X-Git-Url: https://www.chiark.greenend.org.uk/ucgi/~ianmdlvl/git?p=elogind.git;a=blobdiff_plain;f=src%2Fjournal%2Fjournald-server.c;h=019c3a649acae87f7dfd68292e479e3af701b4ab;hp=d3a1c574bd6d3ed3aa3273ed92c8e7a3a2a6f5f0;hb=ea69bd41c5923f4f278a09bb7d8cb1abcfa122e1;hpb=348ced909724a1331b85d57aede80a102a00e428 diff --git a/src/journal/journald-server.c b/src/journal/journald-server.c index d3a1c574b..019c3a649 100644 --- a/src/journal/journald-server.c +++ b/src/journal/journald-server.c @@ -50,6 +50,7 @@ #include "journald-stream.h" #include "journald-console.h" #include "journald-native.h" +#include "journald-audit.h" #include "journald-server.h" #ifdef HAVE_ACL @@ -67,10 +68,11 @@ #define DEFAULT_SYNC_INTERVAL_USEC (5*USEC_PER_MINUTE) #define DEFAULT_RATE_LIMIT_INTERVAL (30*USEC_PER_SEC) #define DEFAULT_RATE_LIMIT_BURST 1000 +#define DEFAULT_MAX_FILE_USEC USEC_PER_MONTH #define RECHECK_AVAILABLE_SPACE_USEC (30*USEC_PER_SEC) -static const char* const storage_table[] = { +static const char* const storage_table[_STORAGE_MAX] = { [STORAGE_AUTO] = "auto", [STORAGE_VOLATILE] = "volatile", [STORAGE_PERSISTENT] = "persistent", @@ -80,10 +82,10 @@ static const char* const storage_table[] = { DEFINE_STRING_TABLE_LOOKUP(storage, Storage); DEFINE_CONFIG_PARSE_ENUM(config_parse_storage, storage, Storage, "Failed to parse storage setting"); -static const char* const split_mode_table[] = { - [SPLIT_NONE] = "none", +static const char* const split_mode_table[_SPLIT_MAX] = { + [SPLIT_LOGIN] = "login", [SPLIT_UID] = "uid", - [SPLIT_LOGIN] = "login" + [SPLIT_NONE] = "none", }; DEFINE_STRING_TABLE_LOOKUP(split_mode, SplitMode); @@ -202,15 +204,15 @@ void server_fix_perms(Server *s, JournalFile *f, uid_t uid) { r = fchmod(f->fd, 0640); if (r < 0) - log_warning("Failed to fix access mode on %s, ignoring: %s", f->path, strerror(-r)); + log_warning_errno(r, "Failed to fix access mode on %s, ignoring: %m", f->path); #ifdef HAVE_ACL - if (uid <= 0) + if (uid <= SYSTEM_UID_MAX) return; acl = acl_get_fd(f->fd); if (!acl) { - log_warning("Failed to read ACL on %s, ignoring: %m", f->path); + log_warning_errno(errno, "Failed to read ACL on %s, ignoring: %m", f->path); return; } @@ -220,7 +222,7 @@ void server_fix_perms(Server *s, JournalFile *f, uid_t uid) { if (acl_create_entry(&acl, &entry) < 0 || acl_set_tag_type(entry, ACL_USER) < 0 || acl_set_qualifier(entry, &uid) < 0) { - log_warning("Failed to patch ACL on %s, ignoring: %m", f->path); + log_warning_errno(errno, "Failed to patch ACL on %s, ignoring: %m", f->path); goto finish; } } @@ -230,12 +232,12 @@ void server_fix_perms(Server *s, JournalFile *f, uid_t uid) { if (acl_get_permset(entry, &permset) < 0 || acl_add_perm(permset, ACL_READ) < 0 || calc_acl_mask_if_needed(&acl) < 0) { - log_warning("Failed to patch ACL on %s, ignoring: %m", f->path); + log_warning_errno(errno, "Failed to patch ACL on %s, ignoring: %m", f->path); goto finish; } if (acl_set_fd(f->fd, acl) < 0) - log_warning("Failed to set ACL on %s, ignoring: %m", f->path); + log_warning_errno(errno, "Failed to set ACL on %s, ignoring: %m", f->path); finish: acl_free(acl); @@ -258,24 +260,24 @@ static JournalFile* find_journal(Server *s, uid_t uid) { if (s->runtime_journal) return s->runtime_journal; - if (uid <= 0) + if (uid <= SYSTEM_UID_MAX) return s->system_journal; r = sd_id128_get_machine(&machine); if (r < 0) return s->system_journal; - f = hashmap_get(s->user_journals, UINT32_TO_PTR(uid)); + f = ordered_hashmap_get(s->user_journals, UINT32_TO_PTR(uid)); if (f) return f; - if (asprintf(&p, "/var/log/journal/" SD_ID128_FORMAT_STR "/user-%lu.journal", - SD_ID128_FORMAT_VAL(machine), (unsigned long) uid) < 0) + if (asprintf(&p, "/var/log/journal/" SD_ID128_FORMAT_STR "/user-"UID_FMT".journal", + SD_ID128_FORMAT_VAL(machine), uid) < 0) return s->system_journal; - while (hashmap_size(s->user_journals) >= USER_JOURNALS_MAX) { + while (ordered_hashmap_size(s->user_journals) >= USER_JOURNALS_MAX) { /* Too many open? Then let's close one */ - f = hashmap_steal_first(s->user_journals); + f = ordered_hashmap_steal_first(s->user_journals); assert(f); journal_file_close(f); } @@ -286,7 +288,7 @@ static JournalFile* find_journal(Server *s, uid_t uid) { server_fix_perms(s, f, uid); - r = hashmap_put(s->user_journals, UINT32_TO_PTR(uid), f); + r = ordered_hashmap_put(s->user_journals, UINT32_TO_PTR(uid), f); if (r < 0) { journal_file_close(f); return s->system_journal; @@ -295,6 +297,30 @@ static JournalFile* find_journal(Server *s, uid_t uid) { return f; } +static int do_rotate( + Server *s, + JournalFile **f, + const char* name, + bool seal, + uint32_t uid) { + + int r; + assert(s); + + if (!