) /* In Windows we need FILE_SHARE_DELETE, so that the file can be * renamed/replaced later on in rrd_create.c * This is only possible using CreateFileA() first and not using open() alone */ HANDLE handle; handle = CreateFileA(file_name, dwDesiredAccess, FILE_SHARE_READ | FILE_SHARE_WRITE | FILE_SHARE_DELETE, NULL, dwCreationDisposition, FILE_ATTRIBUTE_NORMAL, NULL); if (handle == INVALID_HANDLE_VALUE) { LPVOID lpMsgBuf = NULL; FormatMessage(FORMAT_MESSAGE_ALLOCATE_BUFFER | FORMAT_MESSAGE_FROM_SYSTEM | FORMAT_MESSAGE_IGNORE_INSERTS, NULL, GetLastError(), 0, (LPTSTR) & lpMsgBuf, 0, NULL); rrd_set_error("opening '%s': %s", file_name, (LPTSTR) lpMsgBuf); LocalFree(lpMsgBuf); goto out_free; } if ((rrd_simple_file->fd = _open_osfhandle((intptr_t) handle, flags)) < 0) { rrd_set_error("opening '%s': %s", file_name, rrd_strerror(errno)); goto out_free; } #else if ((rrd_simple_file->fd = open(file_name, flags, 0666)) < 0) { rrd_set_error("opening '%s': %s", file_name, rrd_strerror(errno)); goto out_free; } #endif #ifdef HAVE_MMAP #ifdef HAVE_BROKEN_MS_ASYNC if (rdwr & RRD_READWRITE) { /* some unices, the files mtime does not get updated on memory mapped files, in order to help them, we update the timestamp at this point. The thing happens pretty 'close' to the open call so the chances of a race should be minimal. Maybe ask your vendor to fix your OS ... */ utime(file_name, NULL); } #endif #endif if (rdwr & RRD_LOCK) { if (rrd_rwlock(rrd_file, rdwr & RRD_READWRITE) != 0) { rrd_set_error("could not lock RRD"); goto out_close; } } /* Better try to avoid seeks as much as possible. stat may be heavy but * many concurrent seeks are even worse. */ if (newfile_size == 0 && ((fstat(rrd_simple_file->fd, &statb)) < 0)) { rrd_set_error("fstat '%s': %s", file_name, rrd_strerror(errno)); goto out_close; } if (newfile_size == 0) { rrd_file->file_len = statb.st_size; } else { rrd_file->file_len = newfile_size; #ifdef HAVE_POSIX_FALLOCATE /* man: posix_fallocate() returns zero on success, * or an error number on failure. Note that errno is not set. */ int fret = posix_fallocate(rrd_simple_file->fd, 0, newfile_size); if (fret) { rrd_set_error("posix_fallocate '%s': %s", file_name, rrd_strerror(fret)); goto out_close; } else { goto no_lseek_necessary; } #endif if (lseek(rrd_simple_file->fd, newfile_size - 1, SEEK_SET) == -1) { rrd_set_error("lseek '%s': %s", file_name, rrd_strerror(errno)); goto out_close; } if (write(rrd_simple_file->fd, "\0", 1) == -1) { /* poke */ rrd_set_error("write '%s': %s", file_name, rrd_strerror(errno)); goto out_close; } if (lseek(rrd_simple_file->fd, 0, SEEK_SET) == -1) { rrd_set_error("lseek '%s': %s", file_name, rrd_strerror(errno)); goto out_close; } } #ifdef HAVE_POSIX_FALLOCATE no_lseek_necessary: #endif #ifdef HAVE_MMAP #ifndef HAVE_POSIX_FALLOCATE /* force allocating the file on the underlying filesystem to prevent any * future bus error when the filesystem is full and attempting to write * trough the file mapping. Filling the file using memset on the file * mapping can also lead some bus error, so we use the old fashioned * write(). */ if (rdwr & RRD_CREAT) { char buf[4096]; unsigned i; memset(buf, DNAN, sizeof buf); lseek(rrd_simple_file->fd, offset, SEEK_SET); for (i = 0; i < (newfile_size - 1) / sizeof buf; ++i) { if (write(rrd_simple_file->fd, buf, sizeof buf) == -1) { rrd_set_error("write '%s': %s", file_name, rrd_strerror(errno)); goto out_close; } } if (write(rrd_simple_file->fd, buf, (newfile_size - 1) % sizeof