summaryrefslogtreecommitdiff
path: root/src/grp-journal/libjournal-core
diff options
context:
space:
mode:
Diffstat (limited to 'src/grp-journal/libjournal-core')
-rw-r--r--src/grp-journal/libjournal-core/.gitignore4
-rw-r--r--src/grp-journal/libjournal-core/Makefile56
-rw-r--r--src/grp-journal/libjournal-core/journald-audit.c565
-rw-r--r--src/grp-journal/libjournal-core/journald-audit.h28
-rw-r--r--src/grp-journal/libjournal-core/journald-console.c116
-rw-r--r--src/grp-journal/libjournal-core/journald-console.h24
-rw-r--r--src/grp-journal/libjournal-core/journald-gperf.gperf46
-rw-r--r--src/grp-journal/libjournal-core/journald-kmsg.c474
-rw-r--r--src/grp-journal/libjournal-core/journald-kmsg.h29
-rw-r--r--src/grp-journal/libjournal-core/journald-native.c502
-rw-r--r--src/grp-journal/libjournal-core/journald-native.h35
-rw-r--r--src/grp-journal/libjournal-core/journald-rate-limit.c272
-rw-r--r--src/grp-journal/libjournal-core/journald-rate-limit.h28
-rw-r--r--src/grp-journal/libjournal-core/journald-server.c2007
-rw-r--r--src/grp-journal/libjournal-core/journald-server.h187
-rw-r--r--src/grp-journal/libjournal-core/journald-stream.c785
-rw-r--r--src/grp-journal/libjournal-core/journald-stream.h32
-rw-r--r--src/grp-journal/libjournal-core/journald-syslog.c455
-rw-r--r--src/grp-journal/libjournal-core/journald-syslog.h33
-rw-r--r--src/grp-journal/libjournal-core/journald-wall.c72
-rw-r--r--src/grp-journal/libjournal-core/journald-wall.h24
-rw-r--r--src/grp-journal/libjournal-core/test-audit-type.c43
-rw-r--r--src/grp-journal/libjournal-core/test-catalog.c264
-rw-r--r--src/grp-journal/libjournal-core/test-compress-benchmark.c180
-rw-r--r--src/grp-journal/libjournal-core/test-compress.c308
-rw-r--r--src/grp-journal/libjournal-core/test-journal-enum.c53
-rw-r--r--src/grp-journal/libjournal-core/test-journal-flush.c75
-rw-r--r--src/grp-journal/libjournal-core/test-journal-init.c64
-rw-r--r--src/grp-journal/libjournal-core/test-journal-interleaving.c307
-rw-r--r--src/grp-journal/libjournal-core/test-journal-match.c76
-rw-r--r--src/grp-journal/libjournal-core/test-journal-send.c102
-rw-r--r--src/grp-journal/libjournal-core/test-journal-stream.c196
-rw-r--r--src/grp-journal/libjournal-core/test-journal-syslog.c45
-rw-r--r--src/grp-journal/libjournal-core/test-journal-verify.c150
-rw-r--r--src/grp-journal/libjournal-core/test-journal.c178
-rw-r--r--src/grp-journal/libjournal-core/test-mmap-cache.c79
36 files changed, 7894 insertions, 0 deletions
diff --git a/src/grp-journal/libjournal-core/.gitignore b/src/grp-journal/libjournal-core/.gitignore
new file mode 100644
index 0000000000..04d5852547
--- /dev/null
+++ b/src/grp-journal/libjournal-core/.gitignore
@@ -0,0 +1,4 @@
+/journald-gperf.c
+/libsystemd-journal.pc
+/audit_type-list.txt
+/audit_type-*-name.*
diff --git a/src/grp-journal/libjournal-core/Makefile b/src/grp-journal/libjournal-core/Makefile
new file mode 100644
index 0000000000..d55aebfb49
--- /dev/null
+++ b/src/grp-journal/libjournal-core/Makefile
@@ -0,0 +1,56 @@
+# -*- Mode: makefile; indent-tabs-mode: t -*-
+#
+# This file is part of systemd.
+#
+# Copyright 2010-2012 Lennart Poettering
+# Copyright 2010-2012 Kay Sievers
+# Copyright 2013 Zbigniew Jędrzejewski-Szmek
+# Copyright 2013 David Strauss
+# Copyright 2016 Luke Shumaker
+#
+# systemd is free software; you can redistribute it and/or modify it
+# under the terms of the GNU Lesser General Public License as published by
+# the Free Software Foundation; either version 2.1 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
+# Lesser General Public License for more details.
+#
+# You should have received a copy of the GNU Lesser General Public License
+# along with systemd; If not, see <http://www.gnu.org/licenses/>.
+include $(dir $(lastword $(MAKEFILE_LIST)))/../../../config.mk
+include $(topsrcdir)/build-aux/Makefile.head.mk
+
+libjournal_core_la_SOURCES = \
+ src/journal/journald-kmsg.c \
+ src/journal/journald-kmsg.h \
+ src/journal/journald-syslog.c \
+ src/journal/journald-syslog.h \
+ src/journal/journald-stream.c \
+ src/journal/journald-stream.h \
+ src/journal/journald-server.c \
+ src/journal/journald-server.h \
+ src/journal/journald-console.c \
+ src/journal/journald-console.h \
+ src/journal/journald-wall.c \
+ src/journal/journald-wall.h \
+ src/journal/journald-native.c \
+ src/journal/journald-native.h \
+ src/journal/journald-audit.c \
+ src/journal/journald-audit.h \
+ src/journal/journald-rate-limit.c \
+ src/journal/journald-rate-limit.h \
+ src/journal/journal-internal.h
+
+nodist_libjournal_core_la_SOURCES = \
+ src/journal/journald-gperf.c
+
+libjournal_core_la_LIBADD = \
+ libshared.la
+
+noinst_LTLIBRARIES += \
+ libjournal-core.la
+
+include $(topsrcdir)/build-aux/Makefile.tail.mk
diff --git a/src/grp-journal/libjournal-core/journald-audit.c b/src/grp-journal/libjournal-core/journald-audit.c
new file mode 100644
index 0000000000..74ad0c017e
--- /dev/null
+++ b/src/grp-journal/libjournal-core/journald-audit.c
@@ -0,0 +1,565 @@
+/***
+ This file is part of systemd.
+
+ Copyright 2014 Lennart Poettering
+
+ systemd is free software; you can redistribute it and/or modify it
+ under the terms of the GNU Lesser General Public License as published by
+ the Free Software Foundation; either version 2.1 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
+ Lesser General Public License for more details.
+
+ You should have received a copy of the GNU Lesser General Public License
+ along with systemd; If not, see <http://www.gnu.org/licenses/>.
+***/
+
+#include "basic/alloc-util.h"
+#include "basic/fd-util.h"
+#include "basic/hexdecoct.h"
+#include "basic/io-util.h"
+#include "basic/missing.h"
+#include "basic/string-util.h"
+#include "sd-journal/audit-type.h"
+
+#include "journald-audit.h"
+
+typedef struct MapField {
+ const char *audit_field;
+ const char *journal_field;
+ int (*map)(const char *field, const char **p, struct iovec **iov, size_t *n_iov_allocated, unsigned *n_iov);
+} MapField;
+
+static int map_simple_field(const char *field, const char **p, struct iovec **iov, size_t *n_iov_allocated, unsigned *n_iov) {
+ _cleanup_free_ char *c = NULL;
+ size_t l = 0, allocated = 0;
+ const char *e;
+
+ assert(field);
+ assert(p);
+ assert(iov);
+ assert(n_iov);
+
+ l = strlen(field);
+ allocated = l + 1;
+ c = malloc(allocated);
+ if (!c)
+ return -ENOMEM;
+
+ memcpy(c, field, l);
+ for (e = *p; *e != ' ' && *e != 0; e++) {
+ if (!GREEDY_REALLOC(c, allocated, l+2))
+ return -ENOMEM;
+
+ c[l++] = *e;
+ }
+
+ c[l] = 0;
+
+ if (!GREEDY_REALLOC(*iov, *n_iov_allocated, *n_iov + 1))
+ return -ENOMEM;
+
+ (*iov)[*n_iov].iov_base = c;
+ (*iov)[*n_iov].iov_len = l;
+ (*n_iov)++;
+
+ *p = e;
+ c = NULL;
+
+ return 1;
+}
+
+static int map_string_field_internal(const char *field, const char **p, struct iovec **iov, size_t *n_iov_allocated, unsigned *n_iov, bool filter_printable) {
+ _cleanup_free_ char *c = NULL;
+ const char *s, *e;
+ size_t l;
+
+ assert(field);
+ assert(p);
+ assert(iov);
+ assert(n_iov);
+
+ /* The kernel formats string fields in one of two formats. */
+
+ if (**p == '"') {
+ /* Normal quoted syntax */
+ s = *p + 1;
+ e = strchr(s, '"');
+ if (!e)
+ return 0;
+
+ l = strlen(field) + (e - s);
+ c = malloc(l+1);
+ if (!c)
+ return -ENOMEM;
+
+ *((char*) mempcpy(stpcpy(c, field), s, e - s)) = 0;
+
+ e += 1;
+
+ } else if (unhexchar(**p) >= 0) {
+ /* Hexadecimal escaping */
+ size_t allocated = 0;
+
+ l = strlen(field);
+ allocated = l + 2;
+ c = malloc(allocated);
+ if (!c)
+ return -ENOMEM;
+
+ memcpy(c, field, l);
+ for (e = *p; *e != ' ' && *e != 0; e += 2) {
+ int a, b;
+ uint8_t x;
+
+ a = unhexchar(e[0]);
+ if (a < 0)
+ return 0;
+
+ b = unhexchar(e[1]);
+ if (b < 0)
+ return 0;
+
+ x = ((uint8_t) a << 4 | (uint8_t) b);
+
+ if (filter_printable && x < (uint8_t) ' ')
+ x = (uint8_t) ' ';
+
+ if (!GREEDY_REALLOC(c, allocated, l+2))
+ return -ENOMEM;
+
+ c[l++] = (char) x;
+ }
+
+ c[l] = 0;
+ } else
+ return 0;
+
+ if (!GREEDY_REALLOC(*iov, *n_iov_allocated, *n_iov + 1))
+ return -ENOMEM;
+
+ (*iov)[*n_iov].iov_base = c;
+ (*iov)[*n_iov].iov_len = l;
+ (*n_iov)++;
+
+ *p = e;
+ c = NULL;
+
+ return 1;
+}
+
+static int map_string_field(const char *field, const char **p, struct iovec **iov, size_t *n_iov_allocated, unsigned *n_iov) {
+ return map_string_field_internal(field, p, iov, n_iov_allocated, n_iov, false);
+}
+
+static int map_string_field_printable(const char *field, const char **p, struct iovec **iov, size_t *n_iov_allocated, unsigned *n_iov) {
+ return map_string_field_internal(field, p, iov, n_iov_allocated, n_iov, true);
+}
+
+static int map_generic_field(const char *prefix, const char **p, struct iovec **iov, size_t *n_iov_allocated, unsigned *n_iov) {
+ const char *e, *f;
+ char *c, *t;
+ int r;
+
+ /* Implements fallback mappings for all fields we don't know */
+
+ for (e = *p; e < *p + 16; e++) {
+
+ if (*e == 0 || *e == ' ')
+ return 0;
+
+ if (*e == '=')
+ break;
+
+ if (!((*e >= 'a' && *e <= 'z') ||
+ (*e >= 'A' && *e <= 'Z') ||
+ (*e >= '0' && *e <= '9') ||
+ *e == '_' || *e == '-'))
+ return 0;
+ }
+
+ if (e <= *p || e >= *p + 16)
+ return 0;
+
+ c = alloca(strlen(prefix) + (e - *p) + 2);
+
+ t = stpcpy(c, prefix);
+ for (f = *p; f < e; f++) {
+ char x;
+
+ if (*f >= 'a' && *f <= 'z')
+ x = (*f - 'a') + 'A'; /* uppercase */
+ else if (*f == '-')
+ x = '_'; /* dashes → underscores */
+ else
+ x = *f;
+
+ *(t++) = x;
+ }
+ strcpy(t, "=");
+
+ e++;
+
+ r = map_simple_field(c, &e, iov, n_iov_allocated, n_iov);
+ if (r < 0)
+ return r;
+
+ *p = e;
+ return r;
+}
+
+/* Kernel fields are those occurring in the audit string before
+ * msg='. All of these fields are trusted, hence carry the "_" prefix.