*f) + return -EINVAL; + + r = journal_file_rotate(f, s->compress, seal); + if (r < 0) + if (*f) + log_error_errno(r, "Failed to rotate %s: %m", (*f)->path); + else + log_error_errno(r, "Failed to create new %s journal: %m", name); + else + server_fix_perms(s, *f, uid); + return r; +} + void server_rotate(Server *s) { JournalFile *f; void *k; @@ -303,42 +329,16 @@ void server_rotate(Server *s) { log_debug("Rotating..."); - if (s->runtime_journal) { - r = journal_file_rotate(&s->runtime_journal, s->compress, false); - if (r < 0) - if (s->runtime_journal) - log_error("Failed to rotate %s: %s", s->runtime_journal->path, strerror(-r)); - else - log_error("Failed to create new runtime journal: %s", strerror(-r)); - else - server_fix_perms(s, s->runtime_journal, 0); - } - - if (s->system_journal) { - r = journal_file_rotate(&s->system_journal, s->compress, s->seal); - if (r < 0) - if (s->system_journal) - log_error("Failed to rotate %s: %s", s->system_journal->path, strerror(-r)); - else - log_error("Failed to create new system journal: %s", strerror(-r)); + do_rotate(s, &s->runtime_journal, "runtime", false, 0); + do_rotate(s, &s->system_journal, "system", s->seal, 0); - else - server_fix_perms(s, s->system_journal, 0); - } - - HASHMAP_FOREACH_KEY(f, k, s->user_journals, i) { - r = journal_file_rotate(&f, s->compress, s->seal); - if (r < 0) - if (f) - log_error("Failed to rotate %s: %s", f->path, strerror(-r)); - else { - log_error("Failed to create user journal: %s", strerror(-r)); - hashmap_remove(s->user_journals, k); - } - else { - hashmap_replace(s->user_journals, k, f); - server_fix_perms(s, f, PTR_TO_UINT32(k)); - } + ORDERED_HASHMAP_FOREACH_KEY(f, k, s->user_journals, i) { + r = do_rotate(s, &f, "user", s->seal, PTR_TO_UINT32(k)); + if (r >= 0) + ordered_hashmap_replace(s->user_journals, k, f); + else if (!f) + /* Old file has been closed and deallocated */ + ordered_hashmap_remove(s->user_journals, k); } } @@ -351,24 +351,43 @@ void server_sync(Server *s) { if (s->system_journal) { r = journal_file_set_offline(s->system_journal); if (r < 0) - log_error("Failed to sync system journal: %s", strerror(-r)); + log_error_errno(r, "Failed to sync system journal: %m"); } - HASHMAP_FOREACH_KEY(f, k, s->user_journals, i) { + ORDERED_HASHMAP_FOREACH_KEY(f, k, s->user_journals, i) { r = journal_file_set_offline(f); if (r < 0) - log_error("Failed to sync user journal: %s", strerror(-r)); + log_error_errno(r, "Failed to sync user journal: %m"); } if (s->sync_event_source) { r = sd_event_source_set_enabled(s->sync_event_source, SD_EVENT_OFF); if (r < 0) - log_error("Failed to disable sync timer source: %s", strerror(-r)); + log_error_errno(r, "Failed to disable sync timer source: %m"); } s->sync_scheduled = false; } +static void do_vacuum( + Server *s, + const char *id, + JournalFile *f, + const char* path, + JournalMetrics *metrics) { + + const char *p; + int r; + + if (!f) + return; + + p = strappenda(path, id); + r = journal_directory_vacuum(p, metrics->max_use, s->max_retention_usec, &s->oldest_file_usec, false); + if (r < 0 && r != -ENOENT) + log_error_errno(r, "Failed to vacuum %s: %m", p); +} + void server_vacuum(Server *s) { char ids[33]; sd_id128_t machine; @@ -380,27 +399,13 @@ void server_vacuum(Server *s) { r = sd_id128_get_machine(&machine); if (r < 0) { - log_error("Failed to get machine ID: %s", strerror(-r)); + log_error_errno(r, "Failed to get machine ID: %m"); return; } - sd_id128_to_string(machine, ids); - if (s->system_journal) { - char *p = strappenda("/var/log/journal/", ids); - - r = journal_directory_vacuum(p, s->system_metrics.max_use, s->max_retention_usec, &s->oldest_file_usec); - if (r < 0 && r != -ENOENT) - log_error("Failed