buf) == -1) { rrd_set_error("write '%s': %s", file_name, rrd_strerror(errno)); goto out_close; } lseek(rrd_simple_file->fd, 0, SEEK_SET); } #endif data = mmap(0, rrd_file->file_len, rrd_simple_file->mm_prot, rrd_simple_file->mm_flags, rrd_simple_file->fd, offset); /* lets see if the first read worked */ if (data == MAP_FAILED) { rrd_set_error("mmaping file '%s': %s", file_name, rrd_strerror(errno)); goto out_close; } rrd->__mmap_start = data; rrd->__mmap_size = rrd_file->file_len; rrd_simple_file->file_start = data; #endif if (rdwr & RRD_CREAT) goto out_done; if (rdwr & RRD_READAHEAD) { /* If perfect READAHEAD is not achieved for whatever reason, caller will not thank us for advising the kernel of RANDOM access below. */ rdwr |= RRD_COPY; } /* In general we need no read-ahead when dealing with rrd_files. When we stop reading, it is highly unlikely that we start up again. In this manner we actually save time and disk access (and buffer cache). Thanks to Dave Plonka for the Idea of using POSIX_FADV_RANDOM here. */ #ifdef USE_MADVISE if (rdwr & RRD_COPY) { /* We will read everything in a moment (copying) */ madvise(data, rrd_file->file_len, MADV_SEQUENTIAL); } else { /* We do not need to read anything in for the moment */ madvise(data, rrd_file->file_len, MADV_RANDOM); } #endif #if !defined(HAVE_MMAP) && defined(HAVE_POSIX_FADVISE) if (rdwr & RRD_COPY) { posix_fadvise(rrd_simple_file->fd, 0, 0, POSIX_FADV_SEQUENTIAL); } else { posix_fadvise(rrd_simple_file->fd, 0, 0, POSIX_FADV_RANDOM); } #endif #ifdef HAVE_LIBRADOS read_check: #endif __rrd_read(rrd->stat_head, stat_head_t, 1); /* lets do some test if we are on track ... */ if (memcmp(rrd->stat_head->cookie, RRD_COOKIE, sizeof(RRD_COOKIE)) != 0) { rrd_set_error("'%s' is not an RRD file", file_name); goto out_close; } if (rrd->stat_head->float_cookie != FLOAT_COOKIE) { rrd_set_error("This RRD was created on another architecture"); goto out_close; } version = atoi(rrd->stat_head->version); if (version > atoi(RRD_VERSION5)) { rrd_set_error("can't handle RRD file version %s", rrd->stat_head->version); goto out_close; } __rrd_read(rrd->ds_def, ds_def_t, rrd->stat_head->ds_cnt); __rrd_read(rrd->rra_def, rra_def_t, rrd->stat_head->rra_cnt); /* handle different format for the live_head */ if (version < 3) { rrd->live_head = (live_head_t *) malloc(sizeof(live_head_t)); if (rrd->live_head == NULL) { rrd_set_error("live_head_t malloc"); goto out_close; } __rrd_read(rrd->legacy_last_up, time_t, 1); rrd->live_head->last_up = *rrd->legacy_last_up; rrd->live_head->last_up_usec = 0; } else { __rrd_read(rrd->live_head, live_head_t, 1); } __rrd_read(rrd->pdp_prep, pdp_prep_t, rrd->stat_head->ds_cnt); __rrd_read(rrd->cdp_prep, cdp_prep_t, rrd->stat_head->rra_cnt * rrd->stat_head->ds_cnt); __rrd_read(rrd->rra_ptr, rra_ptr_t, rrd->stat_head->rra_cnt); rrd_file->header_len = offset; rrd_file->pos = offset; #if defined(HAVE_MMAP) && defined(USE_MADVISE) if (data != MAP_FAILED) { /* MADV_SEQUENTIAL mentions drop-behind. Override it for the header * now we've read it, in case anyone implemented drop-behind. * * Do *not* fall back to fadvise() for !HAVE_MMAP. In that case, * we've copied the header and will not read it again. Doing e.g. * FADV_NORMAL on Linux (4.12) on *any* region would negate the * effect of previous FADV_SEQUENTIAL. */ madvise(data, sysconf(_SC_PAGESIZE), MADV_NORMAL); madvise(data, sysconf(_SC_PAGESIZE), MADV_WILLNEED); } #endif { unsigned long row_cnt = 0; for (ui = 0; ui < rrd->stat_head->rra_cnt; ui++) row_cnt += rrd->rra_def[ui].row_cnt; size_t correct_len = rrd_file->header_len + sizeof(rrd_value_t) * row_cnt * rrd->stat_head->ds_cnt; #ifdef HAVE_LIBRADOS /* skip length checking for rados file */ if (rrd_file->rados) { rrd_file->file_len = correct_len; } #endif if (correct_len > rrd_file->file_len) { rrd_set_error("'%s' is too small (should be %ld bytes)", file_name, (long long) correct_len); goto out_close; } if (rdwr & RRD_READVALUES) { __rrd_read(rrd->rrd_value, rrd_value_t, row_cnt * rrd->stat_head->ds_cnt); if (rrd_seek(rrd_file, rrd_file->header_len, SEEK_SET) != 0) goto out_close; } } out_done: return (rrd_file); out_close: #ifdef HAVE_MMAP if (data != MAP_FAILED) munmap(data, rrd_file->file_len); #endif #ifdef HAVE_LIBRADOS if (rrd_file->rados) rrd_rados_close(rrd_file->rados); #endif if (rrd_simple_file->fd >= 0) { /* keep the original error */ char *e = strdup(rrd_get_error()); close_and_unlock(rrd_simple_file->fd); if (e) { rrd_set_error(e); free(e); } else rrd_set_error("error message was lost (out of memory)"); } out_free: free(rrd_file->pvt); free(rrd_file); return NULL; } #if defined DEBUG && DEBUG > 1 /* Print list of in-core pages of a the current rrd_file. */ static void mincore_print( rrd_file_t *rrd_file, char *mark) { rrd_simple_file_t *rrd_simple_file; rrd_simple_file = (rrd_simple_file_t *) rrd_file->pvt; #ifdef HAVE_MMAP /* pretty print blocks in core */ size_t off; unsigned char *vec; ssize_t _page_size = sysconf(_SC_PAGESIZE); off = rrd_file->file_len + ((rrd_file->file_len + _page_size - 1) / _page_size); vec = malloc(off); if (vec != NULL) { memset(vec, 0, off); if (mincore(rrd_simple_file->file_start, rrd_file->file_len, vec) == 0) { int prev; unsigned is_in = 0, was_in = 0; for (off = 0, prev = 0; off < rrd_file->file_len; ++off) { is_in = vec[off] & 1; /* if lsb set then is core resident */ if (off == 0) was_in = is_in; if (was_in != is_in) { fprintf(stderr, "%s: %sin core: %p len %ld\n", mark, was_in ? "" : "not ", vec + prev, off - prev); was_in = is_in; prev = off; } } fprintf(stderr, "%s: %sin core: %p len %ld\n", mark, was_in ? "" : "not ", vec + prev, off - prev); } else fprintf(stderr, "mincore: %s", rrd_strerror(errno)); } #else fprintf(stderr, "sorry mincore only works with mmap"); #endif } #endif /* defined DEBUG && DEBUG > 1 */ /* * get exclusive lock to whole file. * lock gets removed when we close the file * * returns 0 on success */ int rrd_lock( rrd_file_t *rrd_file) { return rrd_rwlock(rrd_file, 1); } #if defined(_WIN32) && !defined(__CYGWIN__) && !defined(__CYGWIN32__) #define USE_WINDOWS_LOCK 1 #endif #ifdef USE_WINDOWS_LOCK static int rrd_windows_lock( int fd) { int ret; long pos; /* * _locking() is relative to fd position. * We need to consistently lock bytes starting from 0, * so we can successfully unlock on close. * * Note rrd_lock() API doesn't set a specific error message. * Knowing that rrd_lock() (or even rrd_open()) failed should * be specific enough, if someone manages to invoke rrdtool * on something silly like a named pipe or COM1. */ pos = tell(fd); if (pos < 0) return -1; if (lseek(fd, 0, SEEK_SET) < 0) return -1; while (1) { ret = _locking(fd, _LK_NBLCK, LONG_MAX); if (ret == 0) break; /* success */ if (errno != EACCES) break; /* failure */ /* EACCES: someone else has the lock. */ /* * Wait 0.01 seconds before trying again. _locking() * with _LK_LOCK would work similarly but waits 1 second * between tries, which seems less