+ * We try to translate the fields we know into our native names. The
+ * other's are generically mapped to _AUDIT_FIELD_XYZ= */
+static const MapField map_fields_kernel[] = {
+
+ /* First, we map certain well-known audit fields into native
+ * well-known fields */
+ { "pid=", "_PID=", map_simple_field },
+ { "ppid=", "_PPID=", map_simple_field },
+ { "uid=", "_UID=", map_simple_field },
+ { "euid=", "_EUID=", map_simple_field },
+ { "fsuid=", "_FSUID=", map_simple_field },
+ { "gid=", "_GID=", map_simple_field },
+ { "egid=", "_EGID=", map_simple_field },
+ { "fsgid=", "_FSGID=", map_simple_field },
+ { "tty=", "_TTY=", map_simple_field },
+ { "ses=", "_AUDIT_SESSION=", map_simple_field },
+ { "auid=", "_AUDIT_LOGINUID=", map_simple_field },
+ { "subj=", "_SELINUX_CONTEXT=", map_simple_field },
+ { "comm=", "_COMM=", map_string_field },
+ { "exe=", "_EXE=", map_string_field },
+ { "proctitle=", "_CMDLINE=", map_string_field_printable },
+
+ /* Some fields don't map to native well-known fields. However,
+ * we know that they are string fields, hence let's undo
+ * string field escaping for them, though we stick to the
+ * generic field names. */
+ { "path=", "_AUDIT_FIELD_PATH=", map_string_field },
+ { "dev=", "_AUDIT_FIELD_DEV=", map_string_field },
+ { "name=", "_AUDIT_FIELD_NAME=", map_string_field },
+ {}
+};
+
+/* Userspace fields are those occurring in the audit string after
+ * msg='. All of these fields are untrusted, hence carry no "_"
+ * prefix. We map the fields we don't know to AUDIT_FIELD_XYZ= */
+static const MapField map_fields_userspace[] = {
+ { "cwd=", "AUDIT_FIELD_CWD=", map_string_field },
+ { "cmd=", "AUDIT_FIELD_CMD=", map_string_field },
+ { "acct=", "AUDIT_FIELD_ACCT=", map_string_field },
+ { "exe=", "AUDIT_FIELD_EXE=", map_string_field },
+ { "comm=", "AUDIT_FIELD_COMM=", map_string_field },
+ {}
+};
+
+static int map_all_fields(
+ const char *p,
+ const MapField map_fields[],
+ const char *prefix,
+ bool handle_msg,
+ struct iovec **iov,
+ size_t *n_iov_allocated,
+ unsigned *n_iov) {
+
+ int r;
+
+ assert(p);
+ assert(iov);
+ assert(n_iov_allocated);
+ assert(n_iov);
+
+ for (;;) {
+ bool mapped = false;
+ const MapField *m;
+ const char *v;
+
+ p += strspn(p, WHITESPACE);
+
+ if (*p == 0)
+ return 0;
+
+ if (handle_msg) {
+ v = startswith(p, "msg='");
+ if (v) {
+ const char *e;
+ char *c;
+
+ /* Userspace message. It's enclosed in
+ simple quotation marks, is not
+ escaped, but the last field in the
+ line, hence let's remove the
+ quotation mark, and apply the
+ userspace mapping instead of the
+ kernel mapping. */
+
+ e = endswith(v, "'");
+ if (!e)
+ return 0; /* don't continue splitting up if the final quotation mark is missing */
+
+ c = strndupa(v, e - v);
+ return map_all_fields(c, map_fields_userspace, "AUDIT_FIELD_", false, iov, n_iov_allocated, n_iov);
+ }
+ }
+
+ /* Try to map the kernel fields to our own names */
+ for (m = map_fields; m->audit_field; m++) {
+ v = startswith(p, m->audit_field);
+ if (!v)
+ continue;
+
+ r = m->map(m->journal_field, &v, iov, n_iov_allocated, n_iov);
+ if (r < 0)
+ return log_debug_errno(r, "Failed to parse audit array: %m");
+
+ if (r > 0) {
+ mapped = true;
+ p = v;
+ break;
+ }
+ }
+
+ if (!mapped) {
+ r = map_generic_field(prefix, &p, iov, n_iov_allocated, n_iov);
+ if (r < 0)
+ return log_debug_errno(r, "Failed to parse audit array: %m");
+
+ if (r == 0)
+ /* Couldn't process as generic field, let's just skip over it */
+ p += strcspn(p, WHITESPACE);
+ }
+ }
+}
+
+static void process_audit_string(Server *s, int type, const char *data, size_t size) {
+ _cleanup_free_ struct iovec *iov = NULL;
+ size_t n_iov_allocated = 0;
+ unsigned n_iov = 0, k;
+ uint64_t seconds, msec, id;
+ const char *p, *type_name;
+ unsigned z;
+ char id_field[sizeof("_AUDIT_ID=") + DECIMAL_STR_MAX(uint64_t)],
+ type_field[sizeof("_AUDIT_TYPE=") + DECIMAL_STR_MAX(int)],
+ source_time_field[sizeof("_SOURCE_REALTIME_TIMESTAMP=") + DECIMAL_STR_MAX(usec_t)];
+ char *m;
+
+ assert(s);
+
+ if (size <= 0)
+ return;
+
+ if (!data)
+ return;
+
+ /* Note that the input buffer is NUL terminated, but let's
+ * check whether there is a spurious NUL byte */
+ if (memchr(data, 0, size))
+ return;
+
+ p = startswith(data, "audit");
+ if (!p)
+ return;
+
+ if (sscanf(p, "(%" PRIu64 ".%" PRIu64 ":%" PRIu64 "):%n",
+ &seconds,
+ &msec,
+ &id,
+ &k) != 3)
+ return;
+
+ p += k;
+ p += strspn(p, WHITESPACE);
+
+ if (isempty(p))
+ return;
+
+ n_iov_allocated = N_IOVEC_META_FIELDS + 7;
+ iov = new(struct iovec, n_iov_allocated);
+ if (!iov) {
+ log_oom();
+ return;
+ }
+
+ IOVEC_SET_STRING(iov[n_iov++], "_TRANSPORT=audit");
+
+ sprintf(source_time_field, "_SOURCE_REALTIME_TIMESTAMP=%" PRIu64,
+ (usec_t) seconds * USEC_PER_SEC + (usec_t) msec * USEC_PER_MSEC);
+ IOVEC_SET_STRING(iov[n_iov++], source_time_field);
+
+ sprintf(type_field, "_AUDIT_TYPE=%i", type);
+ IOVEC_SET_STRING(iov[n_iov++], type_field);
+
+ sprintf(id_field, "_AUDIT_ID=%" PRIu64, id);
+ IOVEC_SET_STRING(iov[n_iov++], id_field);
+
+ assert_cc(4 == LOG_FAC(LOG_AUTH));
+ IOVEC_SET_STRING(iov[n_iov++], "SYSLOG_FACILITY=4");
+ IOVEC_SET_STRING(iov[n_iov++], "SYSLOG_IDENTIFIER=audit");
+
+ type_name = audit_type_name_alloca(type);
+
+ m = strjoina("MESSAGE=", type_name, " ", p);
+ IOVEC_SET_STRING(iov[n_iov++], m);
+
+ z = n_iov;
+
+ map_all_fields(p, map_fields_kernel, "_AUDIT_FIELD_", true, &iov, &n_iov_allocated, &n_iov);
+
+ if (!GREEDY_REALLOC(iov, n_iov_allocated, n_iov + N_IOVEC_META_FIELDS)) {
+ log_oom();
+ goto finish;
+ }
+
+ server_dispatch_message(s, iov, n_iov, n_iov_allocated, NULL, NULL, NULL, 0, NULL, LOG_NOTICE, 0);
+
+finish:
+ /* free() all entries that map_all_fields() added. All others
+ * are allocated on the stack or are constant. */
+
+ for (; z < n_iov; z++)
+ free(iov[z].iov_base);
+}
+
+void server_process_audit_message(
+ Server *s,
+ const void *buffer,
+ size_t buffer_size,
+ const struct ucred *ucred,
+ const union sockaddr_union *sa,
+ socklen_t salen) {
+
+ const struct nlmsghdr *nl = buffer;
+
+ assert(s);
+
+ if (buffer_size < ALIGN(sizeof(struct nlmsghdr)))
+ return;
+
+ assert(buffer);
+
+ /* Filter out fake data */
+ if (!sa ||
+ salen != sizeof(struct sockaddr_nl) ||
+ sa->nl.nl_family != AF_NETLINK ||
+ sa->nl.nl_pid != 0) {
+ log_debug("Audit netlink message from invalid sender.");
+ return;
+ }
+
+ if (!ucred || ucred->pid != 0) {
+ log_debug("Audit netlink message with invalid credentials.");
+ return;
+ }
+
+ if (!NLMSG_OK(nl, buffer_size)) {
+ log_error("Audit netlink message truncated.");
+ return;
+ }
+
+ /* Ignore special Netlink messages */
+ if (IN_SET(nl->nlmsg_type, NLMSG_NOOP, NLMSG_ERROR))
+ return;
+
+ /* Below AUDIT_FIRST_USER_MSG theer are only control messages, let's ignore those */
+ if (nl->nlmsg_type < AUDIT_FIRST_USER_MSG)
+ return;
+
+ process_audit_string(s, nl->nlmsg_type, NLMSG_DATA(nl), nl->nlmsg_len - ALIGN(sizeof(struct nlmsghdr)));
+}
+
+static int enable_audit(int fd, bool b) {
+ struct {
