summaryrefslogtreecommitdiff
path: root/src
diff options
context:
space:
mode:
authorLennart Poettering <lennart@poettering.net>2011-10-08 02:20:44 +0200
committerLennart Poettering <lennart@poettering.net>2011-10-08 02:21:05 +0200
commitcec736d21ff86c4ac81b4d306ddba2120333818c (patch)
tree4df4f1126307a4b6d066318e2f468847099beb4a /src
parentf4b4781191e8edfb5690e4447166e3ba7bcb48f5 (diff)
journal: implement parallel traversal in client
Diffstat (limited to 'src')
-rw-r--r--src/journal/journal-def.h5
-rw-r--r--src/journal/journal-file.c1191
-rw-r--r--src/journal/journal-file.h (renamed from src/journal/journal-private.h)18
-rw-r--r--src/journal/journalctl.c11
-rw-r--r--src/journal/journald.c11
-rw-r--r--src/journal/sd-journal.c1241
-rw-r--r--src/journal/sd-journal.h27
-rw-r--r--src/journal/test-journal.c2
8 files changed, 1354 insertions, 1152 deletions
diff --git a/src/journal/journal-def.h b/src/journal/journal-def.h
index 2a519fe0db..b3fa1e524f 100644
--- a/src/journal/journal-def.h
+++ b/src/journal/journal-def.h
@@ -48,7 +48,7 @@ enum {
_packed_ struct ObjectHeader {
uint8_t type;
- uint8_t reserved[3];
+ uint8_t reserved[7];
uint64_t size;
uint8_t payload[];
};
@@ -74,6 +74,7 @@ _packed_ struct EntryObject {
uint64_t seqnum;
uint64_t realtime;
uint64_t monotonic;
+ sd_id128_t boot_id;
uint64_t xor_hash;
uint64_t prev_entry_offset;
uint64_t next_entry_offset;
@@ -118,6 +119,7 @@ _packed_ struct Header {
sd_id128_t file_id;
sd_id128_t machine_id;
sd_id128_t boot_id;
+ sd_id128_t seqnum_id;
uint64_t arena_offset;
uint64_t arena_size;
uint64_t arena_max_size;
@@ -133,7 +135,6 @@ _packed_ struct Header {
uint64_t tail_entry_offset;
uint64_t last_bisect_offset;
uint64_t n_objects;
- uint64_t seqnum_base;
uint64_t seqnum;
};
diff --git a/src/journal/journal-file.c b/src/journal/journal-file.c
new file mode 100644
index 0000000000..37e2e37eb1
--- /dev/null
+++ b/src/journal/journal-file.c
@@ -0,0 +1,1191 @@
+/*-*- Mode: C; c-basic-offset: 8; indent-tabs-mode: nil -*-*/
+
+/***
+ This file is part of systemd.
+
+ Copyright 2011 Lennart Poettering
+
+ systemd is free software; you can redistribute it and/or modify it
+ under the terms of the GNU General Public License as published by
+ the Free Software Foundation; either version 2 of the License, or
+ (at your option) any later version.
+
+ systemd is distributed in the hope that it will be useful, but
+ WITHOUT ANY WARRANTY; without even the implied warranty of
+ MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
+ General Public License for more details.
+
+ You should have received a copy of the GNU General Public License
+ along with systemd; If not, see <http://www.gnu.org/licenses/>.
+***/
+
+#include <sys/mman.h>
+#include <errno.h>
+#include <sys/uio.h>
+#include <unistd.h>
+#include <sys/statvfs.h>
+#include <fcntl.h>
+#include <stddef.h>
+
+#include "journal-def.h"
+#include "journal-file.h"
+#include "lookup3.h"
+
+#define DEFAULT_ARENA_MAX_SIZE (16ULL*1024ULL*1024ULL*1024ULL)
+#define DEFAULT_ARENA_MIN_SIZE (256ULL*1024ULL)
+#define DEFAULT_ARENA_KEEP_FREE (1ULL*1024ULL*1024ULL)
+
+#define DEFAULT_HASH_TABLE_SIZE (2047ULL*16ULL)
+#define DEFAULT_BISECT_TABLE_SIZE ((DEFAULT_ARENA_MAX_SIZE/(64ULL*1024ULL))*8ULL)
+
+#define DEFAULT_WINDOW_SIZE (128ULL*1024ULL*1024ULL)
+
+static const char signature[] = { 'L', 'P', 'K', 'S', 'H', 'H', 'R', 'H' };
+
+#define ALIGN64(x) (((x) + 7ULL) & ~7ULL)
+
+void journal_file_close(JournalFile *f) {
+ assert(f);
+
+ if (f->fd >= 0)
+ close_nointr_nofail(f->fd);
+
+ if (f->header)
+ munmap(f->header, PAGE_ALIGN(sizeof(Header)));
+
+ if (f->hash_table_window)
+ munmap(f->hash_table_window, f->hash_table_window_size);
+
+ if (f->bisect_table_window)
+ munmap(f->bisect_table_window, f->bisect_table_window_size);
+
+ if (f->window)
+ munmap(f->window, f->window_size);
+
+ free(f->path);
+ free(f);
+}
+
+static int journal_file_init_header(JournalFile *f) {
+ Header h;
+ ssize_t k;
+ int r;
+
+ assert(f);
+
+ zero(h);
+ memcpy(h.signature, signature, 8);
+ h.arena_offset = htole64(ALIGN64(sizeof(h)));
+ h.arena_max_size = htole64(DEFAULT_ARENA_MAX_SIZE);
+ h.arena_min_size = htole64(DEFAULT_ARENA_MIN_SIZE);
+ h.arena_keep_free = htole64(DEFAULT_ARENA_KEEP_FREE);
+
+ r = sd_id128_randomize(&h.file_id);
+ if (r < 0)
+ return r;
+
+ h.seqnum_id = h.file_id;
+
+ k = pwrite(f->fd, &h, sizeof(h), 0);
+ if (k < 0)
+ return -errno;
+
+ if (k != sizeof(h))
+ return -EIO;
+
+ return 0;
+}
+
+static int journal_file_refresh_header(JournalFile *f) {
+ int r;
+
+ assert(f);
+
+ r = sd_id128_get_machine(&f->header->machine_id);
+ if (r < 0)
+ return r;
+
+ r = sd_id128_get_boot(&f->header->boot_id);
+ if (r < 0)
+ return r;
+
+ f->header->state = htole32(STATE_ONLINE);
+ return 0;
+}
+
+static int journal_file_verify_header(JournalFile *f) {
+ assert(f);
+
+ if (memcmp(f->header, signature, 8))
+ return -EBADMSG;
+
+ if (f->header->incompatible_flags != 0)
+ return -EPROTONOSUPPORT;
+
+ if ((uint64_t) f->last_stat.st_size < (le64toh(f->header->arena_offset) + le64toh(f->header->arena_size)))
+ return -ENODATA;
+
+ if (f->writable) {
+ uint32_t state;
+ sd_id128_t machine_id;
+ int r;
+
+ r = sd_id128_get_machine(&machine_id);
+ if (r < 0)
+ return r;
+
+ if (!sd_id128_equal(machine_id, f->header->machine_id))
+ return -EHOSTDOWN;
+
+ state = le32toh(f->header->state);
+
+ if (state == STATE_ONLINE)
+ log_debug("Journal file %s is already online. Assuming unclean closing. Ignoring.", f->path);
+ else if (state == STATE_ARCHIVED)
+ return -ESHUTDOWN;
+ else if (state != STATE_OFFLINE)
+ log_debug("Journal file %s has unknown state %u. Ignoring.", f->path, state);
+ }
+
+ return 0;
+}
+
+static int journal_file_allocate(JournalFile *f, uint64_t offset, uint64_t size) {
+ uint64_t asize;
+ uint64_t old_size, new_size;
+
+ assert(f);
+
+ if (offset < le64toh(f->header->arena_offset))
+ return -EINVAL;
+
+ new_size = PAGE_ALIGN(offset + size);
+
+ /* We assume that this file is not sparse, and we know that
+ * for sure, since we alway call posix_fallocate()
+ * ourselves */
+
+ old_size =
+ le64toh(f->header->arena_offset) +
+ le64toh(f->header->arena_size);
+
+ if (old_size >= new_size)
+ return 0;
+
+ asize = new_size - le64toh(f->header->arena_offset);
+
+ if (asize > le64toh(f->header->arena_min_size)) {
+ struct statvfs svfs;
+
+ if (fstatvfs(f->fd, &svfs) >= 0) {
+ uint64_t available;
+
+ available = svfs.f_bfree * svfs.f_bsize;
+
+ if (available >= f->header->arena_keep_free)
+ available -= f->header->arena_keep_free;
+ else
+ available = 0;
+
+ if (new_size - old_size > available)
+ return -E2BIG;
+ }
+ }
+
+ if (asize > le64toh(f->header->arena_max_size))
+ return -E2BIG;
+
+ if (posix_fallocate(f->fd, 0, new_size) < 0)
+ return -errno;
+
+ if (fstat(f->fd, &f->last_stat) < 0)
+ return -errno;
+
+ f->header->arena_size = htole64(asize);
+
+ return 0;
+}
+
+static int journal_file_map(
+ JournalFile *f,
+ uint64_t offset,
+ uint64_t size,
+ void **_window,
+ uint64_t *_woffset,
+ uint64_t *_wsize,
+ void **ret) {
+
+ uint64_t woffset, wsize;
+ void *window;
+
+ assert(f);
+ assert(size > 0);
+ assert(ret);
+
+ woffset = offset & ~((uint64_t) page_size() - 1ULL);
+ wsize = size + (offset - woffset);
+ wsize = PAGE_ALIGN(wsize);
+
+ window = mmap(NULL, wsize, f->prot, MAP_SHARED, f->fd, woffset);
+ if (window == MAP_FAILED)
+ return -errno;
+
+ if (_window)
+ *_window = window;
+
+ if (_woffset)
+ *_woffset = woffset;
+
+ if (_wsize)