to vacuum %s: %s", p, strerror(-r)); - } - - if (s->runtime_journal) { - char *p = strappenda("/run/log/journal/", ids); - - r = journal_directory_vacuum(p, s->runtime_metrics.max_use, s->max_retention_usec, &s->oldest_file_usec); - if (r < 0 && r != -ENOENT) - log_error("Failed to vacuum %s: %s", p, strerror(-r)); - } + do_vacuum(s, ids, s->system_journal, "/var/log/journal/", &s->system_metrics); + do_vacuum(s, ids, s->runtime_journal, "/run/log/journal/", &s->runtime_metrics); s->cached_available_space_timestamp = 0; } @@ -449,12 +454,13 @@ static void server_cache_hostname(Server *s) { s->hostname_field = x; } -bool shall_try_append_again(JournalFile *f, int r) { +static bool shall_try_append_again(JournalFile *f, int r) { /* -E2BIG Hit configured limit -EFBIG Hit fs limit -EDQUOT Quota limit hit -ENOSPC Disk full + -EIO I/O error of some kind (mmap) -EHOSTDOWN Other machine -EBUSY Unclean shutdown -EPROTONOSUPPORT Unsupported feature @@ -472,6 +478,8 @@ bool shall_try_append_again(JournalFile *f, int r) { log_info("%s: Unsupported feature, rotating.", f->path); else if (r == -EBADMSG || r == -ENODATA || r == ESHUTDOWN) log_warning("%s: Journal file corrupted, rotating.", f->path); + else if (r == -EIO) + log_warning("%s: IO error, rotating.", f->path); else return false; @@ -514,7 +522,7 @@ static void write_to_journal(Server *s, uid_t uid, struct iovec *iovec, unsigned for (i = 0; i < n; i++) size += iovec[i].iov_len; - log_error("Failed to write entry (%d items, %zu bytes), ignoring: %s", n, size, strerror(-r)); + log_error_errno(r, "Failed to write entry (%d items, %zu bytes), ignoring: %m", n, size); return; } @@ -533,7 +541,7 @@ static void write_to_journal(Server *s, uid_t uid, struct iovec *iovec, unsigned for (i = 0; i < n; i++) size += iovec[i].iov_len; - log_error("Failed to write entry (%d items, %zu bytes) despite vacuuming, ignoring: %s", n, size, strerror(-r)); + log_error_errno(r, "Failed to write entry (%d items, %zu bytes) despite vacuuming, ignoring: %m", n, size); } else server_schedule_sync(s, priority); } @@ -541,8 +549,8 @@ static void write_to_journal(Server *s, uid_t uid, struct iovec *iovec, unsigned static void dispatch_message_real( Server *s, struct iovec *iovec, unsigned n, unsigned m, - struct ucred *ucred, - struct timeval *tv, + const struct ucred *ucred, + const struct timeval *tv, const char *label, size_t label_len, const char *unit_id, int priority, @@ -581,13 +589,13 @@ static void dispatch_message_real( if (ucred) { realuid = ucred->uid; - sprintf(pid, "_PID=%lu", (unsigned long) ucred->pid); + sprintf(pid, "_PID="PID_FMT, ucred->pid); IOVEC_SET_STRING(iovec[n++], pid); - sprintf(uid, "_UID=%lu", (unsigned long) ucred->uid); + sprintf(uid, "_UID="UID_FMT, ucred->uid); IOVEC_SET_STRING(iovec[n++], uid); - sprintf(gid, "_GID=%lu", (unsigned long) ucred->gid); + sprintf(gid, "_GID="GID_FMT, ucred->gid); IOVEC_SET_STRING(iovec[n++], gid); r = get_process_comm(ucred->pid, &t); @@ -621,13 +629,13 @@ static void dispatch_message_real( #ifdef HAVE_AUDIT r = audit_session_from_pid(ucred->pid, &audit); if (r >= 0) { - sprintf(audit_session, "_AUDIT_SESSION=%lu", (unsigned long) audit); + sprintf(audit_session, "_AUDIT_SESSION=%"PRIu32, audit); IOVEC_SET_STRING(iovec[n++], audit_session); } r = audit_loginuid_from_pid(ucred->pid, &loginuid); if (r >= 0) { - sprintf(audit_loginuid, "_AUDIT_LOGINUID=%lu", (unsigned long) loginuid); + sprintf(audit_loginuid, "_AUDIT_LOGINUID="UID_FMT, loginuid); IOVEC_SET_STRING(iovec[n++], audit_loginuid); } #endif @@ -649,7 +657,7 @@ static void dispatch_message_real( if (cg_path_get_owner_uid(c, &owner) >= 0) { owner_valid = true; - sprintf(owner_uid, "_SYSTEMD_OWNER_UID=%lu", (unsigned long) owner); + sprintf(owner_uid, "_SYSTEMD_OWNER_UID="UID_FMT, owner); IOVEC_SET_STRING(iovec[n++], owner_uid); } @@ -684,9 +692,9 @@ static void dispatch_message_real( } #ifdef HAVE_SELINUX - if (use_selinux()) { + if (mac_selinux_use()) { if (label) { - x = alloca(sizeof("_SELINUX_CONTEXT=") + label_len); + x = alloca(strlen("_SELINUX_CONTEXT=") + label_len + 1); *((char*) mempcpy(stpcpy(x, "_SELINUX_CONTEXT="), label, label_len)) = 