+ union {
+ struct nlmsghdr header;
+ uint8_t header_space[NLMSG_HDRLEN];
+ };
+ struct audit_status body;
+ } _packed_ request = {
+ .header.nlmsg_len = NLMSG_LENGTH(sizeof(struct audit_status)),
+ .header.nlmsg_type = AUDIT_SET,
+ .header.nlmsg_flags = NLM_F_REQUEST,
+ .header.nlmsg_seq = 1,
+ .header.nlmsg_pid = 0,
+ .body.mask = AUDIT_STATUS_ENABLED,
+ .body.enabled = b,
+ };
+ union sockaddr_union sa = {
+ .nl.nl_family = AF_NETLINK,
+ .nl.nl_pid = 0,
+ };
+ struct iovec iovec = {
+ .iov_base = &request,
+ .iov_len = NLMSG_LENGTH(sizeof(struct audit_status)),
+ };
+ struct msghdr mh = {
+ .msg_iov = &iovec,
+ .msg_iovlen = 1,
+ .msg_name = &sa.sa,
+ .msg_namelen = sizeof(sa.nl),
+ };
+
+ ssize_t n;
+
+ n = sendmsg(fd, &mh, MSG_NOSIGNAL);
+ if (n < 0)
+ return -errno;
+ if (n != NLMSG_LENGTH(sizeof(struct audit_status)))
+ return -EIO;
+
+ /* We don't wait for the result here, we can't do anything
+ * about it anyway */
+
+ return 0;
+}
+
+int server_open_audit(Server *s) {
+ static const int one = 1;
+ int r;
+
+ if (s->audit_fd < 0) {
+ static const union sockaddr_union sa = {
+ .nl.nl_family = AF_NETLINK,
+ .nl.nl_pid = 0,
+ .nl.nl_groups = AUDIT_NLGRP_READLOG,
+ };
+
+ s->audit_fd = socket(AF_NETLINK, SOCK_RAW|SOCK_CLOEXEC|SOCK_NONBLOCK, NETLINK_AUDIT);
+ if (s->audit_fd < 0) {
+ if (errno == EAFNOSUPPORT || errno == EPROTONOSUPPORT)
+ log_debug("Audit not supported in the kernel.");
+ else
+ log_warning_errno(errno, "Failed to create audit socket, ignoring: %m");
+
+ return 0;
+ }
+
+ if (bind(s->audit_fd, &sa.sa, sizeof(sa.nl)) < 0) {
+ log_warning_errno(errno,
+ "Failed to join audit multicast group. "
+ "The kernel is probably too old or multicast reading is not supported. "
+ "Ignoring: %m");
+ s->audit_fd = safe_close(s->audit_fd);
+ return 0;
+ }
+ } else
+ fd_nonblock(s->audit_fd, 1);
+
+ r = setsockopt(s->audit_fd, SOL_SOCKET, SO_PASSCRED, &one, sizeof(one));
+ if (r < 0)
+ return log_error_errno(errno, "Failed to set SO_PASSCRED on audit socket: %m");
+
+ r = sd_event_add_io(s->event, &s->audit_event_source, s->audit_fd, EPOLLIN, server_process_datagram, s);
+ if (r < 0)
+ return log_error_errno(r, "Failed to add audit fd to event loop: %m");
+
+ /* We are listening now, try to enable audit */
+ r = enable_audit(s->audit_fd, true);
+ if (r < 0)
+ log_warning_errno(r, "Failed to issue audit enable call: %m");
+
+ return 0;
+}
diff --git a/src/grp-journal/libjournal-core/journald-audit.h b/src/grp-journal/libjournal-core/journald-audit.h
new file mode 100644
index 0000000000..98c07e95e4
--- /dev/null
+++ b/src/grp-journal/libjournal-core/journald-audit.h
@@ -0,0 +1,28 @@
+#pragma once
+
+/***
+ This file is part of systemd.
+
+ Copyright 2014 Lennart Poettering
+
+ systemd is free software; you can redistribute it and/or modify it
+ under the terms of the GNU Lesser General Public License as published by
+ the Free Software Foundation; either version 2.1 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
+ Lesser General Public License for more details.
+
+ You should have received a copy of the GNU Lesser General Public License
+ along with systemd; If not, see <http://www.gnu.org/licenses/>.
+***/
+
+#include "basic/socket-util.h"
+
+#include "journald-server.h"
+
+void server_process_audit_message(Server *s, const void *buffer, size_t buffer_size, const struct ucred *ucred, const union sockaddr_union *sa, socklen_t salen);
+
+int server_open_audit(Server*s);
diff --git a/src/grp-journal/libjournal-core/journald-console.c b/src/grp-journal/libjournal-core/journald-console.c
new file mode 100644
index 0000000000..b1e6828dc2
--- /dev/null
+++ b/src/grp-journal/libjournal-core/journald-console.c
@@ -0,0 +1,116 @@
+/***
+ 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 Lesser General Public License as published by
+ the Free Software Foundation; either version 2.1 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
+ Lesser General Public License for more details.
+
+ You should have received a copy of the GNU Lesser General Public License
+ along with systemd; If not, see <http://www.gnu.org/licenses/>.
+***/
+
+#include <fcntl.h>
+#include <sys/socket.h>
+#include <time.h>
+
+#include "basic/alloc-util.h"
+#include "basic/fd-util.h"
+#include "basic/fileio.h"
+#include "basic/formats-util.h"
+#include "basic/io-util.h"
+#include "basic/parse-util.h"
+#include "basic/process-util.h"
+#include "basic/stdio-util.h"
+#include "basic/terminal-util.h"
+
+#include "journald-console.h"
+#include "journald-server.h"
+
+static bool prefix_timestamp(void) {
+
+ static int cached_printk_time = -1;
+
+ if (_unlikely_(cached_printk_time < 0)) {
+ _cleanup_free_ char *p = NULL;
+
+ cached_printk_time =
+ read_one_line_file("/sys/module/printk/parameters/time", &p) >= 0
+ && parse_boolean(p) > 0;
+ }
+
+ return cached_printk_time;
+}
+
+void server_forward_console(
+ Server *s,
+ int priority,
+ const char *identifier,
+ const char *message,
+ const struct ucred *ucred) {
+
+ struct iovec iovec[5];
+ struct timespec ts;
+ char tbuf[sizeof("[] ")-1 + DECIMAL_STR_MAX(ts.tv_sec) + DECIMAL_STR_MAX(ts.tv_nsec)-3 + 1];
+ char header_pid[sizeof("[]: ")-1 + DECIMAL_STR_MAX(pid_t)];
+ int n = 0, fd;
+ _cleanup_free_ char *ident_buf = NULL;
+ const char *tty;
+
+ assert(s);
+ assert(message);
+
+ if (LOG_PRI(priority) > s->max_level_console)
+ return;
+
+ /* First: timestamp */
+ if (prefix_timestamp()) {
+ assert_se(clock_gettime(CLOCK_MONOTONIC, &ts) == 0);
+ xsprintf(tbuf, "[%5"PRI_TIME".%06ld] ",
+ ts.tv_sec,
+ ts.tv_nsec / 1000);
+ IOVEC_SET_STRING(iovec[n++], tbuf);
+ }
+
+ /* Second: identifier and PID */
+ if (ucred) {
+ if (!identifier) {
+ get_process_comm(ucred->pid, &ident_buf);
+ identifier = ident_buf;
+ }
+
+ xsprintf(header_pid, "["PID_FMT"]: ", ucred->pid);
+
+ if (identifier)
+ IOVEC_SET_STRING(iovec[n++], identifier);
+
+ IOVEC_SET_STRING(iovec[n++], header_pid);
+ } else if (identifier) {
+ IOVEC_SET_STRING(iovec[n++], identifier);
+ IOVEC_SET_STRING(iovec[n++], ": ");
+ }
+
+ /* Fourth: message */
+ IOVEC_SET_STRING(iovec[n++], message);
+ IOVEC_SET_STRING(iovec[n++], "\n");
+
+ tty = s->tty_path ? s->tty_path : "/dev/console";
+
+ fd = open_terminal(tty, O_WRONLY|O_NOCTTY|O_CLOEXEC);
+ if (fd < 0) {
+ log_debug_errno(fd, "Failed to open %s for logging: %m", tty);
+ return;
+ }
+
+ if (writev(fd, iovec, n) < 0)
+ log_debug_errno(errno, "Failed to write to %s for logging: %m", tty);
+
+ safe_close(fd);
+}
diff --git a/src/grp-journal/libjournal-core/journald-console.h b/src/grp-journal/libjournal-core/journald-console.h
new file mode 100644
index 0000000000..dda07e2c28
--- /dev/null
+++ b/src/grp-journal/libjournal-core/journald-console.h
@@ -0,0 +1,24 @@
+#pragma once
+
+/***
+ 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 Lesser General Public License as published by
+ the Free Software Foundation; either version 2.1 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
+ Lesser General Public License for more details.