+ *_wsize = wsize;
+
+ *ret = (uint8_t*) window + (offset - woffset);
+
+ return 0;
+}
+
+static int journal_file_move_to(JournalFile *f, uint64_t offset, uint64_t size, void **ret) {
+ void *p;
+ uint64_t delta;
+ int r;
+
+ assert(f);
+ assert(ret);
+
+ if (_likely_(f->window &&
+ f->window_offset <= offset &&
+ f->window_offset+f->window_size >= offset + size)) {
+
+ *ret = (uint8_t*) f->window + (offset - f->window_offset);
+ return 0;
+ }
+
+ if (f->window) {
+ if (munmap(f->window, f->window_size) < 0)
+ return -errno;
+
+ f->window = NULL;
+ f->window_size = f->window_offset = 0;
+ }
+
+ if (size < DEFAULT_WINDOW_SIZE) {
+ /* If the default window size is larger then what was
+ * asked for extend the mapping a bit in the hope to
+ * minimize needed remappings later on. We add half
+ * the window space before and half behind the
+ * requested mapping */
+
+ delta = PAGE_ALIGN((DEFAULT_WINDOW_SIZE - size) / 2);
+
+ if (offset < delta)
+ delta = offset;
+
+ offset -= delta;
+ size += (DEFAULT_WINDOW_SIZE - delta);
+ } else
+ delta = 0;
+
+ r = journal_file_map(f,
+ offset, size,
+ &f->window, &f->window_offset, &f->window_size,
+ & p);
+
+ if (r < 0)
+ return r;
+
+ *ret = (uint8_t*) p + delta;
+ return 0;
+}
+
+static bool verify_hash(Object *o) {
+ uint64_t t;
+
+ assert(o);
+
+ t = le64toh(o->object.type);
+ if (t == OBJECT_DATA) {
+ uint64_t s, h1, h2;
+
+ s = le64toh(o->object.size);
+
+ h1 = le64toh(o->data.hash);
+ h2 = hash64(o->data.payload, s - offsetof(Object, data.payload));
+
+ return h1 == h2;
+ }
+
+ return true;
+}
+
+int journal_file_move_to_object(JournalFile *f, uint64_t offset, int type, Object **ret) {
+ int r;
+ void *t;
+ Object *o;
+ uint64_t s;
+
+ assert(f);
+ assert(ret);
+
+ r = journal_file_move_to(f, offset, sizeof(ObjectHeader), &t);
+ if (r < 0)
+ return r;
+
+ o = (Object*) t;
+ s = le64toh(o->object.size);
+
+ if (s < sizeof(ObjectHeader))
+ return -EBADMSG;
+
+ if (type >= 0 && le64toh(o->object.type) != type)
+ return -EBADMSG;
+
+ if (s > sizeof(ObjectHeader)) {
+ r = journal_file_move_to(f, offset, s, &t);
+ if (r < 0)
+ return r;
+
+ o = (Object*) t;
+ }
+
+ if (!verify_hash(o))
+ return -EBADMSG;
+
+ *ret = o;
+ return 0;
+}
+
+static uint64_t journal_file_seqnum(JournalFile *f) {
+ uint64_t r;
+
+ assert(f);
+
+ r = le64toh(f->header->seqnum) + 1;
+ f->header->seqnum = htole64(r);
+
+ return r;
+}
+
+static int journal_file_append_object(JournalFile *f, uint64_t size, Object **ret, uint64_t *offset) {
+ int r;
+ uint64_t p;
+ Object *tail, *o;
+ void *t;
+
+ assert(f);
+ assert(size >= sizeof(ObjectHeader));
+ assert(offset);
+ assert(ret);
+
+ p = le64toh(f->header->tail_object_offset);
+
+ if (p == 0)
+ p = le64toh(f->header->arena_offset);
+ else {
+ r = journal_file_move_to_object(f, p, -1, &tail);
+ if (r < 0)
+ return r;
+
+ p += ALIGN64(le64toh(tail->object.size));
+ }
+
+ r = journal_file_allocate(f, p, size);
+ if (r < 0)
+ return r;
+
+ r = journal_file_move_to(f, p, size, &t);
+ if (r < 0)
+ return r;
+
+ o = (Object*) t;
+
+ zero(o->object);
+ o->object.type = htole64(OBJECT_UNUSED);
+ zero(o->object.reserved);
+ o->object.size = htole64(size);
+
+ f->header->tail_object_offset = htole64(p);
+ if (f->header->head_object_offset == 0)
+ f->header->head_object_offset = htole64(p);
+
+ f->header->n_objects = htole64(le64toh(f->header->n_objects) + 1);
+
+ *ret = o;
+ *offset = p;
+
+ return 0;
+}
+
+static int journal_file_setup_hash_table(JournalFile *f) {
+ uint64_t s, p;
+ Object *o;
+ int r;
+
+ assert(f);
+
+ s = DEFAULT_HASH_TABLE_SIZE;
+ r = journal_file_append_object(f, offsetof(Object, hash_table.table) + s, &o, &p);
+ if (r < 0)
+ return r;
+
+ o->object.type = htole64(OBJECT_HASH_TABLE);
+ memset(o->hash_table.table, 0, s);
+
+ f->header->hash_table_offset = htole64(p + offsetof(Object, hash_table.table));
+ f->header->hash_table_size = htole64(s);
+
+ return 0;
+}
+
+static int journal_file_setup_bisect_table(JournalFile *f) {
+ uint64_t s, p;
+ Object *o;
+ int r;
+
+ assert(f);
+
+ s = DEFAULT_BISECT_TABLE_SIZE;
+ r = journal_file_append_object(f, offsetof(Object, bisect_table.table) + s, &o, &p);
+ if (r < 0)
+ return r;
+
+ o->object.type = htole64(OBJECT_BISECT_TABLE);
+ memset(o->bisect_table.table, 0, s);
+
+ f->header->bisect_table_offset = htole64(p + offsetof(Object, bisect_table.table));
+ f->header->bisect_table_size = htole64(s);
+
+ return 0;
+}
+
+static int journal_file_map_hash_table(JournalFile *f) {
+ uint64_t s, p;
+ void *t;
+ int r;
+
+ assert(f);
+
+ p = le64toh(f->header->hash_table_offset);
+ s = le64toh(f->header->hash_table_size);
+
+ r = journal_file_map(f,
+ p, s,
+ &f->hash_table_window, NULL, &f->hash_table_window_size,
+ &t);
+ if (r < 0)
+ return r;
+
+ f->hash_table = t;
+ return 0;
+}
+
+static int journal_file_map_bisect_table(JournalFile *f) {
+ uint64_t s, p;
+ void *t;
+ int r;
+
+ assert(f);
+
+ p = le64toh(f->header->bisect_table_offset);
+ s = le64toh(f->header->bisect_table_size);
+
+ r = journal_file_map(f,
+ p, s,
+ &f->bisect_table_window, NULL, &f->bisect_table_window_size,
+ &t);
+
+ if (r < 0)
+ return r;
+
+ f->bisect_table = t;
+ return 0;
+}
+
+static int journal_file_link_data(JournalFile *f, Object *o, uint64_t offset, uint64_t hash_index) {
+ uint64_t p;
+ int r;
+
+ assert(f);
+ assert(o);
+ assert(offset > 0);
+ assert(o->object.type == htole64(OBJECT_DATA));
+
+ o->data.head_entry_offset = o->data.tail_entry_offset = 0;
+ o->data.next_hash_offset = 0;
+
+ p = le64toh(f->hash_table[hash_index].tail_hash_offset);
+ if (p == 0) {
+ /* Only entry in the hash table is easy */
+
+ o->data.prev_hash_offset = 0;
+ f->hash_table[hash_index].head_hash_offset = htole64(offset);
+ } else {
+ o->data.prev_hash_offset = htole64(p);
+
+ /* Temporarily move back to the previous data object,
+ * to patch in pointer */
+
+ r = journal_file_move_to_object(f, p, OBJECT_DATA, &o);
+ if (r < 0)
+ return r;
+
+ o->data.next_hash_offset = offset;
+
+ r = journal_file_move_to_object(f, offset, OBJECT_DATA, &o);
+ if (r < 0)
+ return r;
+ }
+
+ f->hash_table[hash_index].tail_hash_offset = htole64(offset);
+
+ return 0;
+}
+
+static int journal_file_append_data(JournalFile *f, const void *data, uint64_t size, Object **ret, uint64_t *offset) {
+ uint64_t hash, h, p, np;
+ uint64_t osize;
+ Object *o;
+ int r;
+
+ assert(f);
+ assert(data || size == 0);
+
+ osize = offsetof(Object, data.payload) + size;
+
+ hash = hash64(data, size);
+ h = hash % (le64toh(f->header->hash_table_size) / sizeof(HashItem));
+ p = le64toh(f->hash_table[h].head_hash_offset);
+
+ while (p != 0) {
+ /* Look for this data object in the hash table */
+
+ r = journal_file_move_to_object(f, p, OBJECT_DATA, &o);
+ if (r < 0)
+ return r;
+
+ if (le64toh(o->object.size) == osize &&
+ memcmp(o->data.payload, data, size) == 0) {
+
+ if (le64toh(o->data.hash) != hash)
+ return -EBADMSG;
+
+ if (ret)
+ *ret = o;
+
+ if (offset)
+ *offset = p;
+
+ return 0;
+ }
+
+ p = le64toh(o->data.next_hash_offset);
+ }
+
+ r = journal_file_append_object(f, osize, &o, &np);
+ if (r < 0)
+ return r;
+
+ o->object.type = htole64(OBJECT_DATA);
+ o->data.hash = htole64(hash);
+ memcpy(o->data.payload, data, size);
+
+ r = journal_file_link_data(f, o, np, h);
+ if (r < 0)
+ return r;
+
+ if (ret)
+ *ret = o;
+
+ if (offset)
+ *offset = np;
+
+ return 0;
+}
+
+uint64_t journal_file_entry_n_items(Object *o) {
+ assert(o);
+ assert(o->object.type == htole64(OBJECT_ENTRY));
+