0; IOVEC_SET_STRING(iovec[n++], x); @@ -708,13 +716,13 @@ static void dispatch_message_real( if (object_pid) { r = get_process_uid(object_pid, &object_uid); if (r >= 0) { - sprintf(o_uid, "OBJECT_UID=%lu", (unsigned long) object_uid); + sprintf(o_uid, "OBJECT_UID="UID_FMT, object_uid); IOVEC_SET_STRING(iovec[n++], o_uid); } r = get_process_gid(object_pid, &object_gid); if (r >= 0) { - sprintf(o_gid, "OBJECT_GID=%lu", (unsigned long) object_gid); + sprintf(o_gid, "OBJECT_GID="GID_FMT, object_gid); IOVEC_SET_STRING(iovec[n++], o_gid); } @@ -742,13 +750,13 @@ static void dispatch_message_real( #ifdef HAVE_AUDIT r = audit_session_from_pid(object_pid, &audit); if (r >= 0) { - sprintf(o_audit_session, "OBJECT_AUDIT_SESSION=%lu", (unsigned long) audit); + sprintf(o_audit_session, "OBJECT_AUDIT_SESSION=%"PRIu32, audit); IOVEC_SET_STRING(iovec[n++], o_audit_session); } r = audit_loginuid_from_pid(object_pid, &loginuid); if (r >= 0) { - sprintf(o_audit_loginuid, "OBJECT_AUDIT_LOGINUID=%lu", (unsigned long) loginuid); + sprintf(o_audit_loginuid, "OBJECT_AUDIT_LOGINUID="UID_FMT, loginuid); IOVEC_SET_STRING(iovec[n++], o_audit_loginuid); } #endif @@ -766,7 +774,7 @@ static void dispatch_message_real( } if (cg_path_get_owner_uid(c, &owner) >= 0) { - sprintf(o_owner_uid, "OBJECT_SYSTEMD_OWNER_UID=%lu", (unsigned long) owner); + sprintf(o_owner_uid, "OBJECT_SYSTEMD_OWNER_UID="UID_FMT, owner); IOVEC_SET_STRING(iovec[n++], o_owner_uid); } @@ -810,12 +818,11 @@ static void dispatch_message_real( /* Split up strictly by any UID */ journal_uid = realuid; else if (s->split_mode == SPLIT_LOGIN && realuid > 0 && owner_valid && owner > 0) - /* Split up by login UIDs, this avoids creation of - * individual journals for system UIDs. We do this - * only if the realuid is not root, in order not to - * accidentally leak privileged information to the - * user that is logged by a privileged process that is - * part of an unprivileged session.*/ + /* Split up by login UIDs. We do this only if the + * realuid is not root, in order not to accidentally + * leak privileged information to the user that is + * logged by a privileged process that is part of an + * unprivileged session. */ journal_uid = owner; else journal_uid = 0; @@ -845,7 +852,7 @@ void server_driver_message(Server *s, sd_id128_t message_id, const char *format, IOVEC_SET_STRING(iovec[n++], buffer); if (!sd_id128_equal(message_id, SD_ID128_NULL)) { - snprintf(mid, sizeof(mid), MESSAGE_ID(message_id)); + snprintf(mid, sizeof(mid), LOG_MESSAGE_ID(message_id)); char_array_0(mid); IOVEC_SET_STRING(iovec[n++], mid); } @@ -860,8 +867,8 @@ void server_driver_message(Server *s, sd_id128_t message_id, const char *format, void server_dispatch_message( Server *s, struct iovec *iovec, unsigned n, unsigned m, - struct ucred *ucred, - struct timeval *tv, + const struct ucred *ucred, + const struct timeval *tv, const char *label, size_t label_len, const char *unit_id, int priority, @@ -924,23 +931,22 @@ finish: } -static int system_journal_open(Server *s) { +static int system_journal_open(Server *s, bool flush_requested) { int r; char *fn; sd_id128_t machine; char ids[33]; r = sd_id128_get_machine(&machine); - if (r < 0) { - log_error("Failed to get machine id: %s", strerror(-r)); - return r; - } + if (r < 0) + return log_error_errno(r, "Failed to get machine id: %m"); sd_id128_to_string(machine, ids); if (!s->system_journal && (s->storage == STORAGE_PERSISTENT || s->storage == STORAGE_AUTO) && - access("/run/systemd/journal/flushed", F_OK) >= 0) { + (flush_requested + || access("/run/systemd/journal/flushed", F_OK) >= 0)) { /* If in auto mode: first try to create the machine * path, but not the prefix. @@ -961,7 +967,7 @@ static int system_journal_open(Server *s) { server_fix_perms(s, s->system_journal, 0); else if (r < 0) { if (r != -ENOENT && r != -EROFS) - log_warning("Failed to open system journal: %s", strerror(-r)); + log_warning_errno(r, "Failed to open system journal: %m"); r = 0; } @@ -985,7 +991,7 @@ static int system_journal_open(Server *s) { if (r < 0) { if (r != -ENOENT) - log_warning("Failed to open runtime journal: %s", strerror(-r)); + log_warning_errno(r, "Failed to open runtime journal: %m"); r = 0; } @@ -995,14 +1001,15 @@ static int system_journal_open(Server *s) { /* OK, we really need the runtime journal, so create * it if necessary. */ - (void) mkdir_parents(fn, 0755); + (void) mkdir("/run/log", 0755); + (void) mkdir("/run/log/journal", 0755); + (void) mkdir_parents(fn, 0750); + r = journal_file_open_reliably(fn, O_RDWR|O_CREAT, 0640, s->compress, false, &s->runtime_metrics, s->mmap, NULL, &s->runtime_journal); free(fn); - if (r < 0) { - log_error("Failed to open runtime journal: %s", strerror(-r)); - return r; - } + if (r < 0) + return log_error_errno(r, "Failed to open runtime journal: %m"); } if (s->runtime_journal) @@ -1031,7 +1038,7 @@ int server_flush_to_var(Server *s) { if (!s->runtime_journal) return 0; - system_journal_open(s); + system_journal_open(s, true); if (!s->system_journal) return 0; @@ -1045,10 +1052,8 @@ int server_flush_to_var(Server *s) { return r; r = sd_journal_open(&j, SD_JOURNAL_RUNTIME_ONLY); - if (r < 0) { - log_error("Failed to read runtime journal: %s", strerror(-r)); - return r; - } + if (r < 0) + return log_error_errno(r, "Failed to read runtime journal: %m"); sd_journal_set_data_threshold(j, 0); @@ -1063,7 +1068,7 @@ int server_flush_to_var(Server *s) { r = journal_file_move_to_object(f, OBJECT_ENTRY, f->current_offset, &o); if (r < 0) { - log_error("Can't read entry: %s", strerror(-r)); + log_error_errno(r, "Can't read entry: %m"); goto finish; } @@ -1072,7 +1077,7 @@ int server_flush_to_var(Server *s) { continue; if (!shall_try_append_again(s->system_journal, r)) { - log_error("Can't write entry: %s", strerror(-r)); + log_error_errno(r, "Can't write entry: %m"); goto finish; } @@ -1088,7 +1093,7 @@ int server_flush_to_var(Server *s) { log_debug("Retrying write."); r = journal_file_copy_entry(f, s->system_journal, o, f->current_offset, NULL, NULL, NULL); if (r < 0) { - log_error("Can't write entry: %s", strerror(-r)); + log_error_errno(r, "Can't write entry: %m"); goto finish; } } @@ -1109,11 +1114,11 @@ finish: return r; } -int process_datagram(sd_event_source *es, int fd, uint32_t revents, void *userdata) { +int server_process_datagram(sd_event_source *es, int fd, uint32_t revents, void *userdata) { Server *s = userdata; assert(s); - assert(fd == s->native_fd || fd == s->syslog_fd); + assert(fd == s->native_fd || fd == s->syslog_fd || fd == s->audit_fd); if (revents != EPOLLIN) { log_error("Got invalid event from epoll for datagram fd: %"PRIx32, revents); @@ -1131,49 +1136,56 @@ int process_datagram(sd_event_source *es, int fd, uint32_t revents, void *userda union { struct cmsghdr cmsghdr; - /* We use NAME_MAX space for the - * SELinux label here. The kernel - * currently enforces no limit, but - * according to suggestions from the - * SELinux people this will change and - * it will probably be identical to - * NAME_MAX. For now we use that, but - * this should be updated one day when - * the final limit is known.*/ + /* We use NAME_MAX space for the SELinux label + * here. The kernel currently enforces no + * limit, but according to suggestions from + * the SELinux people this will change and it + * will probably be identical to NAME_MAX. For + * now we use that, but this should be updated + * one day when the final limit is known. */ uint8_t buf[CMSG_SPACE(sizeof(struct ucred)) + CMSG_SPACE(sizeof(struct timeval)) + CMSG_SPACE(sizeof(int)) + /* fd */ CMSG_SPACE(NAME_MAX)]; /* selinux label */ } control = {}; + union sockaddr_union sa = {}; struct msghdr msghdr = { .msg_iov = &iovec, .msg_iovlen = 1, .msg_control = &control, .msg_controllen = sizeof(control), + .msg_name = &sa, + .msg_namelen = sizeof(sa), }; ssize_t n; - int v; int *fds = NULL; unsigned n_fds = 0; + int v = 0; + size_t m; - if (ioctl(fd, SIOCINQ, &v) < 0) { - log_error("SIOCINQ failed: %m"); - return -errno; - } + /* Try to get the right size, if we can. (Not all + * sockets support SIOCINQ, hence we just try, but + * don't rely on it. */ + (void) ioctl(fd, SIOCINQ, &v); - if (!GREEDY_REALLOC(s->buffer, s->buffer_size, LINE_MAX + (size_t) v)) + /* Fix it up, if it is too small. We use the same fixed value as auditd here. Awful! */ + m = PAGE_ALIGN(MAX3((size_t) v + 