desirable. */ Sleep(10); } /* restore saved fd position */ if (lseek(fd, pos, SEEK_SET) < 0) return -1; return ret; } #endif static int close_and_unlock( int fd) { int ret = 0; #ifdef USE_WINDOWS_LOCK /* * "If a process closes a file that has outstanding locks, the locks are * unlocked by the operating system. However, the time it takes for the * operating system to unlock these locks depends upon available system * resources. Therefore, it is recommended that your process explicitly * unlock all files it has locked when it terminates." (?!) */ if (lseek(fd, 0, SEEK_SET) < 0) { rrd_set_error("lseek: %s", rrd_strerror(errno)); ret = -1; goto out_close; } ret = _locking(fd, LK_UNLCK, LONG_MAX); if (ret != 0 && errno == EACCES) /* fd was not locked - this is entirely possible, ignore the error */ ret = 0; if (ret != 0) rrd_set_error("unlock file: %s", rrd_strerror(errno)); out_close: #endif if (close(fd) != 0) { ret = -1; rrd_set_error("closing file: %s", rrd_strerror(errno)); } return ret; } static int rrd_rwlock( rrd_file_t *rrd_file, int writelock) { #ifdef DISABLE_FLOCK (void) rrd_file; return 0; #else #ifdef HAVE_LIBRADOS if (rrd_file->rados) { /* * No read lock on rados. It would be complicated by the * use of a short finite lock duration in rrd_rados_lock(). * Also rados does not provide blocking locks. * * Rados users may use snapshots if they need to * e.g. obtain a consistent backup. */ if (writelock) return rrd_rados_lock(rrd_file->rados); else return 0; } #endif int rcstat; rrd_simple_file_t *rrd_simple_file; rrd_simple_file = (rrd_simple_file_t *) rrd_file->pvt; #ifdef USE_WINDOWS_LOCK /* _locking() does not support read locks; we always take a write lock */ rcstat = rrd_windows_lock(rrd_simple_file->fd); /* Silence unused parameter compiler warning */ (void) writelock; #else { struct flock lock; lock.l_type = writelock ? F_WRLCK : /* exclusive write lock or */ F_RDLCK; /* shared read lock */ lock.l_len = 0; /* whole file */ lock.l_start = 0; /* start of file */ lock.l_whence = SEEK_SET; /* end of file */ rcstat = fcntl(rrd_simple_file->fd, F_SETLK, &lock); } #endif return (rcstat); #endif } /* drop cache except for the header and the active pages */ void rrd_dontneed( rrd_file_t *rrd_file, rrd_t *rrd) { rrd_simple_file_t *rrd_simple_file; #if defined USE_MADVISE || defined HAVE_POSIX_FADVISE size_t dontneed_start; size_t rra_start; size_t active_block; size_t i; ssize_t _page_size = sysconf(_SC_PAGESIZE); if (rrd_file == NULL) { #if defined DEBUG && DEBUG fprintf(stderr, "rrd_dontneed: Argument 'rrd_file' is NULL.\n"); #endif return; } rrd_simple_file = (rrd_simple_file_t *) rrd_file->pvt; #if defined DEBUG && DEBUG > 1 mincore_print(rrd_file, "before"); #endif /* ignoring errors from RRDs that are smaller then the file_len+rounding */ rra_start = rrd_file->header_len; dontneed_start = PAGE_START(rra_start) + _page_size; for (i = 0; i < rrd->stat_head->rra_cnt; ++i) { active_block = PAGE_START(rra_start + rrd->rra_ptr[i].cur_row * rrd->stat_head->ds_cnt * sizeof(rrd_value_t)); if (active_block > dontneed_start) { #ifdef USE_MADVISE madvise(rrd_simple_file->file_start + dontneed_start, active_block - dontneed_start - 1, MADV_DONTNEED); #else #ifdef HAVE_POSIX_FADVISE /* in linux at least only fadvise DONTNEED seems to purge pages from cache */ posix_fadvise(rrd_simple_file->fd, dontneed_start, active_block - dontneed_start - 1, POSIX_FADV_DONTNEED); #endif #endif } dontneed_start = active_block; /* do not release 'hot' block