+ return (le64toh(o->object.size) - offsetof(Object, entry.items)) / sizeof(EntryItem);
+}
+
+static int journal_file_link_entry_item(JournalFile *f, Object *o, uint64_t offset, uint64_t i) {
+ uint64_t p, q;
+ int r;
+ assert(f);
+ assert(o);
+ assert(offset > 0);
+
+ p = le64toh(o->entry.items[i].object_offset);
+ if (p == 0)
+ return -EINVAL;
+
+ o->entry.items[i].next_entry_offset = 0;
+
+ /* Move to the data object */
+ r = journal_file_move_to_object(f, p, OBJECT_DATA, &o);
+ if (r < 0)
+ return r;
+
+ q = le64toh(o->data.tail_entry_offset);
+ o->data.tail_entry_offset = htole64(offset);
+
+ if (q == 0)
+ o->data.head_entry_offset = htole64(offset);
+ else {
+ uint64_t n, j;
+
+ /* Move to previous entry */
+ r = journal_file_move_to_object(f, q, OBJECT_ENTRY, &o);
+ if (r < 0)
+ return r;
+
+ n = journal_file_entry_n_items(o);
+ for (j = 0; j < n; j++)
+ if (le64toh(o->entry.items[j].object_offset) == p)
+ break;
+
+ if (j >= n)
+ return -EBADMSG;
+
+ o->entry.items[j].next_entry_offset = offset;
+ }
+
+ /* Move back to original entry */
+ r = journal_file_move_to_object(f, offset, OBJECT_ENTRY, &o);
+ if (r < 0)
+ return r;
+
+ o->entry.items[i].prev_entry_offset = q;
+ return 0;
+}
+
+static int journal_file_link_entry(JournalFile *f, Object *o, uint64_t offset) {
+ uint64_t p, i, n, k, a, b;
+ int r;
+
+ assert(f);
+ assert(o);
+ assert(offset > 0);
+ assert(o->object.type == htole64(OBJECT_ENTRY));
+
+ /* Link up the entry itself */
+ p = le64toh(f->header->tail_entry_offset);
+
+ o->entry.prev_entry_offset = f->header->tail_entry_offset;
+ o->entry.next_entry_offset = 0;
+
+ if (p == 0)
+ f->header->head_entry_offset = htole64(offset);
+ else {
+ /* Temporarily move back to the previous entry, to
+ * patch in pointer */
+
+ r = journal_file_move_to_object(f, p, OBJECT_ENTRY, &o);
+ if (r < 0)
+ return r;
+
+ o->entry.next_entry_offset = htole64(offset);
+
+ r = journal_file_move_to_object(f, offset, OBJECT_ENTRY, &o);
+ if (r < 0)
+ return r;
+ }
+
+ f->header->tail_entry_offset = htole64(offset);
+
+ /* Link up the items */
+ n = journal_file_entry_n_items(o);
+ for (i = 0; i < n; i++) {
+ r = journal_file_link_entry_item(f, o, offset, i);
+ if (r < 0)
+ return r;
+ }
+
+ /* Link up the entry in the bisect table */
+ n = le64toh(f->header->bisect_table_size) / sizeof(uint64_t);
+ k = le64toh(f->header->arena_max_size) / n;
+
+ a = (le64toh(f->header->last_bisect_offset) + k - 1) / k;
+ b = offset / k;
+
+ for (; a <= b; a++)
+ f->bisect_table[a] = htole64(offset);
+
+ f->header->last_bisect_offset = htole64(offset + le64toh(o->object.size));
+
+ return 0;
+}
+
+static int journal_file_append_entry_internal(
+ JournalFile *f,
+ const dual_timestamp *ts,
+ uint64_t xor_hash,
+ const EntryItem items[], unsigned n_items,
+ Object **ret, uint64_t *offset) {
+ uint64_t np;
+ uint64_t osize;
+ Object *o;
+ int r;
+
+ assert(f);
+ assert(items || n_items == 0);
+
+ osize = offsetof(Object, entry.items) + (n_items * sizeof(EntryItem));
+
+ r = journal_file_append_object(f, osize, &o, &np);
+ if (r < 0)
+ return r;
+
+ o->object.type = htole64(OBJECT_ENTRY);
+ o->entry.seqnum = htole64(journal_file_seqnum(f));
+ memcpy(o->entry.items, items, n_items * sizeof(EntryItem));
+ o->entry.realtime = ts ? htole64(ts->realtime) : 0;
+ o->entry.monotonic = ts ? htole64(ts->monotonic) : 0;
+ o->entry.xor_hash = htole64(xor_hash);
+ o->entry.boot_id = f->header->boot_id;
+
+ r = journal_file_link_entry(f, o, np);
+ if (r < 0)
+ return r;
+
+ if (ret)
+ *ret = o;
+
+ if (offset)
+ *offset = np;
+
+ return 0;
+}
+
+int journal_file_append_entry(JournalFile *f, const dual_timestamp *ts, const struct iovec iovec[], unsigned n_iovec, Object **ret, uint64_t *offset) {
+ unsigned i;
+ EntryItem *items;
+ int r;
+ uint64_t xor_hash = 0;
+
+ assert(f);
+ assert(iovec || n_iovec == 0);
+
+ items = new(EntryItem, n_iovec);
+ if (!items)
+ return -ENOMEM;
+
+ for (i = 0; i < n_iovec; i++) {
+ uint64_t p;
+ Object *o;
+
+ r = journal_file_append_data(f, iovec[i].iov_base, iovec[i].iov_len, &o, &p);
+ if (r < 0)
+ goto finish;
+
+ xor_hash ^= le64toh(o->data.hash);
+ items[i].object_offset = htole64(p);
+ }
+
+ r = journal_file_append_entry_internal(f, ts, xor_hash, items, n_iovec, ret, offset);
+
+finish:
+ free(items);
+
+ return r;
+}
+
+int journal_file_move_to_entry(JournalFile *f, uint64_t seqnum, Object **ret, uint64_t *offset) {
+ Object *o;
+ uint64_t lower, upper, p, n, k;
+ int r;
+
+ assert(f);
+
+ n = le64toh(f->header->bisect_table_size) / sizeof(uint64_t);
+ k = le64toh(f->header->arena_max_size) / n;
+
+ lower = 0;
+ upper = le64toh(f->header->last_bisect_offset)/k+1;
+
+ while (lower < upper) {
+ k = (upper + lower) / 2;
+ p = le64toh(f->bisect_table[k]);
+
+ if (p == 0) {
+ upper = k;
+ continue;
+ }
+
+ r = journal_file_move_to_object(f, p, OBJECT_ENTRY, &o);
+ if (r < 0)
+ return r;
+
+ if (o->entry.seqnum == seqnum) {
+ if (ret)
+ *ret = o;
+
+ if (offset)
+ *offset = p;
+
+ return 1;
+ } else if (seqnum < o->entry.seqnum)
+ upper = k;
+ else if (seqnum > o->entry.seqnum)
+ lower = k+1;
+ }
+
+ assert(lower == upper);
+
+ if (lower <= 0)
+ return 0;
+
+ /* The object we are looking for is between
+ * bisect_table[lower-1] and bisect_table[lower] */
+
+ p = le64toh(f->bisect_table[lower-1]);
+
+ for (;;) {
+ r = journal_file_move_to_object(f, p, OBJECT_ENTRY, &o);
+ if (r < 0)
+ return r;
+
+ if (o->entry.seqnum == seqnum) {
+ if (ret)
+ *ret = o;
+
+ if (offset)
+ *offset = p;
+
+ return 1;
+
+ } if (seqnum < o->entry.seqnum)
+ return 0;
+
+ if (o->entry.next_entry_offset == 0)
+ return 0;
+
+ p = le64toh(o->entry.next_entry_offset);
+ }
+
+ return 0;
+}
+
+int journal_file_next_entry(JournalFile *f, Object *o, Object **ret, uint64_t *offset) {
+ uint64_t np;
+ int r;
+
+ assert(f);
+
+ if (!o)
+ np = le64toh(f->header->head_entry_offset);
+ else {
+ if (le64toh(o->object.type) != OBJECT_ENTRY)
+ return -EINVAL;
+
+ np = le64toh(o->entry.next_entry_offset);
+ }
+
+ if (np == 0)
+ return 0;
+
+ r = journal_file_move_to_object(f, np, OBJECT_ENTRY, &o);
+ if (r < 0)
+ return r;
+
+ if (ret)
+ *ret = o;
+
+ if (offset)
+ *offset = np;
+
+ return 1;
+}
+
+int journal_file_prev_entry(JournalFile *f, Object *o, Object **ret, uint64_t *offset) {
+ uint64_t np;
+ int r;
+
+ assert(f);
+
+ if (!o)
+ np = le64toh(f->header->tail_entry_offset);
+ else {
+ if (le64toh(o->object.type) != OBJECT_ENTRY)
+ return -EINVAL;
+
+ np = le64toh(o->entry.prev_entry_offset);
+ }
+
+ if (np == 0)
+ return 0;
+
+ r = journal_file_move_to_object(f, np, OBJECT_ENTRY, &o);
+ if (r < 0)
+ return r;
+
+ if (ret)
+ *ret = o;
+
+ if (offset)
+ *offset = np;
+
+ return 1;
+}
+
+int journal_file_find_first_entry(JournalFile *f, const void *data, uint64_t size, Object **ret, uint64_t *offset) {
+ uint64_t p, osize, hash, h;
+ int r;
+
+ assert(f);
+ assert(data || size == 0);
+
+ osize = offsetof(Object, data.payload) + size;
+
+ hash = hash64(data, size);
+ h = hash % (le64toh(f->header->hash_table_size) / sizeof(HashItem));
+ p = le64toh(f->hash_table[h].head_hash_offset);
+
+ while (p != 0) {
+ Object *o;
+
+ r = journal_file_move_to_object(f, p, OBJECT_DATA, &o);
+ if (r < 0)
+ return r;
+
+ if (le64toh(o->object.size) == osize &&
+ memcmp(o->data.payload, data, size) == 0) {
+
+ if (le64toh(o->data.hash) != hash)
+ return -EBADMSG;
+
+ if (o->data.head_entry_offset == 0)
+ return 0;
+
+ p = le64toh(o->data.head_entry_offset);
+ r = journal_file_move_to_object(f, p, OBJECT_ENTRY, &o);
