X-Git-Url: http://www.chiark.greenend.org.uk/ucgi/~ianmdlvl/git?p=elogind.git;a=blobdiff_plain;f=src%2Fjournal%2Fjournald-server.c;h=9732e1b25ed4cec5d33eb6dc942b791d1e0de21f;hp=ae65f028956c633daaa1a7819b58f63b590b33ae;hb=d682b3a7e7c7c2941a4d3e193f1e330dbc9fae89;hpb=968f319679d9069af037240d0c3bcd126181cdac diff --git a/src/journal/journald-server.c b/src/journal/journald-server.c index ae65f0289..9732e1b25 100644 --- a/src/journal/journald-server.c +++ b/src/journal/journald-server.c @@ -65,8 +65,8 @@ #define USER_JOURNALS_MAX 1024 #define DEFAULT_SYNC_INTERVAL_USEC (5*USEC_PER_MINUTE) -#define DEFAULT_RATE_LIMIT_INTERVAL (10*USEC_PER_SEC) -#define DEFAULT_RATE_LIMIT_BURST 200 +#define DEFAULT_RATE_LIMIT_INTERVAL (30*USEC_PER_SEC) +#define DEFAULT_RATE_LIMIT_BURST 1000 #define RECHECK_AVAILABLE_SPACE_USEC (30*USEC_PER_SEC) @@ -89,21 +89,22 @@ static const char* const split_mode_table[] = { DEFINE_STRING_TABLE_LOOKUP(split_mode, SplitMode); DEFINE_CONFIG_PARSE_ENUM(config_parse_split_mode, split_mode, SplitMode, "Failed to parse split mode setting"); -static uint64_t available_space(Server *s) { +static uint64_t available_space(Server *s, bool verbose) { char ids[33]; _cleanup_free_ char *p = NULL; - const char *f; sd_id128_t machine; struct statvfs ss; - uint64_t sum = 0, avail = 0, ss_avail = 0; + uint64_t sum = 0, ss_avail = 0, avail = 0; int r; _cleanup_closedir_ DIR *d = NULL; usec_t ts; + const char *f; JournalMetrics *m; ts = now(CLOCK_MONOTONIC); - if (s->cached_available_space_timestamp + RECHECK_AVAILABLE_SPACE_USEC > ts) + if (s->cached_available_space_timestamp + RECHECK_AVAILABLE_SPACE_USEC > ts + && !verbose) return s->cached_available_space; r = sd_id128_get_machine(&machine); @@ -156,38 +157,27 @@ static uint64_t available_space(Server *s) { sum += (uint64_t) st.st_blocks * 512UL; } - avail = sum >= m->max_use ? 0 : m->max_use - sum; - ss_avail = ss.f_bsize * ss.f_bavail; + avail = ss_avail > m->keep_free ? ss_avail - m->keep_free : 0; - ss_avail = ss_avail < m->keep_free ? 0 : ss_avail - m->keep_free; - - if (ss_avail < avail) - avail = ss_avail; - - s->cached_available_space = avail; + s->cached_available_space = MIN(m->max_use, avail) > sum ? MIN(m->max_use, avail) - sum : 0; s->cached_available_space_timestamp = ts; - return avail; -} - -static void server_read_file_gid(Server *s) { - const char *g = "systemd-journal"; - int r; - - assert(s); - - if (s->file_gid_valid) - return; - - r = get_group_creds(&g, &s->file_gid); - if (r < 0) - log_warning("Failed to resolve '%s' group: %s", g, strerror(-r)); + if (verbose) { + char fb1[FORMAT_BYTES_MAX], fb2[FORMAT_BYTES_MAX], fb3[FORMAT_BYTES_MAX], + fb4[FORMAT_BYTES_MAX], fb5[FORMAT_BYTES_MAX]; + + server_driver_message(s, SD_MESSAGE_JOURNAL_USAGE, + "%s journal is using %s (max %s, leaving %s of free %s, current limit %s).", + s->system_journal ? "Permanent" : "Runtime", + format_bytes(fb1, sizeof(fb1), sum), + format_bytes(fb2, sizeof(fb2), m->max_use), + format_bytes(fb3, sizeof(fb3), m->keep_free), + format_bytes(fb4, sizeof(fb4), ss_avail), + format_bytes(fb5, sizeof(fb5), MIN(m->max_use, avail))); + } - /* if