X-Git-Url: http://www.chiark.greenend.org.uk/ucgi/~ianmdlvl/git?p=elogind.git;a=blobdiff_plain;f=src%2Fjournal%2Fjournal-file.c;h=8cbdbb9f9634557119011ec7c4ba1057c094ad43;hp=3cb8f79947489948c083b6563072818bdf259385;hb=1fc605b0e130149a44abfa38c33f4535cfe548ea;hpb=507f22bd0172bff5e5d98145b1419bd472a2c57f diff --git a/src/journal/journal-file.c b/src/journal/journal-file.c index 3cb8f7994..8cbdbb9f9 100644 --- a/src/journal/journal-file.c +++ b/src/journal/journal-file.c @@ -26,10 +26,7 @@ #include #include #include - -#ifdef HAVE_XATTR -#include -#endif +#include #include "journal-def.h" #include "journal-file.h" @@ -68,7 +65,10 @@ /* How many entries to keep in the entry array chain cache at max */ #define CHAIN_CACHE_MAX 20 -int journal_file_set_online(JournalFile *f) { +/* How much to increase the journal file size at once each time we allocate something new. */ +#define FILE_SIZE_INCREASE (8ULL*1024ULL*1024ULL) /* 8MB */ + +static int journal_file_set_online(JournalFile *f) { assert(f); if (!f->writable) @@ -130,17 +130,15 @@ void journal_file_close(JournalFile *f) { if (f->header) munmap(f->header, PAGE_ALIGN(sizeof(Header))); - if (f->fd >= 0) - close_nointr_nofail(f->fd); - + safe_close(f->fd); free(f->path); if (f->mmap) mmap_cache_unref(f->mmap); - hashmap_free_free(f->chain_cache); + ordered_hashmap_free_free(f->chain_cache); -#ifdef HAVE_XZ +#if defined(HAVE_XZ) || defined(HAVE_LZ4) free(f->compress_buffer); #endif @@ -160,21 +158,21 @@ void journal_file_close(JournalFile *f) { } static int journal_file_init_header(JournalFile *f, JournalFile *template) { - Header h; + Header h = {}; ssize_t k; int r; assert(f); - zero(h); memcpy(h.signature, HEADER_SIGNATURE, 8); h.header_size = htole64(ALIGN64(sizeof(h))); - h.incompatible_flags = - htole32(f->compress ? HEADER_INCOMPATIBLE_COMPRESSED : 0); + h.incompatible_flags |= htole32( + f->compress_xz * HEADER_INCOMPATIBLE_COMPRESSED_XZ | + f->compress_lz4 * HEADER_INCOMPATIBLE_COMPRESSED_LZ4); - h.compatible_flags = - htole32(f->seal ? HEADER_COMPATIBLE_SEALED : 0); + h.compatible_flags = htole32( + f->seal * HEADER_COMPATIBLE_SEALED); r = sd_id128_randomize(&h.file_id); if (r < 0) @@ -218,13 +216,14 @@ static int journal_file_refresh_header(JournalFile *f) { journal_file_set_online(f); /* Sync the online state to disk */ - msync(f->header, PAGE_ALIGN(sizeof(Header)), MS_SYNC); - fdatasync(f->fd); + fsync(f->fd); return 0; } static int journal_file_verify_header(JournalFile *f) { + uint32_t flags; + assert(f); if (memcmp(f->header->signature, HEADER_SIGNATURE, 8)) @@ -232,24 +231,30 @@ static int journal_file_verify_header(JournalFile *f) { /* In both read and write mode we refuse to open files with * incompatible flags we don't know */ -#ifdef HAVE_XZ - if ((le32toh(f->header->incompatible_flags) & ~HEADER_INCOMPATIBLE_COMPRESSED) != 0) - return -EPROTONOSUPPORT; -#else - if (f->header->incompatible_flags != 0) + flags = le32toh(f->header->incompatible_flags); + if (flags & ~HEADER_INCOMPATIBLE_SUPPORTED) { + if (flags & ~HEADER_INCOMPATIBLE_ANY) + log_debug("Journal file %s has unknown incompatible flags %"PRIx32, + f->path, flags & ~HEADER_INCOMPATIBLE_ANY); + flags = (flags & HEADER_INCOMPATIBLE_ANY) & ~HEADER_INCOMPATIBLE_SUPPORTED; + if (flags) + log_debug("Journal file %s uses incompatible flags %"PRIx32 + " disabled at compilation time.", f->path, flags); return -EPROTONOSUPPORT; -#endif + } /* When open for writing we refuse to open files with * compatible flags, too */ - if (f->writable) { -#ifdef HAVE_GCRYPT - if ((le32toh(f->header->compatible_flags) & ~HEADER_COMPATIBLE_SEALED) != 0) - return -EPROTONOSUPPORT; -#else - if (f->header->compatible_flags != 0) - return -EPROTONOSUPPORT; -#endif + flags = le32toh(f->header->compatible_flags); + if (f->writable && (flags & ~HEADER_COMPATIBLE_SUPPORTED)) { + if (flags & ~HEADER_COMPATIBLE_ANY) + log_debug("Journal file %s has unknown compatible flags %"PRIx32, + f->path, flags & ~HEADER_COMPATIBLE_ANY); + flags = (flags & HEADER_COMPATIBLE_ANY) & ~HEADER_COMPATIBLE_SUPPORTED; + if (flags) + log_debug("Journal file %s uses compatible flags %"PRIx32 + " disabled at compilation time.", f->path, flags); + return -EPROTONOSUPPORT; } if (f->header->state >= _STATE_MAX) @@ -274,12 +279,6 @@ static int journal_file_verify_header(JournalFile *f) { !VALID64(le64toh(f->header->entry_array_offset))) return -ENODATA; - if (le64toh(f->header->data_hash_table_offset) < le64toh(f->header->header_size) || - le64toh(f->header->field_hash_table_offset) < le64toh(f->header->header_size) || - le64toh(f->header->tail_object_offset) < le64toh(f->header->header_size) || - le64toh(f->header->entry_array_offset) < le64toh(f->header->header_size)) - return -ENODATA; - if (f->writable) { uint8_t state; sd_id128_t machine_id; @@ -305,7 +304,8 @@ static int journal_file_verify_header(JournalFile *f) { } } - f->compress = JOURNAL_HEADER_COMPRESSED(f->header); + f->compress_xz = JOURNAL_HEADER_COMPRESSED_XZ(f->header); + f->compress_lz4 = JOURNAL_HEADER_COMPRESSED_LZ4(f->header); f->seal = JOURNAL_HEADER_SEALED(f->header); @@ -333,12 +333,10 @@ static int journal_file_allocate(JournalFile *f, uint64_t offset, uint64_t size) if (new_size <= old_size) return 0; - if (f->metrics.max_size > 0 && - new_size > f->metrics.max_size) + if (f->metrics.max_size > 0 && new_size > f->metrics.max_size) return -E2BIG; - if (new_size > f->metrics.min_size && - f->metrics.keep_free > 0) { + if (new_size > f->metrics.min_size && f->metrics.keep_free > 0) { struct statvfs svfs; if (fstatvfs(f->fd, &svfs) >= 0) { @@ -356,6 +354,11 @@ static int journal_file_allocate(JournalFile *f, uint64_t offset, uint64_t size) } } + /* Increase by larger blocks at once */ + new_size = ((new_size+FILE_SIZE_INCREASE-1) / FILE_SIZE_INCREASE) * FILE_SIZE_INCREASE; + if (f->metrics.max_size > 0 && new_size > f->metrics.max_size) + new_size = f->metrics.max_size; + /* Note that the glibc fallocate() fallback is very inefficient, hence we try to minimize the allocation area as we can. */ @@ -371,7 +374,13 @@ static int journal_file_allocate(JournalFile *f, uint64_t offset, uint64_t size) return 0; } -static int journal_file_move_to(JournalFile *f, int context, bool keep_always, uint64_t offset, uint64_t size, void **ret) { +static unsigned type_to_context(ObjectType type) { + /* One context for each type, plus one catch-all for the rest */ + assert_cc(_OBJECT_TYPE_MAX <= MMAP_CACHE_MAX_CONTEXTS); + return type > OBJECT_UNUSED && type < _OBJECT_TYPE_MAX ? type : 0; +} + +static int journal_file_move_to(JournalFile *f, ObjectType type, bool keep_always, uint64_t offset, uint64_t size, void **ret) { assert(f); assert(ret); @@ -388,12 +397,12 @@ static int journal_file_move_to(JournalFile *f, int context, bool keep_always, u return -EADDRNOTAVAIL; } - return mmap_cache_get(f->mmap, f->fd, f->prot, context, keep_always, offset, size, &f->last_stat, ret); + return mmap_cache_get(f->mmap, f->fd, f->prot, type_to_context(type), keep_always, offset, size, &f->last_stat, ret); } static uint64_t minimum_header_size(Object *o) { - static uint64_t table[] = { + static const uint64_t table[] = { [OBJECT_DATA] = sizeof(DataObject), [OBJECT_FIELD] = sizeof(FieldObject), [OBJECT_ENTRY] = sizeof(EntryObject), @@ -409,12 +418,11 @@ static uint64_t minimum_header_size(Object *o) { return table[o->object.type]; } -int journal_file_move_to_object(JournalFile *f, int type, uint64_t offset, Object **ret) { +int journal_file_move_to_object(JournalFile *f, ObjectType type, uint64_t offset, Object **ret) { int r; void *t; Object *o; uint64_t s; - unsigned