+ if (r < 0)
+ return r;
+
+ if (ret)
+ *ret = o;
+
+ if (offset)
+ *offset = p;
+
+ return 1;
+ }
+
+ p = le64toh(o->data.next_hash_offset);
+ }
+
+ return 0;
+}
+
+int journal_file_find_last_entry(JournalFile *f, const void *data, uint64_t size, Object **ret, uint64_t *offset) {
+ uint64_t p, osize, hash, h;
+ int r;
+
+ assert(f);
+ assert(data || size == 0);
+
+ osize = offsetof(Object, data.payload) + size;
+
+ hash = hash64(data, size);
+ h = hash % (le64toh(f->header->hash_table_size) / sizeof(HashItem));
+ p = le64toh(f->hash_table[h].tail_hash_offset);
+
+ while (p != 0) {
+ Object *o;
+
+ r = journal_file_move_to_object(f, p, OBJECT_DATA, &o);
+ if (r < 0)
+ return r;
+
+ if (le64toh(o->object.size) == osize &&
+ memcmp(o->data.payload, data, size) == 0) {
+
+ if (le64toh(o->data.hash) != hash)
+ return -EBADMSG;
+
+ if (o->data.tail_entry_offset == 0)
+ return 0;
+
+ p = le64toh(o->data.tail_entry_offset);
+ r = journal_file_move_to_object(f, p, OBJECT_ENTRY, &o);
+ if (r < 0)
+ return r;
+
+ if (ret)
+ *ret = o;
+
+ if (offset)
+ *offset = p;
+
+ return 1;
+ }
+
+ p = le64toh(o->data.prev_hash_offset);
+ }
+
+ return 0;
+}
+
+void journal_file_dump(JournalFile *f) {
+ char a[33], b[33], c[33];
+ Object *o;
+ int r;
+ uint64_t p;
+
+ assert(f);
+
+ printf("File ID: %s\n"
+ "Machine ID: %s\n"
+ "Boot ID: %s\n"
+ "Arena size: %llu\n",
+ 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),
+ (unsigned long long) le64toh(f->header->arena_size));
+
+ p = le64toh(f->header->head_object_offset);
+ while (p != 0) {
+ r = journal_file_move_to_object(f, p, -1, &o);
+ if (r < 0)
+ goto fail;
+
+ switch (o->object.type) {
+
+ case OBJECT_UNUSED:
+ printf("Type: OBJECT_UNUSED\n");
+ break;
+
+ case OBJECT_DATA:
+ printf("Type: OBJECT_DATA\n");
+ break;
+
+ case OBJECT_ENTRY:
+ printf("Type: OBJECT_ENTRY %llu\n", (unsigned long long) le64toh(o->entry.seqnum));
+ break;
+
+ case OBJECT_HASH_TABLE:
+ printf("Type: OBJECT_HASH_TABLE\n");
+ break;
+
+ case OBJECT_BISECT_TABLE:
+ printf("Type: OBJECT_BISECT_TABLE\n");
+ break;
+ }
+
+ if (p == le64toh(f->header->tail_object_offset))
+ p = 0;
+ else
+ p = p + ALIGN64(le64toh(o->object.size));
+ }
+
+ return;
+fail:
+ log_error("File corrupt");
+}
+
+int journal_file_open(
+ const char *fname,
+ int flags,
+ mode_t mode,
+ JournalFile **ret) {
+
+ JournalFile *f;
+ int r;
+ bool newly_created = false;
+
+ assert(fname);
+
+ if ((flags & O_ACCMODE) != O_RDONLY &&
+ (flags & O_ACCMODE) != O_RDWR)
+ return -EINVAL;
+
+ f = new0(JournalFile, 1);
+ if (!f)
+ return -ENOMEM;
+
+ f->writable = (flags & O_ACCMODE) != O_RDONLY;
+ f->prot = prot_from_flags(flags);
+
+ f->fd = open(fname, flags|O_CLOEXEC, mode);
+ if (f->fd < 0) {
+ r = -errno;
+ goto fail;
+ }
+
+ f->path = strdup(fname);
+ if (!f->path) {
+ r = -ENOMEM;
+ goto fail;
+ }
+
+ if (fstat(f->fd, &f->last_stat) < 0) {
+ r = -errno;
+ goto fail;
+ }
+
+ if (f->last_stat.st_size == 0 && f->writable) {
+ newly_created = true;
+
+ r = journal_file_init_header(f);
+ if (r < 0)
+ goto fail;
+
+ if (fstat(f->fd, &f->last_stat) < 0) {
+ r = -errno;
+ goto fail;
+ }
+ }
+
+ if (f->last_stat.st_size < (off_t) sizeof(Header)) {
+ r = -EIO;
+ goto fail;
+ }
+
+ f->header = mmap(NULL, PAGE_ALIGN(sizeof(Header)), prot_from_flags(flags), MAP_SHARED, f->fd, 0);
+ if (f->header == MAP_FAILED) {
+ f->header = NULL;
+ r = -errno;
+ goto fail;
+ }
+
+ if (!newly_created) {
+ r = journal_file_verify_header(f);
+ if (r < 0)
+ goto fail;
+ }
+
+ if (f->writable) {
+ r = journal_file_refresh_header(f);
+ if (r < 0)
+ goto fail;
+ }
+
+ if (newly_created) {
+
+ r = journal_file_setup_hash_table(f);
+ if (r < 0)
+ goto fail;
+
+ r = journal_file_setup_bisect_table(f);
+ if (r < 0)
+ goto fail;
+ }
+
+ r = journal_file_map_hash_table(f);
+ if (r < 0)
+ goto fail;
+
+ r = journal_file_map_bisect_table(f);
+ if (r < 0)
+ goto fail;
+
+ if (ret)
+ *ret = f;
+
+ return 0;
+
+fail:
+ journal_file_close(f);
+
+ return r;
+}
diff --git a/src/journal/journal-private.h b/src/journal/journal-file.h
index 3277d29542..55cc7153af 100644
--- a/src/journal/journal-private.h
+++ b/src/journal/journal-file.h
@@ -1,7 +1,7 @@
/*-*- Mode: C; c-basic-offset: 8; indent-tabs-mode: nil -*-*/
-#ifndef foojournalprivatehfoo
-#define foojournalprivatehfoo
+#ifndef foojournalfilehfoo
+#define foojournalfilehfoo
/***
This file is part of systemd.
@@ -24,7 +24,6 @@
#include <inttypes.h>
-#include "sd-journal.h"
#include "journal-def.h"
#include "util.h"
#include "sd-id128.h"
@@ -50,24 +49,25 @@ typedef struct JournalFile {
uint64_t window_offset;
uint64_t window_size;
- Object *current;
uint64_t current_offset;
} JournalFile;
-typedef struct JournalCoursor {
- sd_id128_t file_id;
- sd_id128_t boot_id;
+typedef struct JournalCursor {
+ uint8_t version;
+ uint8_t reserved[7];
uint64_t seqnum;
+ sd_id128_t seqnum_id;
+ sd_id128_t boot_id;
uint64_t monotonic;
uint64_t realtime;
uint64_t xor_hash;
-} JournalCoursor;
+} JournalCursor;
int journal_file_open(const char *fname, int flags, mode_t mode, JournalFile **ret);
void journal_file_close(JournalFile *j);
-int journal_file_move_to_object(JournalFile *f, uint64_t offset, Object **ret);
+int journal_file_move_to_object(JournalFile *f, uint64_t offset, int type, Object **ret);
uint64_t journal_file_entry_n_items(Object *o);
diff --git a/src/journal/journalctl.c b/src/journal/journalctl.c
index 7bcd842f6d..5f17f45cac 100644
--- a/src/journal/journalctl.c
+++ b/src/journal/journalctl.c
@@ -23,7 +23,7 @@
#include <errno.h>
#include <stddef.h>
-#include "journal-private.h"
+#include "journal-file.h"
int main(int argc, char *argv[]) {
int r;
@@ -62,21 +62,16 @@ int main(int argc, char *argv[]) {
uint64_t p, l;
p = le64toh(o->entry.items[i].object_offset);
- r = journal_file_move_to_object(f, p, &o);
+ r = journal_file_move_to_object(f, p, OBJECT_DATA, &o);
if (r < 0) {
log_error("Failed to move to data: %s", strerror(-r));
goto finish;
}
- if (le64toh(o->object.type) != OBJECT_DATA) {
- log_error("Invalid file");
- goto finish;
- }
-
l = o->object.size - offsetof(Object, data.payload);
printf("\t[%.*s]\n", (int) l, o->data.payload);
- r = journal_file_move_to_object(f, offset, &o);
+ r = journal_file_move_to_object(f, offset, OBJECT_ENTRY, &o);
if (r < 0) {
log_error("Failed to move back to entry: %s", strerror(-r));
goto finish;
diff --git a/src/journal/journald.c b/src/journal/journald.c
index e9ac3a832e..d65451df58 100644
--- a/src/journal/journald.c
+++ b/src/journal/journald.c
@@ -29,7 +29,7 @@
#include <acl/libacl.h>
#include "hashmap.h"
-#include "journal-private.h"
+#include "journal-file.h"
#include "sd-daemon.h"
#include "socket-util.h"
#include "acl-util.h"
@@ -282,7 +282,9 @@ static int process_event(Server *s, struct epoll_event *ev) {
log_debug("Received SIG%s", signal_to_string(sfsi.ssi_signo));
return 0;
- } else {
+ }
+
+ if (ev->data.fd == s->syslog_fd) {
for (;;) {
char buf[LINE_MAX+1];
struct msghdr msghdr;
@@ -339,9 +341,12 @@ static int process_event(Server *s, struct epoll_event *ev) {
process_message(s, strstrip(buf), ucred, tv);
}
+
+ return 1;
}
- return 1;
+ log_error("Unknown event.");
+ return 0;
}
static int server_init(Server *s) {
diff --git a/src/journal/sd-journal.c b/src/journal/sd-journal.c
index 89bf545837..8426b3bf9e 100644
--- a/src/journal/sd-journal.c
+++ b/src/journal/sd-journal.c
@@ -19,1206 +19,223 @@
along with systemd; If not, see <http://www.gnu.org/licenses/>.