+
+ You should have received a copy of the GNU Lesser General Public License
+ along with systemd; If not, see <http://www.gnu.org/licenses/>.
+***/
+
+#include "journald-server.h"
+
+void server_forward_console(Server *s, int priority, const char *identifier, const char *message, const struct ucred *ucred);
diff --git a/src/grp-journal/libjournal-core/journald-gperf.gperf b/src/grp-journal/libjournal-core/journald-gperf.gperf
new file mode 100644
index 0000000000..7fecd7a964
--- /dev/null
+++ b/src/grp-journal/libjournal-core/journald-gperf.gperf
@@ -0,0 +1,46 @@
+%{
+#include <stddef.h>
+#include <sys/socket.h>
+#include "conf-parser.h"
+#include "journald-server.h"
+%}
+struct ConfigPerfItem;
+%null_strings
+%language=ANSI-C
+%define slot-name section_and_lvalue
+%define hash-function-name journald_gperf_hash
+%define lookup-function-name journald_gperf_lookup
+%readonly-tables
+%omit-struct-type
+%struct-type
+%includes
+%%
+Journal.Storage, config_parse_storage, 0, offsetof(Server, storage)
+Journal.Compress, config_parse_bool, 0, offsetof(Server, compress)
+Journal.Seal, config_parse_bool, 0, offsetof(Server, seal)
+Journal.SyncIntervalSec, config_parse_sec, 0, offsetof(Server, sync_interval_usec)
+# The following is a legacy name for compatibility
+Journal.RateLimitInterval, config_parse_sec, 0, offsetof(Server, rate_limit_interval)
+Journal.RateLimitIntervalSec,config_parse_sec, 0, offsetof(Server, rate_limit_interval)
+Journal.RateLimitBurst, config_parse_unsigned, 0, offsetof(Server, rate_limit_burst)
+Journal.SystemMaxUse, config_parse_iec_uint64, 0, offsetof(Server, system_metrics.max_use)
+Journal.SystemMaxFileSize, config_parse_iec_uint64, 0, offsetof(Server, system_metrics.max_size)
+Journal.SystemKeepFree, config_parse_iec_uint64, 0, offsetof(Server, system_metrics.keep_free)
+Journal.SystemMaxFiles, config_parse_uint64, 0, offsetof(Server, system_metrics.n_max_files)
+Journal.RuntimeMaxUse, config_parse_iec_uint64, 0, offsetof(Server, runtime_metrics.max_use)
+Journal.RuntimeMaxFileSize, config_parse_iec_uint64, 0, offsetof(Server, runtime_metrics.max_size)
+Journal.RuntimeKeepFree, config_parse_iec_uint64, 0, offsetof(Server, runtime_metrics.keep_free)
+Journal.RuntimeMaxFiles, config_parse_uint64, 0, offsetof(Server, runtime_metrics.n_max_files)
+Journal.MaxRetentionSec, config_parse_sec, 0, offsetof(Server, max_retention_usec)
+Journal.MaxFileSec, config_parse_sec, 0, offsetof(Server, max_file_usec)
+Journal.ForwardToSyslog, config_parse_bool, 0, offsetof(Server, forward_to_syslog)
+Journal.ForwardToKMsg, config_parse_bool, 0, offsetof(Server, forward_to_kmsg)
+Journal.ForwardToConsole, config_parse_bool, 0, offsetof(Server, forward_to_console)
+Journal.ForwardToWall, config_parse_bool, 0, offsetof(Server, forward_to_wall)
+Journal.TTYPath, config_parse_path, 0, offsetof(Server, tty_path)
+Journal.MaxLevelStore, config_parse_log_level, 0, offsetof(Server, max_level_store)
+Journal.MaxLevelSyslog, config_parse_log_level, 0, offsetof(Server, max_level_syslog)
+Journal.MaxLevelKMsg, config_parse_log_level, 0, offsetof(Server, max_level_kmsg)
+Journal.MaxLevelConsole, config_parse_log_level, 0, offsetof(Server, max_level_console)
+Journal.MaxLevelWall, config_parse_log_level, 0, offsetof(Server, max_level_wall)
+Journal.SplitMode, config_parse_split_mode, 0, offsetof(Server, split_mode)
diff --git a/src/grp-journal/libjournal-core/journald-kmsg.c b/src/grp-journal/libjournal-core/journald-kmsg.c
new file mode 100644
index 0000000000..946318c184
--- /dev/null
+++ b/src/grp-journal/libjournal-core/journald-kmsg.c
@@ -0,0 +1,474 @@
+/***
+ 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 Lesser General Public License as published by
+ the Free Software Foundation; either version 2.1 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
+ Lesser General Public License for more details.
+
+ You should have received a copy of the GNU Lesser General Public License
+ along with systemd; If not, see <http://www.gnu.org/licenses/>.
+***/
+
+#include <fcntl.h>
+#include <sys/epoll.h>
+#include <sys/mman.h>
+#include <sys/socket.h>
+#include <unistd.h>
+
+#include <libudev.h>
+#include <systemd/sd-messages.h>
+
+#include "basic/escape.h"
+#include "basic/fd-util.h"
+#include "basic/formats-util.h"
+#include "basic/io-util.h"
+#include "basic/parse-util.h"
+#include "basic/process-util.h"
+#include "basic/stdio-util.h"
+#include "basic/string-util.h"
+
+#include "journald-kmsg.h"
+#include "journald-server.h"
+#include "journald-syslog.h"
+
+void server_forward_kmsg(
+ Server *s,
+ int priority,
+ const char *identifier,
+ const char *message,
+ const struct ucred *ucred) {
+
+ struct iovec iovec[5];
+ char header_priority[DECIMAL_STR_MAX(priority) + 3],
+ header_pid[sizeof("[]: ")-1 + DECIMAL_STR_MAX(pid_t) + 1];
+ int n = 0;
+ char *ident_buf = NULL;
+
+ assert(s);
+ assert(priority >= 0);
+ assert(priority <= 999);
+ assert(message);
+
+ if (_unlikely_(LOG_PRI(priority) > s->max_level_kmsg))
+ return;
+
+ if (_unlikely_(s->dev_kmsg_fd < 0))
+ return;
+
+ /* Never allow messages with kernel facility to be written to
+ * kmsg, regardless where the data comes from. */
+ priority = syslog_fixup_facility(priority);
+
+ /* First: priority field */
+ xsprintf(header_priority, "<%i>", priority);
+ IOVEC_SET_STRING(iovec[n++], header_priority);
+
+ /* Second: identifier and PID */
+ if (ucred) {
+ if (!identifier) {
+ get_process_comm(ucred->pid, &ident_buf);
+ identifier = ident_buf;
+ }
+
+ xsprintf(header_pid, "["PID_FMT"]: ", ucred->pid);
+
+ if (identifier)
+ IOVEC_SET_STRING(iovec[n++], identifier);
+
+ IOVEC_SET_STRING(iovec[n++], header_pid);
+ } else if (identifier) {
+ IOVEC_SET_STRING(iovec[n++], identifier);
+ IOVEC_SET_STRING(iovec[n++], ": ");
+ }
+
+ /* Fourth: message */
+ IOVEC_SET_STRING(iovec[n++], message);
+ IOVEC_SET_STRING(iovec[n++], "\n");
+
+ if (writev(s->dev_kmsg_fd, iovec, n) < 0)
+ log_debug_errno(errno, "Failed to write to /dev/kmsg for logging: %m");
+
+ free(ident_buf);
+}
+
+static bool is_us(const char *pid) {
+ pid_t t;
+
+ assert(pid);
+
+ if (parse_pid(pid, &t) < 0)
+ return false;
+
+ return t == getpid();
+}
+
+static void dev_kmsg_record(Server *s, const char *p, size_t l) {
+ struct iovec iovec[N_IOVEC_META_FIELDS + 7 + N_IOVEC_KERNEL_FIELDS + 2 + N_IOVEC_UDEV_FIELDS];
+ char *message = NULL, *syslog_priority = NULL, *syslog_pid = NULL, *syslog_facility = NULL, *syslog_identifier = NULL, *source_time = NULL;
+ int priority, r;
+ unsigned n = 0, z = 0, j;
+ unsigned long long usec;
+ char *identifier = NULL, *pid = NULL, *e, *f, *k;
+ uint64_t serial;
+ size_t pl;
+ char *kernel_device = NULL;
+
+ assert(s);
+ assert(p);
+
+ if (l <= 0)
+ return;
+
+ e = memchr(p, ',', l);
+ if (!e)
+ return;
+ *e = 0;
+
+ r = safe_atoi(p, &priority);
+ if (r < 0 || priority < 0 || priority > 999)
+ return;
+
+ if (s->forward_to_kmsg && (priority & LOG_FACMASK) != LOG_KERN)
+ return;
+
+ l -= (e - p) + 1;
+ p = e + 1;
+ e = memchr(p, ',', l);
+ if (!e)
+ return;
+ *e = 0;
+
+ r = safe_atou64(p, &serial);
+ if (r < 0)
+ return;
+
+ if (s->kernel_seqnum) {
+ /* We already read this one? */
+ if (serial < *s->kernel_seqnum)
+ return;
+
+ /* Did we lose any? */
+ if (serial > *s->kernel_seqnum)
+ server_driver_message(s, SD_MESSAGE_JOURNAL_MISSED,
+ LOG_MESSAGE("Missed %"PRIu64" kernel messages",
+ serial - *s->kernel_seqnum),