1, + (size_t) LINE_MAX, + ALIGN(sizeof(struct nlmsghdr)) + ALIGN((size_t) MAX_AUDIT_MESSAGE_LENGTH)) + 1); + + if (!GREEDY_REALLOC(s->buffer, s->buffer_size, m)) return log_oom(); iovec.iov_base = s->buffer; - iovec.iov_len = s->buffer_size; + iovec.iov_len = s->buffer_size - 1; /* Leave room for trailing NUL we add later */ n = recvmsg(fd, &msghdr, MSG_DONTWAIT|MSG_CMSG_CLOEXEC); if (n < 0) { if (errno == EINTR || errno == EAGAIN) return 0; - log_error("recvmsg() failed: %m"); + log_error_errno(errno, "recvmsg() failed: %m"); return -errno; } @@ -1198,26 +1210,34 @@ int process_datagram(sd_event_source *es, int fd, uint32_t revents, void *userda } } + /* And a trailing NUL, just in case */ + s->buffer[n] = 0; + if (fd == s->syslog_fd) { - if (n > 0 && n_fds == 0) { - s->buffer[n] = 0; + if (n > 0 && n_fds == 0) server_process_syslog_message(s, strstrip(s->buffer), ucred, tv, label, label_len); - } else if (n_fds > 0) + else if (n_fds > 0) log_warning("Got file descriptors via syslog socket. Ignoring."); - } else { + } else if (fd == s->native_fd) { if (n > 0 && n_fds == 0) server_process_native_message(s, s->buffer, n, ucred, tv, label, label_len); else if (n == 0 && n_fds == 1) server_process_native_file(s, fds[0], ucred, tv, label, label_len); else if (n_fds > 0) log_warning("Got too many file descriptors via native socket. Ignoring."); + + } else { + assert(fd == s->audit_fd); + + if (n > 0 && n_fds == 0) + server_process_audit_message(s, s->buffer, n, ucred, &sa, msghdr.msg_namelen); + else if (n_fds > 0) + log_warning("Got file descriptors via audit socket. Ignoring."); } close_many(fds, n_fds); } - - return 0; } static int dispatch_sigusr1(sd_event_source *es, const struct signalfd_siginfo *si, void *userdata) { @@ -1227,9 +1247,11 @@ static int dispatch_sigusr1(sd_event_source *es, const struct signalfd_siginfo * log_info("Received request to flush runtime journal from PID %"PRIu32, si->ssi_pid); - touch("/run/systemd/journal/flushed"); server_flush_to_var(s); server_sync(s); + server_vacuum(s); + + touch("/run/systemd/journal/flushed"); return 0; } @@ -1251,7 +1273,7 @@ static int dispatch_sigterm(sd_event_source *es, const struct signalfd_siginfo * assert(s); - log_info("Received SIG%s", signal_to_string(si->ssi_signo)); + log_received_signal(LOG_INFO, si); sd_event_exit(s->event, 0); return 0; @@ -1267,19 +1289,19 @@ static int setup_signals(Server *s) { sigset_add_many(&mask, SIGINT, SIGTERM, SIGUSR1, SIGUSR2, -1); assert_se(sigprocmask(SIG_SETMASK, &mask, NULL) == 0); - r = sd_event_add_signal(s->event, SIGUSR1, dispatch_sigusr1, s, &s->sigusr1_event_source); + r = sd_event_add_signal(s->event, &s->sigusr1_event_source, SIGUSR1, dispatch_sigusr1, s); if (r < 0) return r; - r = sd_event_add_signal(s->event, SIGUSR2, dispatch_sigusr2, s, &s->sigusr2_event_source); + r = sd_event_add_signal(s->event, &s->sigusr2_event_source, SIGUSR2, dispatch_sigusr2, s); if (r < 0) return r; - r = sd_event_add_signal(s->event, SIGTERM, dispatch_sigterm, s, &s->sigterm_event_source); + r = sd_event_add_signal(s->event, &s->sigterm_event_source, SIGTERM, dispatch_sigterm, s); if (r < 0) return r; - r = sd_event_add_signal(s->event, SIGINT, dispatch_sigterm, s, &s->sigint_event_source); + r = sd_event_add_signal(s->event, &s->sigint_event_source, SIGINT, dispatch_sigterm, s); if (r < 0) return r; @@ -1288,15 +1310,15 @@ static int setup_signals(Server *s) { static int server_parse_proc_cmdline(Server *s) { _cleanup_free_ char *line = NULL; - char *w, *state; + const char *w, *state; size_t l; int r; r = proc_cmdline(&line); - if (r < 0) - log_warning("Failed to read /proc/cmdline, ignoring: %s", strerror(-r)); - if (r <= 0) + if (r < 0) { + log_warning_errno(r, "Failed to read /proc/cmdline, ignoring: %m"); return 0; + } FOREACH_WORD_QUOTED(w, l, line, state) { _cleanup_free_ char *word; @@ -1323,35 +1345,28 @@ static int server_parse_proc_cmdline(Server *s) { log_warning("Failed to parse forward to console switch %s. Ignoring.", word + 36); else s->forward_to_console = r; + } else if (startswith(word, "systemd.journald.forward_to_wall=")) { + r = parse_boolean(word + 33); + if (r < 0) + log_warning("Failed to parse forward