***/
-#include <sys/mman.h>
#include <errno.h>
-#include <sys/uio.h>
-#include <unistd.h>
-#include <sys/statvfs.h>
#include <fcntl.h>
-#include <stddef.h>
#include "sd-journal.h"
#include "journal-def.h"
-#include "journal-private.h"
-#include "lookup3.h"
-#include "list.h"
+#include "journal-file.h"
#include "hashmap.h"
+#include "list.h"
-#define DEFAULT_ARENA_MAX_SIZE (16ULL*1024ULL*1024ULL*1024ULL)
-#define DEFAULT_ARENA_MIN_SIZE (256ULL*1024ULL)
-#define DEFAULT_ARENA_KEEP_FREE (1ULL*1024ULL*1024ULL)
+typedef struct Match Match;
-#define DEFAULT_HASH_TABLE_SIZE (2047ULL*16ULL)
-#define DEFAULT_BISECT_TABLE_SIZE ((DEFAULT_ARENA_MAX_SIZE/(64ULL*1024ULL))*8ULL)
+struct Match {
+ char *data;
+ size_t size;
+ uint64_t hash;
-#define DEFAULT_WINDOW_SIZE (128ULL*1024ULL*1024ULL)
+ LIST_FIELDS(Match, matches);
+};
struct sd_journal {
Hashmap *files;
-};
-
-static const char signature[] = { 'L', 'P', 'K', 'S', 'H', 'H', 'R', 'H' };
-
-#define ALIGN64(x) (((x) + 7ULL) & ~7ULL)
-
-void journal_file_close(JournalFile *f) {
- assert(f);
-
- if (f->fd >= 0)
- close_nointr_nofail(f->fd);
-
- if (f->header)
- munmap(f->header, PAGE_ALIGN(sizeof(Header)));
-
- if (f->hash_table_window)
- munmap(f->hash_table_window, f->hash_table_window_size);
-
- if (f->bisect_table_window)
- munmap(f->bisect_table_window, f->bisect_table_window_size);
-
- if (f->window)
- munmap(f->window, f->window_size);
-
- free(f->path);
- free(f);
-}
-
-static int journal_file_init_header(JournalFile *f) {
- Header h;
- ssize_t k;
- int r;
-
- assert(f);
-
- zero(h);
- memcpy(h.signature, signature, 8);
- h.arena_offset = htole64(ALIGN64(sizeof(h)));
- h.arena_max_size = htole64(DEFAULT_ARENA_MAX_SIZE);
- h.arena_min_size = htole64(DEFAULT_ARENA_MIN_SIZE);
- h.arena_keep_free = htole64(DEFAULT_ARENA_KEEP_FREE);
-
- r = sd_id128_randomize(&h.file_id);
- if (r < 0)
- return r;
-
- k = pwrite(f->fd, &h, sizeof(h), 0);
- if (k < 0)
- return -errno;
-
- if (k != sizeof(h))
- return -EIO;
-
- return 0;
-}
-
-static int journal_file_refresh_header(JournalFile *f) {
- int r;
-
- assert(f);
-
- r = sd_id128_get_machine(&f->header->machine_id);
- if (r < 0)
- return r;
-
- r = sd_id128_get_boot(&f->header->boot_id);
- if (r < 0)
- return r;
-
- f->header->state = htole32(STATE_ONLINE);
- return 0;
-}
-
-static int journal_file_verify_header(JournalFile *f) {
- assert(f);
-
- if (memcmp(f->header, signature, 8))
- return -EBADMSG;
-
- if (f->header->incompatible_flags != 0)
- return -EPROTONOSUPPORT;
-
- if ((uint64_t) f->last_stat.st_size < (le64toh(f->header->arena_offset) + le64toh(f->header->arena_size)))
- return -ENODATA;
-
- if (f->writable) {
- uint32_t state;
- sd_id128_t machine_id;
- int r;
-
- r = sd_id128_get_machine(&machine_id);
- if (r < 0)
- return r;
-
- if (!sd_id128_equal(machine_id, f->header->machine_id))
- return -EHOSTDOWN;
-
- state = le32toh(f->header->state);
-
- if (state == STATE_ONLINE)
- log_debug("Journal file %s is already online. Assuming unclean closing. Ignoring.", f->path);
- else if (state == STATE_ARCHIVED)
- return -ESHUTDOWN;
- else if (state != STATE_OFFLINE)
- log_debug("Journal file %s has unknown state %u. Ignoring.", f->path, state);
- }
-
- return 0;
-}
-
-static int journal_file_allocate(JournalFile *f, uint64_t offset, uint64_t size) {
- uint64_t asize;
- uint64_t old_size, new_size;
-
- assert(f);
-
- if (offset < le64toh(f->header->arena_offset))
- return -EINVAL;
-
- new_size = PAGE_ALIGN(offset + size);
-
- /* We assume that this file is not sparse, and we know that
- * for sure, since we alway call posix_fallocate()
- * ourselves */
-
- old_size =
- le64toh(f->header->arena_offset) +
- le64toh(f->header->arena_size);
-
- if (old_size >= new_size)
- return 0;
-
- asize = new_size - le64toh(f->header->arena_offset);
- if (asize > le64toh(f->header->arena_min_size)) {
- struct statvfs svfs;
-
- if (fstatvfs(f->fd, &svfs) >= 0) {
- uint64_t available;
-
- available = svfs.f_bfree * svfs.f_bsize;
-
- if (available >= f->header->arena_keep_free)
- available -= f->header->arena_keep_free;
- else
- available = 0;
-
- if (new_size - old_size > available)
- return -E2BIG;
- }
- }
-
- if (asize > le64toh(f->header->arena_max_size))
- return -E2BIG;
-
- if (posix_fallocate(f->fd, 0, new_size) < 0)
- return -errno;
-
- if (fstat(f->fd, &f->last_stat) < 0)
- return -errno;
-
- f->header->arena_size = htole64(asize);
-
- return 0;
-}
-
-static int journal_file_map(
- JournalFile *f,
- uint64_t offset,
- uint64_t size,
- void **_window,
- uint64_t *_woffset,
- uint64_t *_wsize,
- void **ret) {
-
- uint64_t woffset, wsize;
- void *window;
-
- assert(f);
- assert(size > 0);
- assert(ret);
-
- woffset = offset & ~((uint64_t) page_size() - 1ULL);
- wsize = size + (offset - woffset);
- wsize = PAGE_ALIGN(wsize);
-
- window = mmap(NULL, wsize, f->prot, MAP_SHARED, f->fd, woffset);
- if (window == MAP_FAILED)
- return -errno;
-
- if (_window)
- *_window = window;
-
- if (_woffset)
- *_woffset = woffset;
-
- if (_wsize)
- *_wsize = wsize;
-
- *ret = (uint8_t*) window + (offset - woffset);
-
- return 0;
-}
-
-static int journal_file_move_to(JournalFile *f, uint64_t offset, uint64_t size, void **ret) {
- void *p;
- uint64_t delta;
- int r;
-
- assert(f);
- assert(ret);
-
- if (_likely_(f->window &&
- f->window_offset <= offset &&
- f->window_offset+f->window_size >= offset + size)) {
-
- *ret = (uint8_t*) f->window + (offset - f->window_offset);
- return 0;
- }
-
- if (f->window) {
- if (munmap(f->window, f->window_size) < 0)
- return -errno;
-
- f->window = NULL;
- f->window_size = f->window_offset = 0;
- }
-
- if (size < DEFAULT_WINDOW_SIZE) {
- /* If the default window size is larger then what was
- * asked for extend the mapping a bit in the hope to
- * minimize needed remappings later on. We add half
- * the window space before and half behind the
- * requested mapping */
-
- delta = PAGE_ALIGN((DEFAULT_WINDOW_SIZE - size) / 2);
-
- if (offset < delta)
- delta = offset;
-
- offset -= delta;
- size += (DEFAULT_WINDOW_SIZE - delta);
- } else
- delta = 0;
-
- r = journal_file_map(f,
- offset, size,
- &f->window, &f->window_offset, &f->window_size,
- & p);
-
- if (r < 0)
- return r;
-
- *ret = (uint8_t*) p + delta;
- return 0;
-}
-
-static bool verify_hash(Object *o) {
- uint64_t t;
-
- assert(o);
-
- t = le64toh(o->object.type);
- if (t == OBJECT_DATA) {
- uint64_t s, h1, h2;
-
- s = le64toh(o->object.size);
-
- h1 = le64toh(o->data.hash);
- h2 = hash64(o->data.payload, s - offsetof(Object, data.payload));
-
- return h1 == h2;
- }
-
- return true;
-}
-
-int journal_file_move_to_object(JournalFile *f, uint64_t offset, Object **ret) {
- int r;
- void *t;
- Object *o;
- uint64_t s;
-
- assert(f);
- assert(ret);
+ JournalFile *current_file;
- r = journal_file_move_to(f, offset, sizeof(ObjectHeader), &t);
- if (r < 0)
- return r;
+ LIST_HEAD(Match, matches);
+};
- o = (Object*) t;
- s = le64toh(o->object.size);
+int sd_journal_add_match(sd_journal *j, const char *field, const void *data, size_t size) {
+ Match *m;
+ char *e;
- if (s < sizeof(ObjectHeader))
- return -EBADMSG;
+ assert(j);
+ assert(field);
+ assert(data || size == 0);
- if (s > sizeof(ObjectHeader)) {
- r = journal_file_move_to(f, offset, s, &t);
- if (r < 0)
- return r;
+ m = new0(Match, 1);
+ if (!m)
+ return -ENOMEM;
- o = (Object*) t;
+ m->size = strlen(field) + 1 + size;
+ m->data = malloc(m->size);
+ if (!m->data) {
+ free(m);
+ return -ENOMEM;
}
- if (!verify_hash(o))
- return -EBADMSG;
+ e = stpcpy(m->data, field);
+ *(e++) = '=';
+ memcpy(e, data, size);
- *ret = o;