+ NULL);
+
+ /* Make sure we never read this one again. Note that
+ * we always store the next message serial we expect
+ * here, simply because this makes handling the first
+ * message with serial 0 easy. */
+ *s->kernel_seqnum = serial + 1;
+ }
+
+ l -= (e - p) + 1;
+ p = e + 1;
+ f = memchr(p, ';', l);
+ if (!f)
+ return;
+ /* Kernel 3.6 has the flags field, kernel 3.5 lacks that */
+ e = memchr(p, ',', l);
+ if (!e || f < e)
+ e = f;
+ *e = 0;
+
+ r = safe_atollu(p, &usec);
+ if (r < 0)
+ return;
+
+ l -= (f - p) + 1;
+ p = f + 1;
+ e = memchr(p, '\n', l);
+ if (!e)
+ return;
+ *e = 0;
+
+ pl = e - p;
+ l -= (e - p) + 1;
+ k = e + 1;
+
+ for (j = 0; l > 0 && j < N_IOVEC_KERNEL_FIELDS; j++) {
+ char *m;
+ /* Metadata fields attached */
+
+ if (*k != ' ')
+ break;
+
+ k++, l--;
+
+ e = memchr(k, '\n', l);
+ if (!e)
+ return;
+
+ *e = 0;
+
+ if (cunescape_length_with_prefix(k, e - k, "_KERNEL_", UNESCAPE_RELAX, &m) < 0)
+ break;
+
+ if (startswith(m, "_KERNEL_DEVICE="))
+ kernel_device = m + 15;
+
+ IOVEC_SET_STRING(iovec[n++], m);
+ z++;
+
+ l -= (e - k) + 1;
+ k = e + 1;
+ }
+
+ if (kernel_device) {
+ struct udev_device *ud;
+
+ ud = udev_device_new_from_device_id(s->udev, kernel_device);
+ if (ud) {
+ const char *g;
+ struct udev_list_entry *ll;
+ char *b;
+
+ g = udev_device_get_devnode(ud);
+ if (g) {
+ b = strappend("_UDEV_DEVNODE=", g);
+ if (b) {
+ IOVEC_SET_STRING(iovec[n++], b);
+ z++;
+ }
+ }
+
+ g = udev_device_get_sysname(ud);
+ if (g) {
+ b = strappend("_UDEV_SYSNAME=", g);
+ if (b) {
+ IOVEC_SET_STRING(iovec[n++], b);
+ z++;
+ }
+ }
+
+ j = 0;
+ ll = udev_device_get_devlinks_list_entry(ud);
+ udev_list_entry_foreach(ll, ll) {
+
+ if (j > N_IOVEC_UDEV_FIELDS)
+ break;
+
+ g = udev_list_entry_get_name(ll);
+ if (g) {
+ b = strappend("_UDEV_DEVLINK=", g);
+ if (b) {
+ IOVEC_SET_STRING(iovec[n++], b);
+ z++;
+ }
+ }
+
+ j++;
+ }
+
+ udev_device_unref(ud);
+ }
+ }
+
+ if (asprintf(&source_time, "_SOURCE_MONOTONIC_TIMESTAMP=%llu", usec) >= 0)
+ IOVEC_SET_STRING(iovec[n++], source_time);
+
+ IOVEC_SET_STRING(iovec[n++], "_TRANSPORT=kernel");
+
+ if (asprintf(&syslog_priority, "PRIORITY=%i", priority & LOG_PRIMASK) >= 0)
+ IOVEC_SET_STRING(iovec[n++], syslog_priority);
+
+ if (asprintf(&syslog_facility, "SYSLOG_FACILITY=%i", LOG_FAC(priority)) >= 0)
+ IOVEC_SET_STRING(iovec[n++], syslog_facility);
+
+ if ((priority & LOG_FACMASK) == LOG_KERN)
+ IOVEC_SET_STRING(iovec[n++], "SYSLOG_IDENTIFIER=kernel");
+ else {
+ pl -= syslog_parse_identifier((const char**) &p, &identifier, &pid);
+
+ /* Avoid any messages we generated ourselves via
+ * log_info() and friends. */
+ if (pid && is_us(pid))
+ goto finish;
+
+ if (identifier) {
+ syslog_identifier = strappend("SYSLOG_IDENTIFIER=", identifier);
+ if (syslog_identifier)
+ IOVEC_SET_STRING(iovec[n++], syslog_identifier);
+ }
+
+ if (pid) {
+ syslog_pid = strappend("SYSLOG_PID=", pid);
+ if (syslog_pid)
+ IOVEC_SET_STRING(iovec[n++], syslog_pid);
+ }
+ }
+
+ if (cunescape_length_with_prefix(p, pl, "MESSAGE=", UNESCAPE_RELAX, &message) >= 0)
+ IOVEC_SET_STRING(iovec[n++], message);
+
+ server_dispatch_message(s, iovec, n, ELEMENTSOF(iovec), NULL, NULL, NULL, 0, NULL, priority, 0);
+
+finish:
+ for (j = 0; j < z; j++)
+ free(iovec[j].iov_base);
+
+ free(message);
+ free(syslog_priority);
+ free(syslog_identifier);
+ free(syslog_pid);
+ free(syslog_facility);
+ free(source_time);
+ free(identifier);
+ free(pid);
+}
+
+static int server_read_dev_kmsg(Server *s) {
+ char buffer[8192+1]; /* the kernel-side limit per record is 8K currently */
+ ssize_t l;
+
+ assert(s);
+ assert(s->dev_kmsg_fd >= 0);
+
+ l = read(s->dev_kmsg_fd, buffer, sizeof(buffer) - 1);
+ if (l == 0)
+ return 0;
+ if (l < 0) {
+ /* Old kernels who don't allow reading from /dev/kmsg
+ * return EINVAL when we try. So handle this cleanly,
+ * but don' try to ever read from it again. */
+ if (errno == EINVAL) {
+ s->dev_kmsg_event_source = sd_event_source_unref(s->dev_kmsg_event_source);
+ return 0;
+ }
+
+ if (errno == EAGAIN || errno == EINTR || errno == EPIPE)
+ return 0;
+
+ return log_error_errno(errno, "Failed to read from kernel: %m");
+ }
+
+ dev_kmsg_record(s, buffer, l);
+ return 1;
+}
+
+int server_flush_dev_kmsg(Server *s) {
+ int r;
+
+ assert(s);
+
+ if (s->dev_kmsg_fd < 0)
+ return 0;
+
+ if (!s->dev_kmsg_readable)
+ return 0;
+
+ log_debug("Flushing /dev/kmsg...");
+
+ for (;;) {
+ r = server_read_dev_kmsg(s);
+ if (r < 0)
+ return r;
+
+ if (r == 0)
+ break;
+ }
+
+ return 0;
+}
+
+static int dispatch_dev_kmsg(sd_event_source *es, int fd, uint32_t revents, void *userdata) {
+ Server *s = userdata;
+
+ assert(es);
+ assert(fd == s->dev_kmsg_fd);
+ assert(s);
+
+ if (revents & EPOLLERR)
+ log_warning("/dev/kmsg buffer overrun, some messages lost.");
+
+ if (!(revents & EPOLLIN))
+ log_error("Got invalid event from epoll for /dev/kmsg: %"PRIx32, revents);
+
+ return server_read_dev_kmsg(s);
+}
+
+int server_open_dev_kmsg(Server *s) {
+ int r;
+
+ assert(s);
+
+ s->dev_kmsg_fd = open("/dev/kmsg", O_RDWR|O_CLOEXEC|O_NONBLOCK|O_NOCTTY);
+ if (s->dev_kmsg_fd < 0) {
+ log_full(errno == ENOENT ? LOG_DEBUG : LOG_WARNING,
+ "Failed to open /dev/kmsg, ignoring: %m");
+ return 0;
+ }
+
+ r = sd_event_add_io(s->event, &s->dev_kmsg_event_source, s->dev_kmsg_fd, EPOLLIN, dispatch_dev_kmsg, s);
+ if (r < 0) {
+
+ /* This will fail with EPERM on older kernels where
+ * /dev/kmsg is not readable. */
+ if (r == -EPERM) {
+ r = 0;
+ goto fail;
+ }
+
+ log_error_errno(r, "Failed to add /dev/kmsg fd to event loop: %m");
+ goto fail;
+ }
+
+ r = sd_event_source_set_priority(s->dev_kmsg_event_source, SD_EVENT_PRIORITY_IMPORTANT+10);
+ if (r < 0) {
+ log_error_errno(r, "Failed to adjust priority of kmsg event source: %m");
+ goto fail;
+ }
+
+ s->dev_kmsg_readable = true;
+
+ return 0;
+
+fail:
+ s->dev_kmsg_event_source = sd_event_source_unref(s->dev_kmsg_event_source);
+ s->dev_kmsg_fd = safe_close(s->dev_kmsg_fd);
+
+ return r;
+}
+
+int server_open_kernel_seqnum(Server *s) {
+ _cleanup_close_ int fd;
+ uint64_t *p;
+ int r;
+
+ assert(s);
+
+ /* We store the seqnum we last read in an mmaped file. That
+ * way we can just use it like a variable, but it is
+ * persistent and automatically flushed at reboot. */
+
+ fd = open("/run/systemd/journal/kernel-seqnum", O_RDWR|O_CREAT|O_CLOEXEC|O_NOCTTY|O_NOFOLLOW, 0644);
+ if (fd < 0) {
+ log_error_errno(errno, "Failed to open /run/systemd/journal/kernel-seqnum, ignoring: %m");
+ return 0;
+ }
+
+ r = posix_fallocate(fd, 0, sizeof(uint64_t));
+ if (r != 0) {
+ log_error_errno(r, "Failed to allocate sequential number file, ignoring: %m");
+ return 0;
+ }
+
+ p = mmap(NULL, sizeof(uint64_t), PROT_READ|PROT_WRITE, MAP_SHARED, fd, 0);
+ if (p == MAP_FAILED) {
+ log_error_errno(errno, "Failed to map sequential number file, ignoring: %m");
+ return 0;
+ }
+
+ s->kernel_seqnum = p;
+
+ return 0;
+}
diff --git a/src/grp-journal/libjournal-core/journald-kmsg.h b/src/grp-journal/libjournal-core/journald-kmsg.h
new file mode 100644
index 0000000000..dab49f1e8c
--- /dev/null
+++ b/src/grp-journal/libjournal-core/journald-kmsg.h
@@ -0,0 +1,29 @@
+#pragma once
+
+/***
+ 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 Lesser General Public License as published by
+ the Free Software Foundation; either version 2.1 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
+ Lesser General Public License for more details.