to wall switch %s. Ignoring.", word + 33); + else + s->forward_to_wall = r; } else if (startswith(word, "systemd.journald")) log_warning("Invalid systemd.journald parameter. Ignoring."); } + /* do not warn about state here, since probably systemd already did */ return 0; } static int server_parse_config_file(Server *s) { - static const char fn[] = "/etc/systemd/journald.conf"; - _cleanup_fclose_ FILE *f = NULL; - int r; - assert(s); - f = fopen(fn, "re"); - if (!f) { - if (errno == ENOENT) - return 0; - - log_warning("Failed to open configuration file %s: %m", fn); - return -errno; - } - - r = config_parse(NULL, fn, f, "Journal\0", config_item_perf_lookup, - (void*) journald_gperf_lookup, false, false, s); - if (r < 0) - log_warning("Failed to parse configuration file: %s", strerror(-r)); - - return r; + return config_parse_many("/etc/systemd/journald.conf", + CONF_DIRS_NULSTR("systemd/journald.conf"), + "Journal\0", + config_item_perf_lookup, journald_gperf_lookup, + false, s); } static int server_dispatch_sync(sd_event_source *es, usec_t t, void *userdata) { @@ -1380,14 +1395,19 @@ int server_schedule_sync(Server *s, int priority) { if (s->sync_interval_usec > 0) { usec_t when; - r = sd_event_get_now_monotonic(s->event, &when); + r = sd_event_now(s->event, CLOCK_MONOTONIC, &when); if (r < 0) return r; when += s->sync_interval_usec; if (!s->sync_event_source) { - r = sd_event_add_monotonic(s->event, when, 0, server_dispatch_sync, s, &s->sync_event_source); + r = sd_event_add_time( + s->event, + &s->sync_event_source, + CLOCK_MONOTONIC, + when, 0, + server_dispatch_sync, s); if (r < 0) return r; @@ -1423,22 +1443,26 @@ static int server_open_hostname(Server *s) { assert(s); s->hostname_fd = open("/proc/sys/kernel/hostname", O_RDONLY|O_CLOEXEC|O_NDELAY|O_NOCTTY); - if (s->hostname_fd < 0) { - log_error("Failed to open /proc/sys/kernel/hostname: %m"); - return -errno; - } + if (s->hostname_fd < 0) + return log_error_errno(errno, "Failed to open /proc/sys/kernel/hostname: %m"); - r = sd_event_add_io(s->event, s->hostname_fd, 0, dispatch_hostname_change, s, &s->hostname_event_source); + r = sd_event_add_io(s->event, &s->hostname_event_source, s->hostname_fd, 0, dispatch_hostname_change, s); if (r < 0) { - log_error("Failed to register hostname fd in event loop: %s", strerror(-r)); - return r; + /* kernels prior to 3.2 don't support polling this file. Ignore + * the failure. */ + if (r == -EPERM) { + log_warning("Failed to register hostname fd in event loop: %s. Ignoring.", + strerror(-r)); + s->hostname_fd = safe_close(s->hostname_fd); + return 0; + } + + return log_error_errno(r, "Failed to register hostname fd in event loop: %m"); } r = sd_event_source_set_priority(s->hostname_event_source, SD_EVENT_PRIORITY_IMPORTANT-10); - if (r < 0) { - log_error("Failed to adjust priority of host name event source: %s", strerror(-r)); - return r; - } + if (r < 0) + return log_error_errno(r, "Failed to adjust priority of host name event source: %m"); return 0; } @@ -1449,7 +1473,7 @@ int server_init(Server *s) { assert(s); zero(*s); - s->syslog_fd = s->native_fd = s->stdout_fd = s->dev_kmsg_fd = s->hostname_fd = -1; + s->syslog_fd = s->native_fd = s->stdout_fd = s->dev_kmsg_fd = s->audit_fd = s->hostname_fd = -1; s->compress = true; s->seal = true; @@ -1459,12 +1483,15 @@ int server_init(Server *s) { s->rate_limit_interval = DEFAULT_RATE_LIMIT_INTERVAL; s->rate_limit_burst = DEFAULT_RATE_LIMIT_BURST; - s->forward_to_syslog = true; + s->forward_to_wall = true; + + s->max_file_usec = DEFAULT_MAX_FILE_USEC; s->max_level_store = LOG_DEBUG; s->max_level_syslog = LOG_DEBUG; s->max_level_kmsg = LOG_NOTICE; s->max_level_console = LOG_INFO; + s->max_level_wall = LOG_EMERG; memset(&s->system_metrics, 0xFF, sizeof(s->system_metrics)); memset(&s->runtime_metrics, 0xFF, sizeof(s->runtime_metrics)); @@ -1472,15 +1499,14 @@ int server_init(Server *s) { server_parse_config_file(s); server_parse_proc_cmdline(s); if (!!s->rate_limit_interval ^ !!s->rate_limit_burst) { - log_debug("Setting both rate limit interval and burst from %llu,%u to 0,0", - (long long unsigned) s->rate_limit_interval, - s->rate_limit_burst); + log_debug("Setting both