+ LIST_PREPEND(Match, matches, j->matches, m);
return 0;
}
-static uint64_t journal_file_seqnum(JournalFile *f) {
- uint64_t r;
-
- assert(f);
-
- r = le64toh(f->header->seqnum) + 1;
- f->header->seqnum = htole64(r);
-
- return r;
-}
-
-static int journal_file_append_object(JournalFile *f, uint64_t size, Object **ret, uint64_t *offset) {
- int r;
- uint64_t p;
- Object *tail, *o;
- void *t;
-
- assert(f);
- assert(size >= sizeof(ObjectHeader));
- assert(offset);
- assert(ret);
-
- p = le64toh(f->header->tail_object_offset);
+void sd_journal_flush_matches(sd_journal *j) {
+ assert(j);
- if (p == 0)
- p = le64toh(f->header->arena_offset);
- else {
- r = journal_file_move_to_object(f, p, &tail);
- if (r < 0)
- return r;
+ while (j->matches) {
+ Match *m = j->matches;
- p += ALIGN64(le64toh(tail->object.size));
+ LIST_REMOVE(Match, matches, j->matches, m);
+ free(m->data);
+ free(m);
}
-
- r = journal_file_allocate(f, p, size);
- if (r < 0)
- return r;
-
- r = journal_file_move_to(f, p, size, &t);
- if (r < 0)
- return r;
-
- o = (Object*) t;
-
- zero(o->object);
- o->object.type = htole64(OBJECT_UNUSED);
- zero(o->object.reserved);
- o->object.size = htole64(size);
-
- f->header->tail_object_offset = htole64(p);
- if (f->header->head_object_offset == 0)
- f->header->head_object_offset = htole64(p);
-
- f->header->n_objects = htole64(le64toh(f->header->n_objects) + 1);
-
- *ret = o;
- *offset = p;
-
- return 0;
-}
-
-static int journal_file_setup_hash_table(JournalFile *f) {
- uint64_t s, p;
- Object *o;
- int r;
-
- assert(f);
-
- s = DEFAULT_HASH_TABLE_SIZE;
- r = journal_file_append_object(f, offsetof(Object, hash_table.table) + s, &o, &p);
- if (r < 0)
- return r;
-
- o->object.type = htole64(OBJECT_HASH_TABLE);
- memset(o->hash_table.table, 0, s);
-
- f->header->hash_table_offset = htole64(p + offsetof(Object, hash_table.table));
- f->header->hash_table_size = htole64(s);
-
- return 0;
-}
-
-static int journal_file_setup_bisect_table(JournalFile *f) {
- uint64_t s, p;
- Object *o;
- int r;
-
- assert(f);
-
- s = DEFAULT_BISECT_TABLE_SIZE;
- r = journal_file_append_object(f, offsetof(Object, bisect_table.table) + s, &o, &p);
- if (r < 0)
- return r;
-
- o->object.type = htole64(OBJECT_BISECT_TABLE);
- memset(o->bisect_table.table, 0, s);
-
- f->header->bisect_table_offset = htole64(p + offsetof(Object, bisect_table.table));
- f->header->bisect_table_size = htole64(s);
-
- return 0;
-}
-
-static int journal_file_map_hash_table(JournalFile *f) {
- uint64_t s, p;
- void *t;
- int r;
-
- assert(f);
-
- p = le64toh(f->header->hash_table_offset);
- s = le64toh(f->header->hash_table_size);
-
- r = journal_file_map(f,
- p, s,
- &f->hash_table_window, NULL, &f->hash_table_window_size,
- &t);
- if (r < 0)
- return r;
-
- f->hash_table = t;
- return 0;
}
-static int journal_file_map_bisect_table(JournalFile *f) {
- uint64_t s, p;
- void *t;
- int r;
-
- assert(f);
-
- p = le64toh(f->header->bisect_table_offset);
- s = le64toh(f->header->bisect_table_size);
+static int compare_order(JournalFile *af, Object *ao, uint64_t ap,
+ JournalFile *bf, Object *bo, uint64_t bp) {
- r = journal_file_map(f,
- p, s,
- &f->bisect_table_window, NULL, &f->bisect_table_window_size,
- &t);
-
- if (r < 0)
- return r;
+ uint64_t a, b;
- f->bisect_table = t;
- return 0;
-}
+ if (sd_id128_equal(af->header->seqnum_id, bf->header->seqnum_id)) {
-static int journal_file_link_data(JournalFile *f, Object *o, uint64_t offset, uint64_t hash_index) {
- uint64_t p;
- int r;
+ /* If this is from the same seqnum source, compare
+ * seqnums */
+ a = le64toh(ao->entry.seqnum);
+ b = le64toh(bo->entry.seqnum);
- assert(f);
- assert(o);
- assert(offset > 0);
- assert(o->object.type == htole64(OBJECT_DATA));
- o->data.head_entry_offset = o->data.tail_entry_offset = 0;
- o->data.next_hash_offset = 0;
+ } else if (sd_id128_equal(ao->entry.boot_id, bo->entry.boot_id)) {
- p = le64toh(f->hash_table[hash_index].tail_hash_offset);
- if (p == 0) {
- /* Only entry in the hash table is easy */
+ /* If the boot id matches compare monotonic time */
+ a = le64toh(ao->entry.monotonic);
+ b = le64toh(bo->entry.monotonic);
- o->data.prev_hash_offset = 0;
- f->hash_table[hash_index].head_hash_offset = htole64(offset);
} else {
- o->data.prev_hash_offset = htole64(p);
- /* Temporarily move back to the previous data object,
- * to patch in pointer */
-
- r = journal_file_move_to_object(f, p, &o);
- if (r < 0)
- return r;
-
- o->data.next_hash_offset = offset;
-
- r = journal_file_move_to_object(f, offset, &o);
- if (r < 0)
- return r;
+ /* Otherwise compare UTC time */
+ a = le64toh(ao->entry.realtime);
+ b = le64toh(ao->entry.realtime);
}
- f->hash_table[hash_index].tail_hash_offset = htole64(offset);
-
- return 0;
+ return
+ a < b ? -1 :
+ a > b ? +1 : 0;
}
-static int journal_file_append_data(JournalFile *f, const void *data, uint64_t size, Object **ret, uint64_t *offset) {
- uint64_t hash, h, p, np;
- uint64_t osize;
- Object *o;
+int sd_journal_next(sd_journal *j) {
+ JournalFile *f, *new_current = NULL;
+ Iterator i;
int r;
+ uint64_t new_offset = 0;
+ Object *new_entry = NULL;
- assert(f);
- assert(data || size == 0);
-
- osize = offsetof(Object, data.payload) + size;
+ assert(j);
- hash = hash64(data, size);
- h = hash % (le64toh(f->header->hash_table_size) / sizeof(HashItem));
- p = le64toh(f->hash_table[h].head_hash_offset);
+ HASHMAP_FOREACH(f, j->files, i) {
+ Object *o;
+ uint64_t p;
- while (p != 0) {
- /* Look for this data object in the hash table */
+ if (f->current_offset > 0) {
+ r = journal_file_move_to_object(f, f->current_offset, OBJECT_ENTRY, &o);
+ if (r < 0)
+ return r;
+ } else
+ o = NULL;
- r = journal_file_move_to_object(f, p, &o);
+ r = journal_file_next_entry(f, o, &o, &p);
if (r < 0)
return r;
+ else if (r == 0)
+ continue;
- if (le64toh(o->object.type) != OBJECT_DATA)
- return -EBADMSG;
-
- if (le64toh(o->object.size) == osize &&
- memcmp(o->data.payload, data, size) == 0) {
-
- if (le64toh(o->data.hash) != hash)
- return -EBADMSG;
-
- if (ret)
- *ret = o;
-
- if (offset)
- *offset = p;
-
- return 0;
+ if (!new_current || compare_order(new_current, new_entry, new_offset, f, o, p) > 0) {
+ new_current = f;
+ new_entry = o;
+ new_offset = p;
}
-
- p = le64toh(o->data.next_hash_offset);
}
- r = journal_file_append_object(f, osize, &o, &np);
- if (r < 0)
- return r;
-
- o->object.type = htole64(OBJECT_DATA);
- o->data.hash = htole64(hash);
- memcpy(o->data.payload, data, size);
-
- r = journal_file_link_data(f, o, np, h);
- if (r < 0)
- return r;
-
- if (ret)
- *ret = o;
-
- if (offset)
- *offset = np;
-
- return 0;
-}
-
-uint64_t journal_file_entry_n_items(Object *o) {
- assert(o);
- assert(o->object.type == htole64(OBJECT_ENTRY));
-
- return (le64toh(o->object.size) - offsetof(Object, entry.items)) / sizeof(EntryItem);
-}
-
-static int journal_file_link_entry_item(JournalFile *f, Object *o, uint64_t offset, uint64_t i) {
- uint64_t p, q;
- int r;
- assert(f);
- assert(o);
- assert(offset > 0);
-
- p = le64toh(o->entry.items[i].object_offset);
- if (p == 0)
- return -EINVAL;
-
- o->entry.items[i].next_entry_offset = 0;
-
- /* Move to the data object */
- r = journal_file_move_to_object(f, p, &o);
- if (r < 0)
- return r;
-
- if (o->object.type != htole64(OBJECT_DATA))
- return -EBADMSG;
-
- q = le64toh(o->data.tail_entry_offset);
- o->data.tail_entry_offset = htole64(offset);
-
- if (q == 0)
- o->data.head_entry_offset = htole64(offset);
- else {
- uint64_t n, j;
-
- /* Move to previous entry */
- r = journal_file_move_to_object(f, q, &o);
- if (r < 0)
- return r;
-