+
+ You should have received a copy of the GNU Lesser General Public License
+ along with systemd; If not, see <http://www.gnu.org/licenses/>.
+***/
+
+#include "journald-server.h"
+
+int server_open_dev_kmsg(Server *s);
+int server_flush_dev_kmsg(Server *s);
+
+void server_forward_kmsg(Server *s, int priority, const char *identifier, const char *message, const struct ucred *ucred);
+
+int server_open_kernel_seqnum(Server *s);
diff --git a/src/grp-journal/libjournal-core/journald-native.c b/src/grp-journal/libjournal-core/journald-native.c
new file mode 100644
index 0000000000..9bfea935ad
--- /dev/null
+++ b/src/grp-journal/libjournal-core/journald-native.c
@@ -0,0 +1,502 @@
+/***
+ 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 Lesser General Public License as published by
+ the Free Software Foundation; either version 2.1 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
+ Lesser General Public License for more details.
+
+ You should have received a copy of the GNU Lesser General Public License
+ along with systemd; If not, see <http://www.gnu.org/licenses/>.
+***/
+
+#include <stddef.h>
+#include <sys/epoll.h>
+#include <sys/mman.h>
+#include <sys/statvfs.h>
+#include <unistd.h>
+
+#include "basic/alloc-util.h"
+#include "basic/fd-util.h"
+#include "basic/fs-util.h"
+#include "basic/io-util.h"
+#include "basic/memfd-util.h"
+#include "basic/parse-util.h"
+#include "basic/path-util.h"
+#include "basic/selinux-util.h"
+#include "basic/socket-util.h"
+#include "basic/string-util.h"
+
+#include "journald-console.h"
+#include "journald-kmsg.h"
+#include "journald-native.h"
+#include "journald-server.h"
+#include "journald-syslog.h"
+#include "journald-wall.h"
+
+bool valid_user_field(const char *p, size_t l, bool allow_protected) {
+ const char *a;
+
+ /* We kinda enforce POSIX syntax recommendations for
+ environment variables here, but make a couple of additional
+ requirements.
+
+ http://pubs.opengroup.org/onlinepubs/000095399/basedefs/xbd_chap08.html */
+
+ /* No empty field names */
+ if (l <= 0)
+ return false;
+
+ /* Don't allow names longer than 64 chars */
+ if (l > 64)
+ return false;
+
+ /* Variables starting with an underscore are protected */
+ if (!allow_protected && p[0] == '_')
+ return false;
+
+ /* Don't allow digits as first character */
+ if (p[0] >= '0' && p[0] <= '9')
+ return false;
+
+ /* Only allow A-Z0-9 and '_' */
+ for (a = p; a < p + l; a++)
+ if ((*a < 'A' || *a > 'Z') &&
+ (*a < '0' || *a > '9') &&
+ *a != '_')
+ return false;
+
+ return true;
+}
+
+static bool allow_object_pid(const struct ucred *ucred) {
+ return ucred && ucred->uid == 0;
+}
+
+void server_process_native_message(
+ Server *s,
+ const void *buffer, size_t buffer_size,
+ const struct ucred *ucred,
+ const struct timeval *tv,
+ const char *label, size_t label_len) {
+
+ struct iovec *iovec = NULL;
+ unsigned n = 0, j, tn = (unsigned) -1;
+ const char *p;
+ size_t remaining, m = 0, entry_size = 0;
+ int priority = LOG_INFO;
+ char *identifier = NULL, *message = NULL;
+ pid_t object_pid = 0;
+
+ assert(s);
+ assert(buffer || buffer_size == 0);
+
+ p = buffer;
+ remaining = buffer_size;
+
+ while (remaining > 0) {
+ const char *e, *q;
+
+ e = memchr(p, '\n', remaining);
+
+ if (!e) {
+ /* Trailing noise, let's ignore it, and flush what we collected */
+ log_debug("Received message with trailing noise, ignoring.");
+ break;
+ }
+
+ if (e == p) {
+ /* Entry separator */
+
+ if (entry_size + n + 1 > ENTRY_SIZE_MAX) { /* data + separators + trailer */
+ log_debug("Entry is too big with %u properties and %zu bytes, ignoring.", n, entry_size);
+ continue;
+ }
+
+ server_dispatch_message(s, iovec, n, m, ucred, tv, label, label_len, NULL, priority, object_pid);
+ n = 0;
+ priority = LOG_INFO;
+ entry_size = 0;
+
+ p++;
+ remaining--;
+ continue;
+ }
+
+ if (*p == '.' || *p == '#') {
+ /* Ignore control commands for now, and
+ * comments too. */
+ remaining -= (e - p) + 1;
+ p = e + 1;
+ continue;
+ }
+
+ /* A property follows */
+
+ /* n existing properties, 1 new, +1 for _TRANSPORT */
+ if (!GREEDY_REALLOC(iovec, m, n + 2 + N_IOVEC_META_FIELDS + N_IOVEC_OBJECT_FIELDS)) {
+ log_oom();
+ break;
+ }
+
+ q = memchr(p, '=', e - p);
+ if (q) {
+ if (valid_user_field(p, q - p, false)) {
+ size_t l;
+
+ l = e - p;
+
+ /* If the field name starts with an
+ * underscore, skip the variable,
+ * since that indidates a trusted
+ * field */
+ iovec[n].iov_base = (char*) p;
+ iovec[n].iov_len = l;
+ entry_size += iovec[n].iov_len;
+ n++;
+
+ /* We need to determine the priority
+ * of this entry for the rate limiting
+ * logic */
+ if (l == 10 &&
+ startswith(p, "PRIORITY=") &&
+ p[9] >= '0' && p[9] <= '9')
+ priority = (priority & LOG_FACMASK) | (p[9] - '0');
+
+ else if (l == 17 &&
+ startswith(p, "SYSLOG_FACILITY=") &&
+ p[16] >= '0' && p[16] <= '9')
+ priority = (priority & LOG_PRIMASK) | ((p[16] - '0') << 3);
+
+ else if (l == 18 &&
+ startswith(p, "SYSLOG_FACILITY=") &&
+ p[16] >= '0' && p[16] <= '9' &&
+ p[17] >= '0' && p[17] <= '9')
+ priority = (priority & LOG_PRIMASK) | (((p[16] - '0')*10 + (p[17] - '0')) << 3);
+
+ else if (l >= 19 &&
+ startswith(p, "SYSLOG_IDENTIFIER=")) {
+ char *t;
+
+ t = strndup(p + 18, l - 18);
+ if (t) {
+ free(identifier);
+ identifier = t;
+ }
+
+ } else if (l >= 8 &&
+ startswith(p, "MESSAGE=")) {
+ char *t;
+
+ t = strndup(p + 8, l - 8);
+ if (t) {
+ free(message);
+ message = t;
+ }
+
+ } else if (l > strlen("OBJECT_PID=") &&
+ l < strlen("OBJECT_PID=") + DECIMAL_STR_MAX(pid_t) &&
+ startswith(p, "OBJECT_PID=") &&
+ allow_object_pid(ucred)) {
+ char buf[DECIMAL_STR_MAX(pid_t)];
+ memcpy(buf, p + strlen("OBJECT_PID="), l - strlen("OBJECT_PID="));
+ buf[l-strlen("OBJECT_PID=")] = '\0';
+
+ /* ignore error */
+ parse_pid(buf, &object_pid);
+ }
+ }
+
+ remaining -= (e - p) + 1;
+ p = e + 1;
+ continue;
+ } else {
+ le64_t l_le;
+ uint64_t l;
+ char *k;
+
+ if (remaining < e - p + 1 + sizeof(uint64_t) + 1) {
+ log_debug("Failed to parse message, ignoring.");
+ break;
+ }
+
+ memcpy(&l_le, e + 1, sizeof(uint64_t));
+ l = le64toh(l_le);
+
+ if (l > DATA_SIZE_MAX) {
+ log_debug("Received binary data block of %"PRIu64" bytes is too large, ignoring.", l);
+ break;
+ }
+
+ if ((uint64_t) remaining < e - p + 1 + sizeof(uint64_t) + l + 1 ||
+ e[1+sizeof(uint64_t)+l] != '\n') {
+ log_debug("Failed to parse message, ignoring.");
+ break;
+ }
+
+ k = malloc((e - p) + 1 + l);
+ if (!k) {
+ log_oom();
+ break;
+ }
+
+ memcpy(k, p, e - p);
+ k[e - p] = '=';
+ memcpy(k + (e - p) + 1, e + 1 + sizeof(uint64_t), l);
+
+ if (valid_user_field(p, e - p, false)) {
+ iovec[n].iov_base = k;
+ iovec[n].iov_len = (e - p) + 1 + l;
+ entry_size += iovec[n].iov_len;
+ n++;
+ } else
+ free(k);
+
+ remaining -= (e - p) + 1 + sizeof(uint64_t) + l + 1;
+ p = e + 1 + sizeof(uint64_t) + l + 1;
+ }
+ }
+
+ if (n <= 0)
+ goto finish;
+
+ tn = n++;
+ IOVEC_SET_STRING(iovec[tn], "_TRANSPORT=journal");
+ entry_size += strlen("_TRANSPORT=journal");
+
+ if (entry_size + n + 1 > ENTRY_SIZE_MAX) { /* data + separators + trailer */
+ log_debug("Entry is too big with %u properties and %zu bytes, ignoring.",
+ n, entry_size);
+ goto finish;
+ }
+
+ if (message) {
+ if (s->forward_to_syslog)
+ server_forward_syslog(s, priority, identifier, message, ucred, tv);
+
+ if (s->forward_to_kmsg)
+ server_forward_kmsg(s, priority, identifier, message, ucred);
+
+ if (s->forward_to_console)
+ server_forward_console(s, priority, identifier, message, ucred);
+
+ if (s->forward_to_wall)