context; assert(f); assert(ret); @@ -423,10 +431,7 @@ int journal_file_move_to_object(JournalFile *f, int type, uint64_t offset, Objec if (!VALID64(offset)) return -EFAULT; - /* One context for each type, plus one catch-all for the rest */ - context = type > 0 && type < _OBJECT_TYPE_MAX ? type : 0; - - r = journal_file_move_to(f, context, false, offset, sizeof(ObjectHeader), &t); + r = journal_file_move_to(f, type, false, offset, sizeof(ObjectHeader), &t); if (r < 0) return r; @@ -442,11 +447,11 @@ int journal_file_move_to_object(JournalFile *f, int type, uint64_t offset, Objec if (s < minimum_header_size(o)) return -EBADMSG; - if (type > 0 && o->object.type != type) + if (type > OBJECT_UNUSED && o->object.type != type) return -EBADMSG; if (s > sizeof(ObjectHeader)) { - r = journal_file_move_to(f, o->object.type, false, offset, s, &t); + r = journal_file_move_to(f, type, false, offset, s, &t); if (r < 0) return r; @@ -483,14 +488,14 @@ static uint64_t journal_file_entry_seqnum(JournalFile *f, uint64_t *seqnum) { return r; } -int journal_file_append_object(JournalFile *f, int type, uint64_t size, Object **ret, uint64_t *offset) { +int journal_file_append_object(JournalFile *f, ObjectType type, uint64_t size, Object **ret, uint64_t *offset) { int r; uint64_t p; Object *tail, *o; void *t; assert(f); - assert(type > 0 && type < _OBJECT_TYPE_MAX); + assert(type > OBJECT_UNUSED && type < _OBJECT_TYPE_MAX); assert(size >= sizeof(ObjectHeader)); assert(offset); assert(ret); @@ -503,7 +508,7 @@ int journal_file_append_object(JournalFile *f, int type, uint64_t size, Object * if (p == 0) p = le64toh(f->header->header_size); else { - r = journal_file_move_to_object(f, -1, p, &tail); + r = journal_file_move_to_object(f, OBJECT_UNUSED, p, &tail); if (r < 0) return r; @@ -558,7 +563,7 @@ static int journal_file_setup_data_hash_table(JournalFile *f) { if (r < 0) return r; - memset(o->hash_table.items, 0, s); + memzero(o->hash_table.items, s); f->header->data_hash_table_offset = htole64(p + offsetof(Object, hash_table.items)); f->header->data_hash_table_size = htole64(s); @@ -584,7 +589,7 @@ static int journal_file_setup_field_hash_table(JournalFile *f) { if (r < 0) return r; - memset(o->hash_table.items, 0, s); + memzero(o->hash_table.items, s); f->header->field_hash_table_offset = htole64(p + offsetof(Object, hash_table.items)); f->header->field_hash_table_size = htole64(s); @@ -812,9 +817,10 @@ int journal_file_find_data_object_with_hash( if (le64toh(o->data.hash) != hash) goto next; - if (o->object.flags & OBJECT_COMPRESSED) { -#ifdef HAVE_XZ - uint64_t l, rsize; + if (o->object.flags & OBJECT_COMPRESSION_MASK) { +#if defined(HAVE_XZ) || defined(HAVE_LZ4) + uint64_t l; + size_t rsize; l = le64toh(o->object.size); if (l <= offsetof(Object, data.payload)) @@ -822,8 +828,10 @@ int journal_file_find_data_object_with_hash( l -= offsetof(Object, data.payload); - if (!uncompress_blob(o->data.payload, l, &f->compress_buffer, &f->compress_buffer_size, &rsize, 0)) - return -EBADMSG; + r = decompress_blob(o->object.flags & OBJECT_COMPRESSION_MASK, + o->data.payload, l, &f->compress_buffer, &f->compress_buffer_size, &rsize, 0); + if (r < 0) + return r; if (rsize == size && memcmp(f->compress_buffer, data, size) == 0) { @@ -839,7 +847,6 @@ int journal_file_find_data_object_with_hash( #else return -EPROTONOSUPPORT; #endif - } else if (le64toh(o->object.size) == osize && memcmp(o->data.payload, data, size) == 0) { @@ -907,6 +914,8 @@ static int journal_file_append_field( osize = offsetof(Object, field.payload) + size; r = journal_file_append_object(f, OBJECT_FIELD, osize, &o, &p); + if (r < 0) + return r; o->field.hash = htole64(hash); memcpy(o->field.payload, field, size); @@ -944,8 +953,7 @@ static int journal_file_append_data( uint64_t hash, p; uint64_t osize; Object *o; - int