- if (o->object.type != htole64(OBJECT_ENTRY))
- return -EBADMSG;
-
- n = journal_file_entry_n_items(o);
- for (j = 0; j < n; j++)
- if (le64toh(o->entry.items[j].object_offset) == p)
- break;
-
- if (j >= n)
- return -EBADMSG;
-
- o->entry.items[j].next_entry_offset = offset;
- }
-
- /* Move back to original entry */
- r = journal_file_move_to_object(f, offset, &o);
- if (r < 0)
- return r;
-
- o->entry.items[i].prev_entry_offset = q;
- return 0;
-}
-
-static int journal_file_link_entry(JournalFile *f, Object *o, uint64_t offset) {
- uint64_t p, i, n, k, a, b;
- int r;
-
- assert(f);
- assert(o);
- assert(offset > 0);
- assert(o->object.type == htole64(OBJECT_ENTRY));
-
- /* Link up the entry itself */
- p = le64toh(f->header->tail_entry_offset);
-
- o->entry.prev_entry_offset = f->header->tail_entry_offset;
- o->entry.next_entry_offset = 0;
-
- if (p == 0)
- f->header->head_entry_offset = htole64(offset);
- else {
- /* Temporarily move back to the previous entry, to
- * patch in pointer */
-
- r = journal_file_move_to_object(f, p, &o);
- if (r < 0)
- return r;
-
- o->entry.next_entry_offset = htole64(offset);
-
- r = journal_file_move_to_object(f, offset, &o);
- if (r < 0)
- return r;
- }
-
- f->header->tail_entry_offset = htole64(offset);
-
- /* Link up the items */
- n = journal_file_entry_n_items(o);
- for (i = 0; i < n; i++) {
- r = journal_file_link_entry_item(f, o, offset, i);
- if (r < 0)
- return r;
+ if (new_current) {
+ j->current_file = new_current;
+ f->current_offset = new_offset;
+ return 1;
}
- /* Link up the entry in the bisect table */
- n = le64toh(f->header->bisect_table_size) / sizeof(uint64_t);
- k = le64toh(f->header->arena_max_size) / n;
-
- a = (le64toh(f->header->last_bisect_offset) + k - 1) / k;
- b = offset / k;
-
- for (; a <= b; a++)
- f->bisect_table[a] = htole64(offset);
-
- f->header->last_bisect_offset = htole64(offset + le64toh(o->object.size));
-
return 0;
}
-static int journal_file_append_entry_internal(
- JournalFile *f,
- const dual_timestamp *ts,
- uint64_t xor_hash,
- const EntryItem items[], unsigned n_items,
- Object **ret, uint64_t *offset) {
- uint64_t np;
- uint64_t osize;
- Object *o;
+int sd_journal_previous(sd_journal *j) {
+ JournalFile *f, *new_current = NULL;
+ Iterator i;
int r;
+ uint64_t new_offset = 0;
+ Object *new_entry = NULL;
- assert(f);
- assert(items || n_items == 0);
-
- osize = offsetof(Object, entry.items) + (n_items * sizeof(EntryItem));
-
- r = journal_file_append_object(f, osize, &o, &np);
- if (r < 0)
- return r;
-
- o->object.type = htole64(OBJECT_ENTRY);
- o->entry.seqnum = htole64(journal_file_seqnum(f));
- memcpy(o->entry.items, items, n_items * sizeof(EntryItem));
- o->entry.realtime = ts ? htole64(ts->realtime) : 0;
- o->entry.monotonic = ts ? htole64(ts->monotonic) : 0;
- o->entry.xor_hash = htole64(xor_hash);
-
- r = journal_file_link_entry(f, o, np);
- if (r < 0)
- return r;
-
- if (ret)
- *ret = o;
-
- if (offset)
- *offset = np;
-
- return 0;
-}
-
-int journal_file_append_entry(JournalFile *f, const dual_timestamp *ts, const struct iovec iovec[], unsigned n_iovec, Object **ret, uint64_t *offset) {
- unsigned i;
- EntryItem *items;
- int r;
- uint64_t xor_hash = 0;
-
- assert(f);
- assert(iovec || n_iovec == 0);
-
- items = new(EntryItem, n_iovec);
- if (!items)
- return -ENOMEM;
+ assert(j);
- for (i = 0; i < n_iovec; i++) {
- uint64_t p;
+ HASHMAP_FOREACH(f, j->files, i) {
Object *o;
+ uint64_t p;
- r = journal_file_append_data(f, iovec[i].iov_base, iovec[i].iov_len, &o, &p);
- if (r < 0)
- goto finish;
-
- xor_hash ^= le64toh(o->data.hash);
- items[i].object_offset = htole64(p);
- }
-
- r = journal_file_append_entry_internal(f, ts, xor_hash, items, n_iovec, ret, offset);
-
-finish:
- free(items);
-
- return r;
-}
-
-int journal_file_move_to_entry(JournalFile *f, uint64_t seqnum, Object **ret, uint64_t *offset) {
- Object *o;
- uint64_t lower, upper, p, n, k;
- int r;
-
- assert(f);
-
- n = le64toh(f->header->bisect_table_size) / sizeof(uint64_t);
- k = le64toh(f->header->arena_max_size) / n;
-
- lower = 0;
- upper = le64toh(f->header->last_bisect_offset)/k+1;
-
- while (lower < upper) {
- k = (upper + lower) / 2;
- p = le64toh(f->bisect_table[k]);
-
- if (p == 0) {
- upper = k;
- continue;
- }
+ if (f->current_offset > 0) {
+ r = journal_file_move_to_object(f, f->current_offset, OBJECT_ENTRY, &o);
+ if (r < 0)
+ return r;
+ } else
+ o = NULL;
- r = journal_file_move_to_object(f, p, &o);
+ r = journal_file_prev_entry(f, o, &o, &p);
if (r < 0)
return r;
+ else if (r == 0)
+ continue;
- if (o->object.type != htole64(OBJECT_ENTRY))
- return -EBADMSG;
-
- if (o->entry.seqnum == seqnum) {
- if (ret)
- *ret = o;
-
- if (offset)
- *offset = p;
-
- return 1;
- } else if (seqnum < o->entry.seqnum)
- upper = k;
- else if (seqnum > o->entry.seqnum)
- lower = k+1;
+ if (!new_current || compare_order(new_current, new_entry, new_offset, f, o, p) > 0) {
+ new_current = f;
+ new_entry = o;
+ new_offset = p;
+ }
}
- assert(lower == upper);
-
- if (lower <= 0)
- return 0;
-
- /* The object we are looking for is between
- * bisect_table[lower-1] and bisect_table[lower] */
-
- p = le64toh(f->bisect_table[lower-1]);
-
- for (;;) {
- r = journal_file_move_to_object(f, p, &o);
- if (r < 0)
- return r;
-
- if (o->entry.seqnum == seqnum) {
- if (ret)
- *ret = o;
-
- if (offset)
- *offset = p;
-
- return 1;
-
- } if (seqnum < o->entry.seqnum)
- return 0;
-
- if (o->entry.next_entry_offset == 0)
- return 0;
-
- p = le64toh(o->entry.next_entry_offset);
+ if (new_current) {
+ j->current_file = new_current;
+ f->current_offset = new_offset;
+ return 1;
}
return 0;
}
-int journal_file_next_entry(JournalFile *f, Object *o, Object **ret, uint64_t *offset) {
- uint64_t np;
- int r;
-
- assert(f);
-
- if (!o)
- np = le64toh(f->header->head_entry_offset);
- else {
- if (le64toh(o->object.type) != OBJECT_ENTRY)
- return -EINVAL;
-
- np = le64toh(o->entry.next_entry_offset);
- }
-
- if (np == 0)
- return 0;
-
- r = journal_file_move_to_object(f, np, &o);
- if (r < 0)
- return r;
-
- if (le64toh(o->object.type) != OBJECT_ENTRY)
- return -EBADMSG;
-
- if (ret)
- *ret = o;
-
- if (offset)
- *offset = np;
-
- return 1;
-}
-
-int journal_file_prev_entry(JournalFile *f, Object *o, Object **ret, uint64_t *offset) {
- uint64_t np;
+int sd_journal_get_cursor(sd_journal *j, void **cursor, size_t *size) {
+ JournalCursor *c;
+ Object *o;
int r;
- assert(f);
-
- if (!o)
- np = le64toh(f->header->tail_entry_offset);
- else {
- if (le64toh(o->object.type) != OBJECT_ENTRY)
- return -EINVAL;
-
- np = le64toh(o->entry.prev_entry_offset);
- }
+ assert(j);
+ assert(cursor);
+ assert(size);
- if (np == 0)
+ if (!j->current_file || !j->current_file->current_offset <= 0)
return 0;
- r = journal_file_move_to_object(f, np, &o);
+ r = journal_file_move_to_object(j->current_file, j->current_file->current_offset, OBJECT_ENTRY, &o);
if (r < 0)
return r;
- if (le64toh(o->object.type) != OBJECT_ENTRY)
- return -EBADMSG;
+ c = new0(JournalCursor, 1);
+ if (!c)
+ return -ENOMEM;
- if (ret)
- *ret = o;
+ c->version = 1;
+ c->seqnum = o->entry.seqnum;
+ c->seqnum_id = j->current_file->header->seqnum_id;
+ c->boot_id = o->entry.boot_id;
+ c->monotonic = o->entry.monotonic;
+ c->realtime = o->entry.realtime;
+ c->xor_hash = o->entry.xor_hash;
- if (offset)
- *offset = np;
+ *cursor = c;
+ *size = sizeof(JournalCursor);
return 1;
}
-int journal_file_find_first_entry(JournalFile *f, const void *data, uint64_t size, Object **ret, uint64_t *offset) {
- uint64_t p, osize, hash, h;
- int r;
-
- assert(f);
- assert(data || size == 0);
-