rate limit interval and burst from "USEC_FMT",%u to 0,0", + s->rate_limit_interval, s->rate_limit_burst); s->rate_limit_interval = s->rate_limit_burst = 0; } mkdir_p("/run/systemd/journal", 0755); - s->user_journals = hashmap_new(trivial_hash_func, trivial_compare_func); + s->user_journals = ordered_hashmap_new(NULL); if (!s->user_journals) return log_oom(); @@ -1489,18 +1515,14 @@ int server_init(Server *s) { return log_oom(); r = sd_event_default(&s->event); - if (r < 0) { - log_error("Failed to create event loop: %s", strerror(-r)); - return r; - } + if (r < 0) + return log_error_errno(r, "Failed to create event loop: %m"); sd_event_set_watchdog(s->event, true); n = sd_listen_fds(true); - if (n < 0) { - log_error("Failed to read listening file descriptors from environment: %s", strerror(-n)); - return n; - } + if (n < 0) + return log_error_errno(n, "Failed to read listening file descriptors from environment: %m"); for (fd = SD_LISTEN_FDS_START; fd < SD_LISTEN_FDS_START + n; fd++) { @@ -1522,7 +1544,8 @@ int server_init(Server *s) { s->stdout_fd = fd; - } else if (sd_is_socket_unix(fd, SOCK_DGRAM, -1, "/dev/log", 0) > 0) { + } else if (sd_is_socket_unix(fd, SOCK_DGRAM, -1, "/dev/log", 0) > 0 || + sd_is_socket_unix(fd, SOCK_DGRAM, -1, "/run/systemd/journal/dev-log", 0) > 0) { if (s->syslog_fd >= 0) { log_error("Too many /dev/log sockets passed."); @@ -1531,9 +1554,24 @@ int server_init(Server *s) { s->syslog_fd = fd; + } else if (sd_is_socket(fd, AF_NETLINK, SOCK_RAW, -1) > 0) { + + if (s->audit_fd >= 0) { + log_error("Too many audit sockets passed."); + return -EINVAL; + } + + s->audit_fd = fd; + } else { - log_error("Unknown socket passed."); - return -EINVAL; + log_warning("Unknown socket passed as file descriptor %d, ignoring.", fd); + + /* Let's close the fd, better be safe than + sorry. The fd might reference some resource + that we really want to release if we don't + make use of it. */ + + safe_close(fd); } } @@ -1553,6 +1591,10 @@ int server_init(Server *s) { if (r < 0) return r; + r = server_open_audit(s); + if (r < 0) + return r; + r = server_open_kernel_seqnum(s); if (r < 0) return r; @@ -1581,7 +1623,7 @@ int server_init(Server *s) { server_cache_boot_id(s); server_cache_machine_id(s); - r = system_journal_open(s); + r = system_journal_open(s, false); if (r < 0) return r; @@ -1599,7 +1641,7 @@ void server_maybe_append_tags(Server *s) { if (s->system_journal) journal_file_maybe_append_tag(s->system_journal, n); - HASHMAP_FOREACH(f, s->user_journals, i) + ORDERED_HASHMAP_FOREACH(f, s->user_journals, i) journal_file_maybe_append_tag(f, n); #endif } @@ -1617,15 +1659,16 @@ void server_done(Server *s) { if (s->runtime_journal) journal_file_close(s->runtime_journal); - while ((f = hashmap_steal_first(s->user_journals))) + while ((f = ordered_hashmap_steal_first(s->user_journals))) journal_file_close(f); - hashmap_free(s->user_journals); + ordered_hashmap_free(s->user_journals); sd_event_source_unref(s->syslog_event_source); sd_event_source_unref(s->native_event_source); sd_event_source_unref(s->stdout_event_source); sd_event_source_unref(s->dev_kmsg_event_source); + sd_event_source_unref(s->audit_event_source); sd_event_source_unref(s->sync_event_source); sd_event_source_unref(s->sigusr1_event_source); sd_event_source_unref(s->sigusr2_event_source); @@ -1634,20 +1677,12 @@ void server_done(Server *s) { sd_event_source_unref(s->hostname_event_source); sd_event_unref(s->event); - if (s->syslog_fd >= 0) - close_nointr_nofail(s->syslog_fd); - - if (s->native_fd >= 0) - close_nointr_nofail(s->native_fd); - - if (s->stdout_fd >= 0) - close_nointr_nofail(s->stdout_fd); - - if (s->dev_kmsg_fd >= 0) - close_nointr_nofail(s->dev_kmsg_fd); - - if (s->hostname_fd >= 0) - close_nointr_nofail(s->hostname_fd); + safe_close(s->syslog_fd); + safe_close(s->native_fd); + safe_close(s->stdout_fd); + safe_close(s->dev_kmsg_fd); + safe_close(s->audit_fd); + safe_close(s->hostname_fd); if (s->rate_limit) journal_rate_limit_free(s->rate_limit); @@ -1658,6 +1693,7 @@ void server_done(Server *s) { free(s->buffer); free(s->tty_path); free(s->cgroup_root); + free(s->hostname_field); if (s->mmap) mmap_cache_unref(s->mmap);