+ server_forward_wall(s, priority, identifier, message, ucred);
+ }
+
+ server_dispatch_message(s, iovec, n, m, ucred, tv, label, label_len, NULL, priority, object_pid);
+
+finish:
+ for (j = 0; j < n; j++) {
+ if (j == tn)
+ continue;
+
+ if (iovec[j].iov_base < buffer ||
+ (const uint8_t*) iovec[j].iov_base >= (const uint8_t*) buffer + buffer_size)
+ free(iovec[j].iov_base);
+ }
+
+ free(iovec);
+ free(identifier);
+ free(message);
+}
+
+void server_process_native_file(
+ Server *s,
+ int fd,
+ const struct ucred *ucred,
+ const struct timeval *tv,
+ const char *label, size_t label_len) {
+
+ struct stat st;
+ bool sealed;
+ int r;
+
+ /* Data is in the passed fd, since it didn't fit in a
+ * datagram. */
+
+ assert(s);
+ assert(fd >= 0);
+
+ /* If it's a memfd, check if it is sealed. If so, we can just
+ * use map it and use it, and do not need to copy the data
+ * out. */
+ sealed = memfd_get_sealed(fd) > 0;
+
+ if (!sealed && (!ucred || ucred->uid != 0)) {
+ _cleanup_free_ char *sl = NULL, *k = NULL;
+ const char *e;
+
+ /* If this is not a sealed memfd, and the peer is unknown or
+ * unprivileged, then verify the path. */
+
+ if (asprintf(&sl, "/proc/self/fd/%i", fd) < 0) {
+ log_oom();
+ return;
+ }
+
+ r = readlink_malloc(sl, &k);
+ if (r < 0) {
+ log_error_errno(r, "readlink(%s) failed: %m", sl);
+ return;
+ }
+
+ e = path_startswith(k, "/dev/shm/");
+ if (!e)
+ e = path_startswith(k, "/tmp/");
+ if (!e)
+ e = path_startswith(k, "/var/tmp/");
+ if (!e) {
+ log_error("Received file outside of allowed directories. Refusing.");
+ return;
+ }
+
+ if (!filename_is_valid(e)) {
+ log_error("Received file in subdirectory of allowed directories. Refusing.");
+ return;
+ }
+ }
+
+ if (fstat(fd, &st) < 0) {
+ log_error_errno(errno, "Failed to stat passed file, ignoring: %m");
+ return;
+ }
+
+ if (!S_ISREG(st.st_mode)) {
+ log_error("File passed is not regular. Ignoring.");
+ return;
+ }
+
+ if (st.st_size <= 0)
+ return;
+
+ if (st.st_size > ENTRY_SIZE_MAX) {
+ log_error("File passed too large. Ignoring.");
+ return;
+ }
+
+ if (sealed) {
+ void *p;
+ size_t ps;
+
+ /* The file is sealed, we can just map it and use it. */
+
+ ps = PAGE_ALIGN(st.st_size);
+ p = mmap(NULL, ps, PROT_READ, MAP_PRIVATE, fd, 0);
+ if (p == MAP_FAILED) {
+ log_error_errno(errno, "Failed to map memfd, ignoring: %m");
+ return;
+ }
+
+ server_process_native_message(s, p, st.st_size, ucred, tv, label, label_len);
+ assert_se(munmap(p, ps) >= 0);
+ } else {
+ _cleanup_free_ void *p = NULL;
+ struct statvfs vfs;
+ ssize_t n;
+
+ if (fstatvfs(fd, &vfs) < 0) {
+ log_error_errno(errno, "Failed to stat file system of passed file, ignoring: %m");
+ return;
+ }
+
+ /* Refuse operating on file systems that have
+ * mandatory locking enabled, see:
+ *
+ * https://github.com/systemd/systemd/issues/1822
+ */
+ if (vfs.f_flag & ST_MANDLOCK) {
+ log_error("Received file descriptor from file system with mandatory locking enable, refusing.");
+ return;
+ }
+
+ /* Make the fd non-blocking. On regular files this has
+ * the effect of bypassing mandatory locking. Of
+ * course, this should normally not be necessary given
+ * the check above, but let's better be safe than
+ * sorry, after all NFS is pretty confusing regarding
+ * file system flags, and we better don't trust it,
+ * and so is SMB. */
+ r = fd_nonblock(fd, true);
+ if (r < 0) {
+ log_error_errno(r, "Failed to make fd non-blocking, ignoring: %m");
+ return;
+ }
+
+ /* The file is not sealed, we can't map the file here, since
+ * clients might then truncate it and trigger a SIGBUS for
+ * us. So let's stupidly read it */
+
+ p = malloc(st.st_size);
+ if (!p) {
+ log_oom();
+ return;
+ }
+
+ n = pread(fd, p, st.st_size, 0);
+ if (n < 0)
+ log_error_errno(errno, "Failed to read file, ignoring: %m");
+ else if (n > 0)
+ server_process_native_message(s, p, n, ucred, tv, label, label_len);
+ }
+}
+
+int server_open_native_socket(Server*s) {
+
+ static const union sockaddr_union sa = {
+ .un.sun_family = AF_UNIX,
+ .un.sun_path = "/run/systemd/journal/socket",
+ };
+ static const int one = 1;
+ int r;
+
+ assert(s);
+
+ if (s->native_fd < 0) {
+ s->native_fd = socket(AF_UNIX, SOCK_DGRAM|SOCK_CLOEXEC|SOCK_NONBLOCK, 0);
+ if (s->native_fd < 0)
+ return log_error_errno(errno, "socket() failed: %m");
+
+ (void) unlink(sa.un.sun_path);
+
+ r = bind(s->native_fd, &sa.sa, SOCKADDR_UN_LEN(sa.un));
+ if (r < 0)
+ return log_error_errno(errno, "bind(%s) failed: %m", sa.un.sun_path);
+
+ (void) chmod(sa.un.sun_path, 0666);
+ } else
+ fd_nonblock(s->native_fd, 1);
+
+ r = setsockopt(s->native_fd, SOL_SOCKET, SO_PASSCRED, &one, sizeof(one));
+ if (r < 0)
+ return log_error_errno(errno, "SO_PASSCRED failed: %m");
+
+#ifdef HAVE_SELINUX
+ if (mac_selinux_have()) {
+ r = setsockopt(s->native_fd, SOL_SOCKET, SO_PASSSEC, &one, sizeof(one));
+ if (r < 0)
+ log_warning_errno(errno, "SO_PASSSEC failed: %m");
+ }
+#endif
+
+ r = setsockopt(s->native_fd, SOL_SOCKET, SO_TIMESTAMP, &one, sizeof(one));
+ if (r < 0)
+ return log_error_errno(errno, "SO_TIMESTAMP failed: %m");
+
+ r = sd_event_add_io(s->event, &s->native_event_source, s->native_fd, EPOLLIN, server_process_datagram, s);
+ if (r < 0)
+ return log_error_errno(r, "Failed to add native server fd to event loop: %m");
+
+ r = sd_event_source_set_priority(s->native_event_source, SD_EVENT_PRIORITY_NORMAL+5);
+ if (r < 0)
+ return log_error_errno(r, "Failed to adjust native event source priority: %m");
+
+ return 0;
+}
diff --git a/src/grp-journal/libjournal-core/journald-native.h b/src/grp-journal/libjournal-core/journald-native.h
new file mode 100644
index 0000000000..c13b80aa4f
--- /dev/null
+++ b/src/grp-journal/libjournal-core/journald-native.h
@@ -0,0 +1,35 @@
+#pragma once
+
+/***
+ 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 Lesser General Public License as published by
+ the Free Software Foundation; either version 2.1 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
+ Lesser General Public License for more details.
+
+ You should have received a copy of the GNU Lesser General Public License
+ along with systemd; If not, see <http://www.gnu.org/licenses/>.
+***/
+
+#include "journald-server.h"
+
+/* Make sure not to make this smaller than the maximum coredump
+ * size. See COREDUMP_MAX in coredump.c */
+#define ENTRY_SIZE_MAX (1024*1024*770u)
+#define DATA_SIZE_MAX (1024*1024*768u)
+
+bool valid_user_field(const char *p, size_t l, bool allow_protected);
+
+void server_process_native_message(Server *s, const void *buffer, size_t buffer_size, const struct ucred *ucred, const struct timeval *tv, const char *label, size_t label_len);
+
+void server_process_native_file(Server *s, int fd, const struct ucred *ucred, const struct timeval *tv, const char *label, size_t label_len);
+
+int server_open_native_socket(Server*s);
diff --git a/src/grp-journal/libjournal-core/journald-rate-limit.c b/src/grp-journal/libjournal-core/journald-rate-limit.c
new file mode 100644
index 0000000000..22603bf8e6
--- /dev/null
+++ b/src/grp-journal/libjournal-core/journald-rate-limit.c
@@ -0,0 +1,272 @@
+/***
+ 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 Lesser General Public License as published by
+ the Free Software Foundation; either version 2.1 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
+ Lesser General Public License for more details.
+
+ You should have received a copy of the GNU Lesser General Public License
+ along with systemd; If not, see <http://www.gnu.org/licenses/>.