- osize = offsetof(Object, data.payload) + size;
-
- hash = hash64(data, size);
- h = hash % (le64toh(f->header->hash_table_size) / sizeof(HashItem));
- p = le64toh(f->hash_table[h].head_hash_offset);
-
- while (p != 0) {
- Object *o;
-
- r = journal_file_move_to_object(f, p, &o);
- if (r < 0)
- return r;
-
- if (le64toh(o->object.type) != OBJECT_DATA)
- return -EBADMSG;
-
- if (le64toh(o->object.size) == osize &&
- memcmp(o->data.payload, data, size) == 0) {
-
- if (le64toh(o->data.hash) != hash)
- return -EBADMSG;
-
- if (o->data.head_entry_offset == 0)
- return 0;
-
- p = le64toh(o->data.head_entry_offset);
- r = journal_file_move_to_object(f, p, &o);
- if (r < 0)
- return r;
-
- if (le64toh(o->object.type) != OBJECT_ENTRY)
- return -EBADMSG;
-
- if (ret)
- *ret = o;
-
- if (offset)
- *offset = p;
-
- return 1;
- }
-
- p = le64toh(o->data.next_hash_offset);
- }
-
- return 0;
-}
-
-int journal_file_find_last_entry(JournalFile *f, const void *data, uint64_t size, Object **ret, uint64_t *offset) {
- uint64_t p, osize, hash, h;
- int r;
-
- assert(f);
- assert(data || size == 0);
-
- osize = offsetof(Object, data.payload) + size;
-
- hash = hash64(data, size);
- h = hash % (le64toh(f->header->hash_table_size) / sizeof(HashItem));
- p = le64toh(f->hash_table[h].tail_hash_offset);
-
- while (p != 0) {
- Object *o;
-
- r = journal_file_move_to_object(f, p, &o);
- if (r < 0)
- return r;
-
- if (le64toh(o->object.type) != OBJECT_DATA)
- return -EBADMSG;
-
- if (le64toh(o->object.size) == osize &&
- memcmp(o->data.payload, data, size) == 0) {
-
- if (le64toh(o->data.hash) != hash)
- return -EBADMSG;
-
- if (o->data.tail_entry_offset == 0)
- return 0;
-
- p = le64toh(o->data.tail_entry_offset);
- r = journal_file_move_to_object(f, p, &o);
- if (r < 0)
- return r;
-
- if (le64toh(o->object.type) != OBJECT_ENTRY)
- return -EBADMSG;
-
- if (ret)
- *ret = o;
-
- if (offset)
- *offset = p;
-
- return 1;
- }
-
- p = le64toh(o->data.prev_hash_offset);
- }
-
- return 0;
-}
-
-void journal_file_dump(JournalFile *f) {
- char a[33], b[33], c[33];
- Object *o;
- int r;
- uint64_t p;
-
- assert(f);
-
- printf("File ID: %s\n"
- "Machine ID: %s\n"
- "Boot ID: %s\n"
- "Arena size: %llu\n",
- 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),
- (unsigned long long) le64toh(f->header->arena_size));
-
- p = le64toh(f->header->head_object_offset);
- while (p != 0) {
- r = journal_file_move_to_object(f, p, &o);
- if (r < 0)
- goto fail;
-
- switch (o->object.type) {
-
- case OBJECT_UNUSED:
- printf("Type: OBJECT_UNUSED\n");
- break;
-
- case OBJECT_DATA:
- printf("Type: OBJECT_DATA\n");
- break;
-
- case OBJECT_ENTRY:
- printf("Type: OBJECT_ENTRY %llu\n", (unsigned long long) le64toh(o->entry.seqnum));
- break;
-
- case OBJECT_HASH_TABLE:
- printf("Type: OBJECT_HASH_TABLE\n");
- break;
-
- case OBJECT_BISECT_TABLE:
- printf("Type: OBJECT_BISECT_TABLE\n");
- break;
- }
-
- if (p == le64toh(f->header->tail_object_offset))
- p = 0;
- else
- p = p + ALIGN64(le64toh(o->object.size));
- }
-
- return;
-fail:
- log_error("File corrupt");
-}
-
-int journal_file_open(
- const char *fname,
- int flags,
- mode_t mode,
- JournalFile **ret) {
-
- JournalFile *f;
- int r;
- bool newly_created = false;
-
- assert(fname);
-
- if ((flags & O_ACCMODE) != O_RDONLY &&
- (flags & O_ACCMODE) != O_RDWR)
- return -EINVAL;
-
- f = new0(JournalFile, 1);
- if (!f)
- return -ENOMEM;
-
- f->writable = (flags & O_ACCMODE) != O_RDONLY;
- f->prot = prot_from_flags(flags);
-
- f->fd = open(fname, flags|O_CLOEXEC, mode);
- if (f->fd < 0) {
- r = -errno;
- goto fail;
- }
-
- f->path = strdup(fname);
- if (!f->path) {
- r = -ENOMEM;
- goto fail;
- }
-
- if (fstat(f->fd, &f->last_stat) < 0) {
- r = -errno;
- goto fail;
- }
-
- if (f->last_stat.st_size == 0 && f->writable) {
- newly_created = true;
-
- r = journal_file_init_header(f);
- if (r < 0)
- goto fail;
-
- if (fstat(f->fd, &f->last_stat) < 0) {
- r = -errno;
- goto fail;
- }
- }
-
- if (f->last_stat.st_size < (off_t) sizeof(Header)) {
- r = -EIO;
- goto fail;
- }
-
- f->header = mmap(NULL, PAGE_ALIGN(sizeof(Header)), prot_from_flags(flags), MAP_SHARED, f->fd, 0);
- if (f->header == MAP_FAILED) {
- f->header = NULL;
- r = -errno;
- goto fail;
- }
-
- if (!newly_created) {
- r = journal_file_verify_header(f);
- if (r < 0)
- goto fail;
- }
-
- if (f->writable) {
- r = journal_file_refresh_header(f);
- if (r < 0)
- goto fail;
- }
-
- if (newly_created) {
-
- r = journal_file_setup_hash_table(f);
- if (r < 0)
- goto fail;
-
- r = journal_file_setup_bisect_table(f);
- if (r < 0)
- goto fail;
- }
-
- r = journal_file_map_hash_table(f);
- if (r < 0)
- goto fail;
-
- r = journal_file_map_bisect_table(f);
- if (r < 0)
- goto fail;
-
- if (ret)
- *ret = f;
-
- return 0;
-
-fail:
- journal_file_close(f);
-
- return r;
+int sd_journal_set_cursor(sd_journal *j, const void *cursor, size_t size) {
+ return -EINVAL;
}
int sd_journal_open(sd_journal **ret) {
diff --git a/src/journal/sd-journal.h b/src/journal/sd-journal.h
index 8170dea87c..55e58601fe 100644
--- a/src/journal/sd-journal.h
+++ b/src/journal/sd-journal.h
@@ -25,13 +25,12 @@
#include <inttypes.h>
#include <sys/types.h>
-#include "sd-id128.h"
-
/* TODO:
*
* - implement rotation
* - check LE/BE conversion for 8bit, 16bit, 32bit values
* - implement parallel traversal
+ * - implement inotify usage on client
* - implement audit gateway
* - implement native gateway
* - extend hash table/bisect table as we go
@@ -45,12 +44,13 @@ void sd_journal_close(sd_journal *j);
int sd_journal_previous(sd_journal *j);
int sd_journal_next(sd_journal *j);
-void* sd_journal_get(sd_journal *j, const char *field, size_t *size);
-uint64_t sd_journal_get_seqnum(sd_journal *j);
-uint64_t sd_journal_get_realtime_usec(sd_journal *j);
-uint64_t sd_journal_get_monotonic_usec(sd_journal *j);
+int sd_journal_get(sd_journal *j, const char *field, const void **data, size_t *size);
+int sd_journal_get_seqnum(sd_journal *j, uint64_t *ret);
+int sd_journal_get_realtime_usec(sd_journal *j, uint64_t *ret);
+int sd_journal_get_monotonic_usec(sd_journal *j, uint64_t *ret);
-int sd_journal_add_match(sd_journal *j, const char *item, size_t *size);
+int sd_journal_add_match(sd_journal *j, const char *field, const void *data, size_t size);
+void sd_journal_flush_matches(sd_journal *j);
int sd_journal_seek_head(sd_journal *j);
int sd_journal_seek_tail(sd_journal *j);
@@ -59,16 +59,9 @@ int sd_journal_seek_seqnum(sd_journal *j, uint64_t seqnum);
int sd_journal_seek_monotonic_usec(sd_journal *j, uint64_t usec);
int sd_journal_seek_realtime_usec(sd_journal *j, uint64_t usec);
-uint64_t sd_journal_get_max_size(sd_journal *j);
-uint64_t sd_journal_get_min_size(sd_journal *j);
-uint64_t sd_journal_get_keep_free(sd_journal *j);
-
-int sd_journal_set_max_size(sd_journal *j, uint64_t size);
-int sd_journal_set_min_size(sd_journal *j, uint64_t size);
-int sd_journal_set_keep_free(sd_journal *j, uint64_t size);
+int sd_journal_get_cursor(sd_journal *j, void **cursor, size_t *size);
+int sd_journal_set_cursor(sd_journal *j, const void *cursor, size_t size);
-sd_id128_t sd_journal_get_file_id(sd_journal *j);
-sd_id128_t sd_journal_get_machine_id(sd_journal *j);
-sd_id128_t sd_journal_get_boot_id(sd_journal *j);
+int sd_journal_get_fd(sd_journal *j);
#endif
diff --git a/src/journal/test-journal.c b/src/journal/test-journal.c
index e0aedc7b83..7028f11f7c 100644
--- a/src/journal/test-journal.c
+++ b/src/journal/test-journal.c
@@ -21,7 +21,7 @@
#include <fcntl.h>
-#include "journal-private.h"
+#include "journal-file.h"
#include "log.h"
int main(int argc, char *argv[]) {