r; - bool compressed = false; + int r, compression = 0; const void *eq; assert(f); @@ -974,23 +982,24 @@ static int journal_file_append_data( o->data.hash = htole64(hash); -#ifdef HAVE_XZ - if (f->compress && +#if defined(HAVE_XZ) || defined(HAVE_LZ4) + if (f->compress_xz && size >= COMPRESSION_SIZE_THRESHOLD) { - uint64_t rsize; + size_t rsize; - compressed = compress_blob(data, size, o->data.payload, &rsize); + compression = compress_blob(data, size, o->data.payload, &rsize); - if (compressed) { + if (compression) { o->object.size = htole64(offsetof(Object, data.payload) + rsize); - o->object.flags |= OBJECT_COMPRESSED; + o->object.flags |= compression; - log_debug("Compressed data object %"PRIu64" -> %"PRIu64, size, rsize); + log_debug("Compressed data object %"PRIu64" -> %zu using %s", + size, rsize, object_compressed_to_string(compression)); } } #endif - if (!compressed && size > 0) + if (!compression && size > 0) memcpy(o->data.payload, data, size); r = journal_file_link_data(f, o, p, hash); @@ -1003,10 +1012,13 @@ static int journal_file_append_data( if (r < 0) return r; - eq = memchr(data, '=', size); + if (!data) + eq = NULL; + else + eq = memchr(data, '=', size); if (eq && eq > data) { + Object *fo = NULL; uint64_t fp; - Object *fo; /* Create field object ... */ r = journal_file_append_field(f, data, (uint8_t*) eq - (uint8_t*) data, &fo, &fp); @@ -1286,7 +1298,7 @@ void journal_file_post_change(JournalFile *f) { __sync_synchronize(); if (ftruncate(f->fd, f->last_stat.st_size) < 0) - log_error("Failed to truncate file to its own size: %m"); + log_error_errno(errno, "Failed to truncate file to its own size: %m"); } static int entry_item_cmp(const void *_a, const void *_b) { @@ -1342,7 +1354,7 @@ int journal_file_append_entry(JournalFile *f, const dual_timestamp *ts, const st /* Order by the position on disk, in order to improve seek * times for rotating media. */ - qsort(items, n_iovec, sizeof(EntryItem), entry_item_cmp); + qsort_safe(items, n_iovec, sizeof(EntryItem), entry_item_cmp); r = journal_file_append_entry_internal(f, ts, xor_hash, items, n_iovec, seqnum, ret, offset); @@ -1352,19 +1364,21 @@ int journal_file_append_entry(JournalFile *f, const dual_timestamp *ts, const st } typedef struct ChainCacheItem { - uint64_t first; /* the array at the begin of the chain */ + uint64_t first; /* the array at the beginning of the chain */ uint64_t array; /* the cached array */ uint64_t begin; /* the first item in the cached array */ uint64_t total; /* the total number of items in all arrays before this one in the chain */ + uint64_t last_index; /* the last index we looked at, to optimize locality when bisecting */ } ChainCacheItem; static void chain_cache_put( - Hashmap *h, + OrderedHashmap *h, ChainCacheItem *ci, uint64_t first, uint64_t array, uint64_t begin, - uint64_t total) { + uint64_t total, + uint64_t last_index) { if (!ci) { /* If the chain item to cache for this chain is the @@ -1372,9 +1386,10 @@ static void chain_cache_put( if (array == first) return; - if (hashmap_size(h) >= CHAIN_CACHE_MAX) - ci = hashmap_steal_first(h); - else { + if (ordered_hashmap_size(h) >= CHAIN_CACHE_MAX) { + ci = ordered_hashmap_steal_first(h); + assert(ci); + } else { ci = new(ChainCacheItem, 1); if (!ci) return; @@ -1382,7 +1397,7 @@ static void chain_cache_put( ci->first = first; - if (hashmap_put(h, &ci->first, ci) < 0) { + if (ordered_hashmap_put(h, &ci->first, ci) < 0) { free(ci); return; } @@ -1392,12 +1407,14 @@ static void chain_cache_put( ci->array = array; ci->begin = begin; ci->total = total; + ci->last_index = last_index; } -static int generic_array_get(JournalFile *f, - uint64_t first, - uint64_t i, - Object **ret, uint64_t *offset) { +static int generic_array_get( + JournalFile *f, + uint64_t first, + uint64_t i, + Object **ret, uint64_t *offset) { Object *o; uint64_t p = 0, a, t = 0; @@ -1409,7 +1426,7 @@ static int generic_array_get(JournalFile *f, a = first; /* Try the chain cache first */ - ci = hashmap_get(f->chain_cache, &first); + ci = ordered_hashmap_get(f->chain_cache, &first); if (ci && i > ci->total) { a = ci->array; i -= ci->total; @@ -1438,7 +1455,7 @@ static int generic_array_get(JournalFile *f, found: /* Let's cache this item for the next invocation */ - chain_cache_put(f->chain_cache, ci, first, a, o->entry_array.items[0], t); + chain_cache_put(f->chain_cache, ci, first, a, le64toh(o->entry_array.items[0]), t, i); r = journal_file_move_to_object(f, OBJECT_ENTRY, p, &o); if (r < 0) @@ -1453,11 +1470,12 @@ found: return 1; } -static int generic_array_get_plus_one(JournalFile *f, - uint64_t extra, - uint64_t first, - uint64_t i, - Object **ret, uint64_t *offset) { +static int generic_array_get_plus_one( + JournalFile *f, + uint64_t extra, + uint64_t first, + uint64_t i, + Object **ret, uint64_t *offset) { Object *o; @@ -1488,17 +1506,18 @@ enum { TEST_RIGHT }; -static int generic_array_bisect(JournalFile *f, - uint64_t first, - uint64_t n, - uint64_t needle, - int (*test_object)(JournalFile *f, uint64_t p, uint64_t needle), - direction_t direction, - Object **ret, - uint64_t *offset, - uint64_t *idx) { - - uint64_t a, p, t = 0, i = 0, last_p = 0; +static int generic_array_bisect( + JournalFile *f, + uint64_t first, + uint64_t n, + uint64_t needle, + int (*test_object)(JournalFile *f, uint64_t p, uint64_t needle), + direction_t direction, + Object **ret, + uint64_t *offset, + uint64_t *idx) { + + uint64_t a, p, t = 0, i = 0, last_p = 0, last_index = (uint64_t) -1; bool subtract_one = false; Object *o, *array = NULL; int r; @@ -1510,7 +1529,7 @@ static int generic_array_bisect(JournalFile *f, /* Start with the first array in the chain */ a = first; - ci = hashmap_get(f->chain_cache, &first); + ci = ordered_hashmap_get(f->chain_cache, &first); if (ci && n > ci->total) { /* Ah, we have iterated this bisection array chain * previously! Let's see if we can skip ahead in the @@ -1523,7 +1542,7 @@ static int generic_array_bisect(JournalFile *f, return r; if (r == TEST_LEFT) { - /* OK, what we are looking for is right of th + /* OK, what we are looking for is right of the * begin of this EntryArray, so let's jump * straight to previously cached array in the * chain */ @@ -1531,6 +1550,7 @@ static int generic_array_bisect(JournalFile *f, a = ci->array; n -= ci->total; t = ci->total; + last_index = ci->last_index; } } @@ -1561,6 +1581,58 @@ static int generic_array_bisect(JournalFile *f, if (r == TEST_RIGHT) { left = 0; right -= 1; + + if (last_index != (uint64_t) -1) { + assert(last_index <= right); + + /* If we cached the last index we + * looked at, let's try to not to jump + * too wildly around and see if we can + * limit the range to look at early to + * the immediate neighbors of the last + * index we looked at. */ + + if (last_index > 0) { + uint64_t x = last_index - 1; + + p = le64toh(array->entry_array.items[x]); + if (p <= 0) + return -EBADMSG; + + r = test_object(f, p, needle); + if (r < 0) + return r; + + if (r == TEST_FOUND) + r = direction == DIRECTION_DOWN ? TEST_RIGHT : TEST_LEFT; + + if (r == TEST_RIGHT) + right = x; + else + left = x + 1; + } + + if (last_index < right) { + uint64_t y = last_index + 1; + + p = le64toh(array->entry_array.items[y]); + if (p <= 0) + return -EBADMSG; + + r = test_object(f, p, needle); + if (r < 0) + return r; + + if (r == TEST_FOUND) + r = direction == DIRECTION_DOWN ? TEST_RIGHT : TEST_LEFT; + + if (r == TEST_RIGHT) + right = y; + else + left = y + 1; + } + } + for (;;) { if (left == right) { if (direction == DIRECTION_UP) @@ -1571,8 +1643,8 @@ static int generic_array_bisect(JournalFile *f, } assert(left < right); - i = (left + right) / 2; + p = le64toh(array->entry_array.items[i]); if (p <= 0) return -EBADMSG; @@ -1591,7 +1663,7 @@ static int generic_array_bisect(JournalFile *f, } } - if (k > n) { + if (k >= n) { if (direction == DIRECTION_UP) { i = n; subtract_one = true; @@ -1605,6 +1677,7 @@ static int generic_array_bisect(JournalFile *f, n -= k; t += k; + last_index = (uint64_t) -1; a = le64toh(array->entry_array.next_entry_array_offset); } @@ -1615,7 +1688,7 @@ found: return 0; /* Let's cache this item for the next invocation */ - chain_cache_put(f->chain_cache, ci, first, a, array->entry_array.items[0], t); + chain_cache_put(f->chain_cache, ci, first, a, le64toh(array->entry_array.items[0]), t, subtract_one ? (i > 0 ? i-1 : (uint64_t) -1) : i); if (subtract_one && i == 0) p = last_p; @@ -1640,16 +1713,18 @@ found: return 1; } -static int generic_array_bisect_plus_one(JournalFile *f, - uint64_t extra, - uint64_t first, - uint64_t n, - uint64_t needle, - int (*test_object)(JournalFile *f, uint64_t p, uint64_t needle), - direction_t direction, - Object **ret, - uint64_t *offset, - uint64_t *idx) { + +static int generic_array_bisect_plus_one( + JournalFile *f, + uint64_t extra, + uint64_t first, + uint64_t n, + uint64_t needle, + int (*test_object)(JournalFile *f, uint64_t p, uint64_t needle), + direction_t direction, + Object **ret, + uint64_t *offset, + uint64_t *idx) { int r; bool step_back = false; @@ -1724,23 +1799,6 @@ _pure_ static int test_object_offset(JournalFile *f, uint64_t p, uint64_t needle return TEST_RIGHT; } -int journal_file_move_to_entry_by_offset( - JournalFile *f, - uint64_t p, - direction_t direction, - Object **ret, - uint64_t *offset) { - - return generic_array_bisect(f, - le64toh(f->header->entry_array_offset), - le64toh(f->header->n_entries), - p, - test_object_offset, - direction, - ret, offset, NULL); -} - - static int test_object_seqnum(JournalFile *f, uint64_t p, uint64_t needle) { Object *o; int r; @@ -1870,13 +1928,17 @@ int journal_file_move_to_entry_by_monotonic( ret, offset, NULL); } +void journal_file_reset_location(JournalFile *f) { + f->current_offset = 0; +} + int journal_file_next_entry( JournalFile *f, Object *o, uint64_t p, direction_t direction, Object **ret, uint64_t *offset) { - uint64_t i, n; + uint64_t i, n, ofs; int r; assert(f); @@ -1917,59 +1979,24 @@ int journal_file_next_entry( } /* And jump to it */ - return generic_array_get(f, - le64toh(f->header->entry_array_offset), - i, - ret, offset); -} - -int journal_file_skip_entry( - JournalFile *f, - Object *o, uint64_t p, - int64_t skip, - Object **ret, uint64_t *offset) { - - uint64_t i, n; - int r; - - assert(f); - assert(o); - assert(p > 0); - - if (o->object.type != OBJECT_ENTRY) - return -EINVAL; - - r = generic_array_bisect(f, - le64toh(f->header->entry_array_offset), - le64toh(f->header->n_entries), - p, - test_object_offset, - DIRECTION_DOWN, - NULL, NULL, - &i); + r = generic_array_get(f, + le64toh(f->header->entry_array_offset), + i, + ret, &ofs); if (r <= 0) return r; - /* Calculate new index */ - if (skip < 0) { - if ((uint64_t) -skip >= i) - i = 0; - else - i = i - (uint64_t) -skip; - } else - i += (uint64_t) skip; - - n = le64toh(f->header->n_entries); - if (n <= 0) + if (p > 0 && + (direction == DIRECTION_DOWN ? ofs <= p : ofs >= p)) { + log_debug("%s: entry array corrupted at entry %"PRIu64, + f->path, i); return -EBADMSG; + } - if (i >= n) - i = n-1; + if (offset) + *offset = ofs; - return generic_array_get(f, - le64toh(f->header->entry_array_offset), - i, - ret, offset); + return 1; } int journal_file_next_entry_for_data( @@ -2141,8 +2168,6 @@ int journal_file_move_to_entry_by_monotonic_for_data( z = q; } - - return 0; } int journal_file_move_to_entry_by_seqnum_for_data( @@ -2208,7 +2233,7 @@ void journal_file_dump(JournalFile *f) { p = le64toh(f->header->header_size); while (p != 0) { - r = journal_file_move_to_object(f, -1, p, &o); + r = journal_file_move_to_object(f, OBJECT_UNUSED, p, &o); if (r < 0) goto fail; @@ -2256,8 +2281,9 @@ void journal_file_dump(JournalFile *f) { break; } - if (o->object.flags & OBJECT_COMPRESSED) - printf("Flags: COMPRESSED\n"); + if (o->object.flags & OBJECT_COMPRESSION_MASK) + printf("Flags: %s\n", + object_compressed_to_string(o->object.flags & OBJECT_COMPRESSION_MASK)); if (p == le64toh(f->header->tail_object_offset)) p = 0; @@ -2270,9 +2296,18 @@ fail: log_error("File corrupt"); } +static const char* format_timestamp_safe(char *buf, size_t l, usec_t t) { + const char *x; + + x = format_timestamp(buf, l, t); + if (x) + return x; + return " --- "; +} + void journal_file_print_header(JournalFile *f) { - char a[33], b[33], c[33]; - char x[FORMAT_TIMESTAMP_MAX], y[FORMAT_TIMESTAMP_MAX]; + char a[33], b[33], c[33], d[33]; + char x[FORMAT_TIMESTAMP_MAX], y[FORMAT_TIMESTAMP_MAX], z[FORMAT_TIMESTAMP_MAX]; struct stat st; char bytes[FORMAT_BYTES_MAX]; @@ -2285,7 +2320,7 @@ void journal_file_print_header(JournalFile *f) { "Sequential Number ID: %s\n" "State: %s\n" "Compatible Flags:%s%s\n" - "Incompatible Flags:%s%s\n" + "Incompatible Flags:%s%s%s\n" "Header size: %"PRIu64"\n" "Arena size: %"PRIu64"\n" "Data Hash Table Size: %"PRIu64"\n" @@ -2295,20 +2330,22 @@ void journal_file_print_header(JournalFile *f) { "Tail Sequential Number: %"PRIu64"\n" "Head Realtime Timestamp: %s\n" "Tail Realtime Timestamp: %s\n" + "Tail Monotonic Timestamp: %s\n" "Objects: %"PRIu64"\n" "Entry Objects: %"PRIu64"\n", f->path, sd_id128_to_string(f->header->file_id, a), sd_id128_to_string(f->header->machine_id, b), sd_id128_to_string(f->header->boot_id, c), - sd_id128_to_string(f->header->seqnum_id, c), + sd_id128_to_string(f->header->seqnum_id, d), f->header->state == STATE_OFFLINE ? "OFFLINE" : f->header->state == STATE_ONLINE ? "ONLINE" : f->header->state == STATE_ARCHIVED ? "ARCHIVED" : "UNKNOWN", JOURNAL_HEADER_SEALED(f->header) ? " SEALED" : "", - (le32toh(f->header->compatible_flags) & ~HEADER_COMPATIBLE_SEALED) ? " ???" : "", - JOURNAL_HEADER_COMPRESSED(f->header) ? " COMPRESSED" : "", - (le32toh(f->header->incompatible_flags) & ~HEADER_INCOMPATIBLE_COMPRESSED) ? " ???" : "", + (le32toh(f->header->compatible_flags) & ~HEADER_COMPATIBLE_ANY) ? " ???" : "", + JOURNAL_HEADER_COMPRESSED_XZ(f->header) ? " COMPRESSED-XZ" : "", + JOURNAL_HEADER_COMPRESSED_LZ4(f->header) ? " COMPRESSED-LZ4" : "", + (le32toh(f->header->incompatible_flags) & ~HEADER_INCOMPATIBLE_ANY) ? " ???" : "", le64toh(f->header->header_size), le64toh(f->header->arena_size), le64toh(f->header->data_hash_table_size) / sizeof(HashItem), @@ -2316,8 +2353,9 @@ void journal_file_print_header(JournalFile *f) { yes_no(journal_file_rotate_suggested(f, 0)), le64toh(f->header->head_entry_seqnum), le64toh(f->header->tail_entry_seqnum), - format_timestamp(x, sizeof(x), le64toh(f->header->head_entry_realtime)), - format_timestamp(y, sizeof(y), le64toh(f->header->tail_entry_realtime)), + format_timestamp_safe(x, sizeof(x), le64toh(f->header->head_entry_realtime)), + format_timestamp_safe(y, sizeof(y), le64toh(f->header->tail_entry_realtime)), + format_timespan(z, sizeof(z), le64toh(f->header->tail_entry_monotonic), USEC_PER_MSEC), le64toh(f->header->n_objects), le64toh(f->header->n_entries)); @@ -2380,8 +2418,10 @@ int journal_file_open( f->flags = flags; f->prot = prot_from_flags(flags); f->writable = (flags & O_ACCMODE) != O_RDONLY; -#ifdef HAVE_XZ - f->compress = compress; +#if defined(HAVE_LZ4) + f->compress_lz4 = compress; +#elif defined(HAVE_XZ) + f->compress_xz = compress; #endif #ifdef HAVE_GCRYPT f->seal = seal; @@ -2403,7 +2443,7 @@ int journal_file_open( goto fail; } - f->chain_cache = hashmap_new(uint64_hash_func, uint64_compare_func); + f->chain_cache = ordered_hashmap_new(&uint64_hash_ops); if (!f->chain_cache) { r = -ENOMEM; goto fail; @@ -2421,7 +2461,6 @@ int journal_file_open( } if (f->last_stat.st_size == 0 && f->writable) { -#ifdef HAVE_XATTR uint64_t crtime; /* Let's attach the creation time to the journal file, @@ -2432,11 +2471,10 @@ int journal_file_open( * currently no usable API to query this, hence let's * emulate this via extended attributes. If extended * attributes are not supported we'll just skip this, - * and rely solely on mtime/atime/ctime of the file.*/ + * and rely solely on mtime/atime/ctime of the file. */ crtime = htole64((uint64_t) now(CLOCK_REALTIME)); fsetxattr(f->fd, "user.crtime_usec", &crtime, sizeof(crtime), XATTR_CREATE); -#endif #ifdef HAVE_GCRYPT /* Try to load the FSPRG state, and if we can't, then @@ -2538,7 +2576,7 @@ fail: } int journal_file_rotate(JournalFile **f, bool compress, bool seal) { - char *p; + _cleanup_free_ char *p = NULL; size_t l; JournalFile *old_file, *new_file = NULL; int r; @@ -2555,22 +2593,15 @@ int journal_file_rotate(JournalFile **f, bool compress, bool seal) { return -EINVAL; l = strlen(old_file->path); - - p = new(char, l + 1 + 32 + 1 + 16 + 1 + 16 + 1); - if (!p) + r = asprintf(&p, "%.*s@" SD_ID128_FORMAT_STR "-%016"PRIx64"-%016"PRIx64".journal", + (int) l - 8, old_file->path, + SD_ID128_FORMAT_VAL(old_file->header->seqnum_id), + le64toh((*f)->header->head_entry_seqnum), + le64toh((*f)->header->head_entry_realtime)); + if (r < 0) return -ENOMEM; - memcpy(p, old_file->path, l - 8); - p[l-8] = '@'; - sd_id128_to_string(old_file->header->seqnum_id, p + l - 8 + 1); - snprintf(p + l - 8 + 1 + 32, 1 + 16 + 1 + 16 + 8 + 1, - "-%016"PRIx64"-%016"PRIx64".journal", - le64toh((*f)->header->head_entry_seqnum), - le64toh((*f)->header->head_entry_realtime)); - r = rename(old_file->path, p); - free(p); - if (r < 0) return -errno; @@ -2596,7 +2627,7 @@ int journal_file_open_reliably( int r; size_t l; - char *p; + _cleanup_free_ char *p = NULL; r = journal_file_open(fname, flags, mode, compress, seal, metrics, mmap_cache, template, ret); @@ -2620,14 +2651,13 @@ int journal_file_open_reliably( /* The file is corrupted. Rotate it away and try it again (but only once) */ l = strlen(fname); - if (asprintf(&p, "%.*s@%016llx-%016llx.journal~", - (int) (l-8), fname, + if (asprintf(&p, "%.*s@%016llx-%016" PRIx64 ".journal~", + (int) l - 8, fname, (unsigned long long) now(CLOCK_REALTIME), - random_ull()) < 0) + random_u64()) < 0) return -ENOMEM; r = rename(fname, p); - free(p); if (r < 0) return -errno; @@ -2655,12 +2685,9 @@ int journal_file_copy_entry(JournalFile *from, JournalFile *to, Object *o, uint6 ts.monotonic = le64toh(o->entry.monotonic); ts.realtime = le64toh(o->entry.realtime); - if (to->tail_entry_monotonic_valid && - ts.monotonic < le64toh(to->header->tail_entry_monotonic)) - return -EINVAL; - n = journal_file_entry_n_items(o); - items = alloca(sizeof(EntryItem) * n); + /* alloca() can't take 0, hence let's allocate at least one */ + items = alloca(sizeof(EntryItem) * MAX(1u, n)); for (i = 0; i < n; i++) { uint64_t l, h; @@ -2686,12 +2713,14 @@ int journal_file_copy_entry(JournalFile *from, JournalFile *to, Object *o, uint6 if ((uint64_t) t != l) return -E2BIG; - if (o->object.flags & OBJECT_COMPRESSED) { -#ifdef HAVE_XZ - uint64_t rsize; + if (o->object.flags & OBJECT_COMPRESSION_MASK) { +#if defined(HAVE_XZ) || defined(HAVE_LZ4) + size_t rsize; - if (!uncompress_blob(o->data.payload, l, &from->compress_buffer, &from->compress_buffer_size, &rsize, 0)) - return -EBADMSG; + r = decompress_blob(o->object.flags & OBJECT_COMPRESSION_MASK, + o->data.payload, l, &from->compress_buffer, &from->compress_buffer_size, &rsize, 0); + if (r < 0) + return r; data = from->compress_buffer; l = rsize;