we couldn't read the gid, then it will be 0, but that's - * fine and we shouldn't try to resolve the group again, so - * let's just pretend it worked right-away. */ - s->file_gid_valid = true; + return s->cached_available_space; } void server_fix_perms(Server *s, JournalFile *f, uid_t uid) { @@ -200,11 +190,9 @@ void server_fix_perms(Server *s, JournalFile *f, uid_t uid) { assert(f); - server_read_file_gid(s); - - r = fchmod_and_fchown(f->fd, 0640, 0, s->file_gid); + r = fchmod(f->fd, 0640); if (r < 0) - log_warning("Failed to fix access mode/rights on %s, ignoring: %s", f->path, strerror(-r)); + log_warning("Failed to fix access mode on %s, ignoring: %s", f->path, strerror(-r)); #ifdef HAVE_ACL if (uid <= 0) @@ -333,8 +321,10 @@ void server_rotate(Server *s) { if (r < 0) if (f) log_error("Failed to rotate %s: %s", f->path, strerror(-r)); - else + 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)); @@ -343,13 +333,12 @@ void server_rotate(Server *s) { } void server_sync(Server *s) { + static const struct itimerspec sync_timer_disable = {}; JournalFile *f; void *k; Iterator i; int r; - static const struct itimerspec sync_timer_disable = {}; - if (s->system_journal) { r = journal_file_set_offline(s->system_journal); if (r < 0) @@ -434,7 +423,7 @@ bool shall_try_append_again(JournalFile *f, int r) { return true; } -static void write_to_journal(Server *s, uid_t uid, struct iovec *iovec, unsigned n) { +static void write_to_journal(Server *s, uid_t uid, struct iovec *iovec, unsigned n, int priority) { JournalFile *f; bool vacuumed = false; int r; @@ -460,7 +449,7 @@ static void write_to_journal(Server *s, uid_t uid, struct iovec *iovec, unsigned r = journal_file_append_entry(f, NULL, iovec, n, &s->seqnum, NULL, NULL); if (r >= 0) { - server_schedule_sync(s); + server_schedule_sync(s, priority); return; } @@ -490,7 +479,8 @@ static void write_to_journal(Server *s, uid_t uid, struct iovec *iovec, unsigned size += iovec[i].iov_len; log_error("Failed to write entry (%d items, %zu bytes) despite vacuuming, ignoring: %s", n, size, strerror(-r)); - } + } else + server_schedule_sync(s, priority); } static void dispatch_message_real( @@ -500,6 +490,7 @@ static void dispatch_message_real( struct timeval *tv, const char *label, size_t label_len, const char *unit_id, + int priority, pid_t object_pid) { char pid[sizeof("_PID=") + DECIMAL_STR_MAX(pid_t)], @@ -514,7 +505,6 @@ static void dispatch_message_real( o_owner_uid[sizeof("OBJECT_SYSTEMD_OWNER_UID=") + DECIMAL_STR_MAX(uid_t)]; uid_t object_uid; gid_t object_gid; - char *x; sd_id128_t id; int r; @@ -569,6 +559,13 @@ static void dispatch_message_real( IOVEC_SET_STRING(iovec[n++], x); } + r = get_process_capeff(ucred->pid, &t); + if (r >= 0) { + x = strappenda("_CAP_EFFECTIVE=", t); + free(t); + IOVEC_SET_STRING(iovec[n++], x); + } + #ifdef HAVE_AUDIT r = audit_session_from_pid(ucred->pid, &audit); if (r >= 0) { @@ -607,37 +604,46 @@ static void dispatch_message_real( if (cg_path_get_unit(c, &t) >= 0) { x = strappenda("_SYSTEMD_UNIT=", t); free(t); - } else if (cg_path_get_user_unit(c, &t) >= 0) { + IOVEC_SET_STRING(iovec[n++], x); + } else if (unit_id && !session) { + x = strappenda("_SYSTEMD_UNIT=", unit_id); + IOVEC_SET_STRING(iovec[n++], x); + } + + if (cg_path_get_user_unit(c, &t) >= 0) { x = strappenda("_SYSTEMD_USER_UNIT=", t); free(t); - } else if (unit_id) { - if (session) - x = strappenda("_SYSTEMD_USER_UNIT=", unit_id); - else - x = strappenda("_SYSTEMD_UNIT=", unit_id); - } else - x = NULL; - - if (x) IOVEC_SET_STRING(iovec[n++], x); + } else if (unit_id && session) { + x = strappenda("_SYSTEMD_USER_UNIT=", unit_id); + IOVEC_SET_STRING(iovec[n++], x); + } + + if (cg_path_get_slice(c, &t) >= 0) { + x = strappenda("_SYSTEMD_SLICE=", t); + free(t); + IOVEC_SET_STRING(iovec[n++], x); + } free(c); } #ifdef HAVE_SELINUX - if (label) { - x = alloca(sizeof("_SELINUX_CONTEXT=") + label_len); + if (use_selinux()) { + if (label) { + x = alloca(sizeof("_SELINUX_CONTEXT=") + label_len); - *((char*) mempcpy(stpcpy(x, "_SELINUX_CONTEXT="), label, label_len)) = 0; - IOVEC_SET_STRING(iovec[n++], x); - } else { - security_context_t con; + *((char*) mempcpy(stpcpy(x, "_SELINUX_CONTEXT="), label, label_len)) = 0; + IOVEC_SET_STRING(iovec[n++], x); + } else { + security_context_t con; - if (getpidcon(ucred->pid, &con) >= 0) { - x = strappenda("_SELINUX_CONTEXT=", con); + if (getpidcon(ucred->pid, &con) >= 0) { + x = strappenda("_SELINUX_CONTEXT=", con); - freecon(con); - IOVEC_SET_STRING(iovec[n++], x); + freecon(con); + IOVEC_SET_STRING(iovec[n++], x); + } } } #endif @@ -712,14 +718,14 @@ static void dispatch_message_real( if (cg_path_get_unit(c, &t) >= 0) { x = strappenda("OBJECT_SYSTEMD_UNIT=", t); free(t); - } else if (cg_path_get_user_unit(c, &t) >= 0) { + IOVEC_SET_STRING(iovec[n++], x); + } + + if (cg_path_get_user_unit(c, &t) >= 0) { x = strappenda("OBJECT_SYSTEMD_USER_UNIT=", t); free(t); - } else - x = NULL; - - if (x) IOVEC_SET_STRING(iovec[n++], x); + } free(c); } @@ -769,7 +775,7 @@ static void dispatch_message_real( else journal_uid = 0; - write_to_journal(s, journal_uid, iovec, n); + write_to_journal(s, journal_uid, iovec, n, priority); } void server_driver_message(Server *s, sd_id128_t message_id, const char *format, ...) { @@ -803,7 +809,7 @@ void server_driver_message(Server *s, sd_id128_t message_id, const char *format, ucred.uid = getuid(); ucred.gid = getgid(); - dispatch_message_real(s, iovec, n, ELEMENTSOF(iovec), &ucred, NULL, NULL, 0, NULL, 0); + dispatch_message_real(s, iovec, n, ELEMENTSOF(iovec), &ucred, NULL, NULL, 0, NULL, LOG_INFO, 0); } void server_dispatch_message( @@ -829,6 +835,11 @@ void server_dispatch_message( if (LOG_PRI(priority) > s->max_level_store) return; + /* Stop early in case the information will not be stored + * in a journal. */ + if (s->storage == STORAGE_NONE) + return; + if (!ucred) goto finish; @@ -853,7 +864,7 @@ void server_dispatch_message( } rl = journal_rate_limit_test(s->rate_limit, path, - priority & LOG_PRIMASK, available_space(s)); + priority & LOG_PRIMASK, available_space(s, false)); if (rl == 0) return; @@ -864,7 +875,7 @@ void server_dispatch_message( "Suppressed %u messages from %s", rl - 1, path); finish: - dispatch_message_real(s, iovec, n, m, ucred, tv, label, label_len, unit_id, object_pid); + dispatch_message_real(s, iovec, n, m, ucred, tv, label, label_len, unit_id, priority, object_pid); } @@ -875,8 +886,10 @@ static int system_journal_open(Server *s) { char ids[33]; r = sd_id128_get_machine(&machine); - if (r < 0) + if (r < 0) { + log_error("Failed to get machine id: %s", strerror(-r)); return r; + } sd_id128_to_string(machine, ids); @@ -899,23 +912,9 @@ static int system_journal_open(Server *s) { fn = strappenda(fn, "/system.journal"); r = journal_file_open_reliably(fn, O_RDWR|O_CREAT, 0640, s->compress, s->seal, &s->system_metrics, s->mmap, NULL, &s->system_journal); - if (r >= 0) { - char fb[FORMAT_BYTES_MAX]; - uint64_t avail; - + if (r >= 0) server_fix_perms(s, s->system_journal, 0); - - server_driver_message(s, SD_ID128_NULL, "Allowing system journal files to grow to %s.", - format_bytes(fb, sizeof(fb), s->system_metrics.max_use)); - - avail = available_space(s); - - if (s->system_metrics.max_use > avail) - server_driver_message(s, SD_ID128_NULL, "Journal size currently limited to %s due to SystemKeepFree.", - format_bytes(fb, sizeof(fb), avail)); - - } else if (r < 0) { - + else if (r < 0) { if (r != -ENOENT && r != -EROFS) log_warning("Failed to open system journal: %s", strerror(-r)); @@ -961,22 +960,12 @@ static int system_journal_open(Server *s) { } } - if (s->runtime_journal) { - char fb[FORMAT_BYTES_MAX]; - uint64_t avail; - + if (s->runtime_journal) server_fix_perms(s, s->runtime_journal, 0); - server_driver_message(s, SD_ID128_NULL, "Allowing runtime journal files to grow to %s.", - format_bytes(fb, sizeof(fb), s->runtime_metrics.max_use)); - - avail = available_space(s); - - if (s->system_metrics.max_use > avail) - server_driver_message(s, SD_ID128_NULL, "Journal size currently limited to %s due to RuntimeKeepFree.", - format_bytes(fb, sizeof(fb), avail)); - } } + available_space(s, true); + return r; } @@ -1002,10 +991,8 @@ int server_flush_to_var(Server *s) { log_debug("Flushing to /var..."); r = sd_id128_get_machine(&machine); - if (r < 0) { - log_error("Failed to get machine id: %s", strerror(-r)); + if (r < 0) return r; - } r = sd_journal_open(&j, SD_JOURNAL_RUNTIME_ONLY); if (r < 0) { @@ -1077,7 +1064,8 @@ int process_event(Server *s, struct epoll_event *ev) { ssize_t n; if (ev->events != EPOLLIN) { - log_error("Got invalid event from epoll."); + log_error("Got invalid event from epoll for %s: %"PRIx32, + "signal fd", ev->events); return -EIO; } @@ -1094,6 +1082,8 @@ int process_event(Server *s, struct epoll_event *ev) { } if (sfsi.ssi_signo == SIGUSR1) { + log_info("Received request to flush runtime journal from PID %"PRIu32, + sfsi.ssi_pid); touch("/run/systemd/journal/flushed"); server_flush_to_var(s); server_sync(s); @@ -1101,6 +1091,8 @@ int process_event(Server *s, struct epoll_event *ev) { } if (sfsi.ssi_signo == SIGUSR2) { + log_info("Received request to rotate journal from PID %"PRIu32, + sfsi.ssi_pid); server_rotate(s); server_vacuum(s); return 1; @@ -1126,8 +1118,12 @@ int process_event(Server *s, struct epoll_event *ev) { } else if (ev->data.fd == s->dev_kmsg_fd) { int r; - if (ev->events != EPOLLIN) { - log_error("Got invalid event from epoll."); + if (ev->events & EPOLLERR) + log_warning("/dev/kmsg buffer overrun, some messages lost."); + + if (!