if update for this RAA will occur * within 10 minutes */ if (rrd->stat_head->pdp_step * rrd->rra_def[i].pdp_cnt - rrd->live_head->last_up % (rrd->stat_head->pdp_step * rrd->rra_def[i].pdp_cnt) < 10 * 60) { dontneed_start += _page_size; } rra_start += rrd->rra_def[i].row_cnt * rrd->stat_head->ds_cnt * sizeof(rrd_value_t); } if (dontneed_start < rrd_file->file_len) { #ifdef USE_MADVISE madvise(rrd_simple_file->file_start + dontneed_start, rrd_file->file_len - dontneed_start, MADV_DONTNEED); #else #ifdef HAVE_POSIX_FADVISE posix_fadvise(rrd_simple_file->fd, dontneed_start, rrd_file->file_len - dontneed_start, POSIX_FADV_DONTNEED); #endif #endif } #if defined DEBUG && DEBUG > 1 mincore_print(rrd_file, "after"); #endif #else /* #if defined USE_MADVISE || defined HAVE_POSIX_FADVISE */ /* Silence compiler warnings about unused variables and parameters */ (void) rrd_simple_file; (void) rrd_file; (void) rrd; #endif /* without madvise and posix_fadvise it does not make much sense todo anything */ } int rrd_close( rrd_file_t *rrd_file) { rrd_simple_file_t *rrd_simple_file; rrd_simple_file = (rrd_simple_file_t *) rrd_file->pvt; int ret = 0; #ifdef HAVE_LIBRADOS if (rrd_file->rados) { if (rrd_rados_close(rrd_file->rados) != 0) ret = -1; } #endif #ifdef HAVE_MMAP if (rrd_simple_file->file_start != NULL) { if (munmap(rrd_simple_file->file_start, rrd_file->file_len) != 0) { ret = -1; rrd_set_error("munmap rrd_file: %s", rrd_strerror(errno)); } } #endif if (rrd_simple_file->fd >= 0) { if (close_and_unlock(rrd_simple_file->fd) != 0) ret = -1; } free(rrd_file->pvt); free(rrd_file); return ret; } /* Set position of rrd_file. */ off_t rrd_seek( rrd_file_t *rrd_file, off_t off, int whence) { #ifdef HAVE_LIBRADOS /* no seek for rados */ if (rrd_file->rados) { rrd_file->pos = off; return 0; } #endif off_t ret = 0; #ifndef HAVE_MMAP rrd_simple_file_t *rrd_simple_file; rrd_simple_file = (rrd_simple_file_t *) rrd_file->pvt; #endif #ifdef HAVE_MMAP if (whence == SEEK_SET) rrd_file->pos = off; else if (whence == SEEK_CUR) rrd_file->pos += off; else if (whence == SEEK_END) rrd_file->pos = rrd_file->file_len + off; #else ret = lseek(rrd_simple_file->fd, off, whence); if (ret < 0) rrd_set_error("lseek: %s", rrd_strerror(errno)); rrd_file->pos = ret; #endif /* mimic fseek, which returns 0 upon success */ return ret < 0; /*XXX: or just ret to mimic lseek */ } /* Get current position in rrd_file. */ off_t rrd_tell( rrd_file_t *rrd_file) { return rrd_file->pos; } /* Read count bytes into buffer buf, starting at rrd_file->pos. * Returns the number of bytes read or <0 on error. */ ssize_t rrd_read( rrd_file_t *rrd_file, void *buf, size_t count) { #ifdef HAVE_LIBRADOS if (rrd_file->rados) { ssize_t ret = rrd_rados_read(rrd_file->rados, buf, count, rrd_file->pos); if (ret > 0) rrd_file->pos += ret; return ret; } #endif rrd_simple_file_t *rrd_simple_file = (rrd_simple_file_t *) rrd_file->pvt; #ifdef HAVE_MMAP size_t _cnt = count; ssize_t _surplus; if (rrd_file->pos > rrd_file->file_len || _cnt == 0) /* EOF */ return 0; if (buf == NULL) return -1; /* EINVAL */ _surplus = rrd_file->pos + _cnt - rrd_file->file_len; if (_surplus > 0) { /* short read */ _cnt -= _surplus; } if (_cnt == 0) return 0; /* EOF */ buf = memcpy(buf, rrd_simple_file->file_start + rrd_file->pos, _cnt); rrd_file->pos += _cnt; /* mimic read() semantics */ return _cnt; #else ssize_t ret; ret = read(rrd_simple_file->fd, buf, count); if (ret > 0) rrd_file->pos += ret; /* mimic read() semantics */ return ret; #endif } /* Write count bytes from buffer buf to the current position * rrd_file->pos of rrd_simple_file->fd. * Returns the number of bytes written or <0 on error. */ ssize_t rrd_write( rrd_file_t *rrd_file, const void *buf, size_t count) { #ifdef HAVE_LIBRADOS if (rrd_file->rados) { size_t ret = rrd_rados_write(rrd_file->rados, buf, count, rrd_file->pos); if (ret > 0) rrd_file->pos += count; return ret; } #endif rrd_simple_file_t *rrd_simple_file = (rrd_simple_file_t *) rrd_file->pvt; #ifdef HAVE_MMAP size_t old_size = rrd_file->file_len; if (count == 0) return 0; if (buf == NULL) return -1; /* EINVAL */ if ((rrd_file->pos + count) > old_size) { rrd_set_error ("attempting to write beyond end of file (%ld + %ld > %ld)", rrd_file->pos, count, old_size); return -1; } /* can't use memcpy since the areas overlap when tuning */ memmove(rrd_simple_file->file_start + rrd_file->pos, buf, count); rrd_file->pos += count; return count; /* mimic write() semantics */ #else ssize_t _sz = write(rrd_simple_file->fd, buf, count); if (_sz > 0) rrd_file->pos += _sz; return _sz; #endif } /* this is a leftover from the old days, it serves no purpose and is therefore turned into a no-op */ void rrd_flush( rrd_file_t UNUSED(*rrd_file)) { } /* Initialize RRD header. */ void rrd_init( rrd_t *rrd) { rrd->stat_head = NULL; rrd->ds_def = NULL; rrd->rra_def = NULL; rrd->live_head = NULL; rrd->legacy_last_up = NULL; rrd->rra_ptr = NULL; rrd->pdp_prep = NULL; rrd->cdp_prep = NULL; rrd->rrd_value = NULL; rrd->__mmap_start = NULL; rrd->__mmap_size = 0; } /* free RRD data, act correctly, regardless of mmap'ped or malloc'd memory. */ static void free_rrd_ptr_if_not_mmapped( void *m, const rrd_t *rrd) { if (m == NULL) return; if (rrd == NULL || rrd->__mmap_start == NULL) { free(m); return; } /* is this ALWAYS correct on all supported platforms ??? */ long ofs = (char *) m - (char *) rrd->__mmap_start; if (ofs < rrd->__mmap_size) { // DO NOT FREE, this memory is mmapped!! return; } free(m); } void rrd_free( rrd_t *rrd) { if (rrd == NULL) return; free_rrd_ptr_if_not_mmapped(rrd->live_head, rrd); rrd->live_head = NULL; free_rrd_ptr_if_not_mmapped(rrd->stat_head, rrd); rrd->stat_head = NULL; free_rrd_ptr_if_not_mmapped(rrd->ds_def, rrd); rrd->ds_def = NULL; free_rrd_ptr_if_not_mmapped(rrd->rra_def, rrd); rrd->rra_def = NULL; free_rrd_ptr_if_not_mmapped(rrd->rra_ptr, rrd); rrd->rra_ptr = NULL; free_rrd_ptr_if_not_mmapped(rrd->pdp_prep, rrd); rrd->pdp_prep = NULL; free_rrd_ptr_if_not_mmapped(rrd->cdp_prep, rrd); rrd->cdp_prep = NULL; free_rrd_ptr_if_not_mmapped(rrd->rrd_value, rrd); rrd->rrd_value = NULL; } /* routine used by external libraries to free memory allocated by * rrd library */ void rrd_freemem( void *mem) { free(mem); } /* * rra_update informs us about the RRAs being updated * The low level storage API may use this information for * aligning RRAs within stripes, or other performance enhancements */ void rrd_notify_row( rrd_file_t UNUSED(*rrd_file), int UNUSED(rra_idx), unsigned long UNUSED(rra_row), time_t UNUSED(rra_time)) { } /* * This function is called when creating a new RRD * The storage implementation can use this opportunity to select * a sensible starting row within the file. * The default implementation is random, to ensure that all RRAs * don't change to a new disk block at the same time */ unsigned long rrd_select_initial_row( rrd_file_t UNUSED(*rrd_file), int UNUSED(rra_idx), rra_def_t *rra) { return rrd_random() % rra->row_cnt; }