(ev->events & EPOLLIN)) { + log_error("Got invalid event from epoll for %s: %"PRIx32, + "/dev/kmsg", ev->events); return -EIO; } @@ -1141,7 +1137,9 @@ int process_event(Server *s, struct epoll_event *ev) { ev->data.fd == s->syslog_fd) { if (ev->events != EPOLLIN) { - log_error("Got invalid event from epoll."); + log_error("Got invalid event from epoll for %s: %"PRIx32, + ev->data.fd == s->native_fd ? "native fd" : "syslog fd", + ev->events); return -EIO; } @@ -1228,8 +1226,8 @@ int process_event(Server *s, struct epoll_event *ev) { label = (char*) CMSG_DATA(cmsg); label_len = cmsg->cmsg_len - CMSG_LEN(0); } else if (cmsg->cmsg_level == SOL_SOCKET && - cmsg->cmsg_type == SO_TIMESTAMP && - cmsg->cmsg_len == CMSG_LEN(sizeof(struct timeval))) + cmsg->cmsg_type == SO_TIMESTAMP && + cmsg->cmsg_len == CMSG_LEN(sizeof(struct timeval))) tv = (struct timeval*) CMSG_DATA(cmsg); else if (cmsg->cmsg_level == SOL_SOCKET && cmsg->cmsg_type == SCM_RIGHTS) { @@ -1239,15 +1237,8 @@ int process_event(Server *s, struct epoll_event *ev) { } if (ev->data.fd == s->syslog_fd) { - char *e; - if (n > 0 && n_fds == 0) { - e = memchr(s->buffer, '\n', n); - if (e) - *e = 0; - else - s->buffer[n] = 0; - + s->buffer[n] = 0; server_process_syslog_message(s, strstrip(s->buffer), ucred, tv, label, label_len); } else if (n_fds > 0) log_warning("Got file descriptors via syslog socket. Ignoring."); @@ -1269,7 +1260,8 @@ int process_event(Server *s, struct epoll_event *ev) { } else if (ev->data.fd == s->stdout_fd) { if (ev->events != EPOLLIN) { - log_error("Got invalid event from epoll."); + log_error("Got invalid event from epoll for %s: %"PRIx32, + "stdout fd", ev->events); return -EIO; } @@ -1280,7 +1272,8 @@ int process_event(Server *s, struct epoll_event *ev) { StdoutStream *stream; if ((ev->events|EPOLLIN|EPOLLHUP) != (EPOLLIN|EPOLLHUP)) { - log_error("Got invalid event from epoll."); + log_error("Got invalid event from epoll for %s: %"PRIx32, + "stdout stream", ev->events); return -EIO; } @@ -1425,11 +1418,17 @@ static int server_open_sync_timer(Server *s) { return 0; } -int server_schedule_sync(Server *s) { +int server_schedule_sync(Server *s, int priority) { int r; assert(s); + if (priority <= LOG_CRIT) { + /* Immediately sync to disk when this is of priority CRIT, ALERT, EMERG */ + server_sync(s); + return 0; + } + if (s->sync_scheduled) return 0; @@ -1455,7 +1454,7 @@ int server_init(Server *s) { zero(*s); s->sync_timer_fd = s->syslog_fd = s->native_fd = s->stdout_fd = - s->signal_fd = s->epoll_fd = s->dev_kmsg_fd = -1; + s->signal_fd = s->epoll_fd = s->dev_kmsg_fd = -1; s->compress = true; s->seal = true;