+***/
+
+#include <errno.h>
+#include <string.h>
+
+#include "basic/alloc-util.h"
+#include "basic/hashmap.h"
+#include "basic/list.h"
+#include "basic/random-util.h"
+#include "basic/string-util.h"
+#include "basic/util.h"
+
+#include "journald-rate-limit.h"
+
+#define POOLS_MAX 5
+#define BUCKETS_MAX 127
+#define GROUPS_MAX 2047
+
+static const int priority_map[] = {
+ [LOG_EMERG] = 0,
+ [LOG_ALERT] = 0,
+ [LOG_CRIT] = 0,
+ [LOG_ERR] = 1,
+ [LOG_WARNING] = 2,
+ [LOG_NOTICE] = 3,
+ [LOG_INFO] = 3,
+ [LOG_DEBUG] = 4
+};
+
+typedef struct JournalRateLimitPool JournalRateLimitPool;
+typedef struct JournalRateLimitGroup JournalRateLimitGroup;
+
+struct JournalRateLimitPool {
+ usec_t begin;
+ unsigned num;
+ unsigned suppressed;
+};
+
+struct JournalRateLimitGroup {
+ JournalRateLimit *parent;
+
+ char *id;
+ JournalRateLimitPool pools[POOLS_MAX];
+ uint64_t hash;
+
+ LIST_FIELDS(JournalRateLimitGroup, bucket);
+ LIST_FIELDS(JournalRateLimitGroup, lru);
+};
+
+struct JournalRateLimit {
+ usec_t interval;
+ unsigned burst;
+
+ JournalRateLimitGroup* buckets[BUCKETS_MAX];
+ JournalRateLimitGroup *lru, *lru_tail;
+
+ unsigned n_groups;
+
+ uint8_t hash_key[16];
+};
+
+JournalRateLimit *journal_rate_limit_new(usec_t interval, unsigned burst) {
+ JournalRateLimit *r;
+
+ assert(interval > 0 || burst == 0);
+
+ r = new0(JournalRateLimit, 1);
+ if (!r)
+ return NULL;
+
+ r->interval = interval;
+ r->burst = burst;
+
+ random_bytes(r->hash_key, sizeof(r->hash_key));
+
+ return r;
+}
+
+static void journal_rate_limit_group_free(JournalRateLimitGroup *g) {
+ assert(g);
+
+ if (g->parent) {
+ assert(g->parent->n_groups > 0);
+
+ if (g->parent->lru_tail == g)
+ g->parent->lru_tail = g->lru_prev;
+
+ LIST_REMOVE(lru, g->parent->lru, g);
+ LIST_REMOVE(bucket, g->parent->buckets[g->hash % BUCKETS_MAX], g);
+
+ g->parent->n_groups--;
+ }
+
+ free(g->id);
+ free(g);
+}
+
+void journal_rate_limit_free(JournalRateLimit *r) {
+ assert(r);
+
+ while (r->lru)
+ journal_rate_limit_group_free(r->lru);
+
+ free(r);
+}
+
+_pure_ static bool journal_rate_limit_group_expired(JournalRateLimitGroup *g, usec_t ts) {
+ unsigned i;
+
+ assert(g);
+
+ for (i = 0; i < POOLS_MAX; i++)
+ if (g->pools[i].begin + g->parent->interval >= ts)
+ return false;
+
+ return true;
+}
+
+static void journal_rate_limit_vacuum(JournalRateLimit *r, usec_t ts) {
+ assert(r);
+
+ /* Makes room for at least one new item, but drop all
+ * expored items too. */
+
+ while (r->n_groups >= GROUPS_MAX ||
+ (r->lru_tail && journal_rate_limit_group_expired(r->lru_tail, ts)))
+ journal_rate_limit_group_free(r->lru_tail);
+}
+
+static JournalRateLimitGroup* journal_rate_limit_group_new(JournalRateLimit *r, const char *id, usec_t ts) {
+ JournalRateLimitGroup *g;
+ struct siphash state;
+
+ assert(r);
+ assert(id);
+
+ g = new0(JournalRateLimitGroup, 1);
+ if (!g)
+ return NULL;
+
+ g->id = strdup(id);
+ if (!g->id)
+ goto fail;
+
+ siphash24_init(&state, r->hash_key);
+ string_hash_func(g->id, &state);
+ g->hash = siphash24_finalize(&state);
+
+ journal_rate_limit_vacuum(r, ts);
+
+ LIST_PREPEND(bucket, r->buckets[g->hash % BUCKETS_MAX], g);
+ LIST_PREPEND(lru, r->lru, g);
+ if (!g->lru_next)
+ r->lru_tail = g;
+ r->n_groups++;
+
+ g->parent = r;
+ return g;
+
+fail:
+ journal_rate_limit_group_free(g);
+ return NULL;
+}
+
+static unsigned burst_modulate(unsigned burst, uint64_t available) {
+ unsigned k;
+
+ /* Modulates the burst rate a bit with the amount of available
+ * disk space */
+
+ k = u64log2(available);
+
+ /* 1MB */
+ if (k <= 20)
+ return burst;
+
+ burst = (burst * (k-20)) / 4;
+
+ /*
+ * Example:
+ *
+ * <= 1MB = rate * 1
+ * 16MB = rate * 2
+ * 256MB = rate * 3
+ * 4GB = rate * 4
+ * 64GB = rate * 5
+ * 1TB = rate * 6
+ */
+
+ return burst;
+}
+
+int journal_rate_limit_test(JournalRateLimit *r, const char *id, int priority, uint64_t available) {
+ uint64_t h;
+ JournalRateLimitGroup *g;
+ JournalRateLimitPool *p;
+ struct siphash state;
+ unsigned burst;
+ usec_t ts;
+
+ assert(id);
+
+ if (!r)
+ return 1;
+
+ if (r->interval == 0 || r->burst == 0)
+ return 1;
+
+ burst = burst_modulate(r->burst, available);
+
+ ts = now(CLOCK_MONOTONIC);
+
+ siphash24_init(&state, r->hash_key);
+ string_hash_func(id, &state);
+ h = siphash24_finalize(&state);
+ g = r->buckets[h % BUCKETS_MAX];
+
+ LIST_FOREACH(bucket, g, g)
+ if (streq(g->id, id))
+ break;
+
+ if (!g) {
+ g = journal_rate_limit_group_new(r, id, ts);
+ if (!g)
+ return -ENOMEM;
+ }
+
+ p = &g->pools[priority_map[priority]];
+
+ if (p->begin <= 0) {
+ p->suppressed = 0;
+ p->num = 1;
+ p->begin = ts;
+ return 1;
+ }
+
+ if (p->begin + r->interval < ts) {
+ unsigned s;
+
+ s = p->suppressed;
+ p->suppressed = 0;
+ p->num = 1;
+ p->begin = ts;
+
+ return 1 + s;
+ }
+
+ if (p->num <= burst) {
+ p->num++;
+ return 1;
+ }
+
+ p->suppressed++;
+ return 0;
+}
diff --git a/src/grp-journal/libjournal-core/journald-rate-limit.h b/src/grp-journal/libjournal-core/journald-rate-limit.h
new file mode 100644
index 0000000000..533dd0f013
--- /dev/null
+++ b/src/grp-journal/libjournal-core/journald-rate-limit.h
@@ -0,0 +1,28 @@
+#pragma once
+
+/***
+ 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 Lesser General Public License as published by
+ the Free Software Foundation; either version 2.1 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
+ Lesser General Public License for more details.
+
+ You should have received a copy of the GNU Lesser General Public License
+ along with systemd; If not, see <http://www.gnu.org/licenses/>.
+***/
+
+#include "basic/util.h"
+
+typedef struct JournalRateLimit JournalRateLimit;
+
+JournalRateLimit *journal_rate_limit_new(usec_t interval, unsigned burst);
+void journal_rate_limit_free(JournalRateLimit *r);
+int journal_rate_limit_test(JournalRateLimit *r, const char *id, int priority, uint64_t available);
diff --git a/src/grp-journal/libjournal-core/journald-server.c b/src/grp-journal/libjournal-core/journald-server.c
new file mode 100644
index 0000000000..8f82d2a838
--- /dev/null
+++ b/src/grp-journal/libjournal-core/journald-server.c
@@ -0,0 +1,2007 @@
+/***
+ 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 Lesser General Public License as published by
+ the Free Software Foundation; either version 2.1 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
+ Lesser General Public License for more details.
+
+ You should have received a copy of the GNU Lesser General Public License
+ along with systemd; If not, see <http://www.gnu.org/licenses/>.
+***/
+
+#ifdef HAVE_SELINUX
+#include <selinux/selinux.h>
+#endif
+#include <sys/ioctl.h>
+#include <sys/mman.h>
+#include <sys/signalfd.h>
+#include <sys/statvfs.h>
+#include <linux/sockios.h>
+
+#include "libudev.h"
+#include "sd-daemon.h"
+#include "sd-journal.h"
+#include "sd-messages.h"
+
+#include "acl-util.h"
+#include "alloc-util.h"
+#include "audit-util.h"
+#include "cgroup-util.h"
+#include "conf-parser.h"
+#include "dirent-util.h"
+#include "extract-word.h"
+#include "fd-util.h"
+#include "fileio.h"
+#include "formats-util.h"
+#include "fs-util.h"
+#include "hashmap.h"
+#include "hostname-util.h"
+#include "io-util.h"
+#include "journal-authenticate.h"
+#include "journal-file.h"
+#include "journal-internal.h"
+#include "journal-vacuum.h"
+#include "journald-audit.h"
+#include "journald-kmsg.h"
+#include "journald-native.h"
+#include "journald-rate-limit.h"
+#include "journald-server.h"
+#include "journald-stream.h"
+#include "journald-syslog.h"
+#include "missing.h"
+#include "mkdir.h"
+#include "parse-util.h"
+#include "proc-cmdline.h"
+#include "process-util.h"
+#include "rm-rf.h"
+#include "selinux-util.h"
+#include "signal-util.h"
+#include "socket-util.h"
+#include "stdio-util.h"
+#include "string-table.h"
+#include "string-util.h"
+#include "user-util.h"
+#include "log.h"
+
+#define USER_JOURNALS_MAX 1024
+
+#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_SPACE_USEC (30*USEC_PER_SEC)
+
+#define NOTIFY_SNDBUF_SIZE (8*1024*1024)
+
+/* The period to insert between posting changes for coalescing */
+#define POST_CHANGE_TIMER_INTERVAL_USEC (250*USEC_PER_MSEC)
+
+static int determine_space_for(
+ Server *s,
+ JournalMetrics *metrics,
+ const char *path,
+ const char *name,
+ bool verbose,
+ bool patch_min_use,
+ uint64_t *available,
+ uint64_t *limit) {
+
+ uint64_t sum = 0, ss_avail, avail;
+ _cleanup_closedir_ DIR *d = NULL;
+ struct dirent *de;
+ struct statvfs ss;
+ const char *p;
+ usec_t ts;
+
+ assert(s);
+ assert(metrics);
+ assert(path);
+ assert(name);
+
+ ts = now(CLOCK_MONOTONIC);
+
+ if (!verbose && s->cached_space_timestamp + RECHECK_SPACE_USEC > ts) {
+
+ if (available)
+ *available = s->cached_space_available;
+ if (limit)
+ *limit = s->cached_space_limit;
+
+ return 0;
+ }
+
+ p = strjoina(path, SERVER_MACHINE_ID(s));
+ d = opendir(p);
+ if (!d)
+ return log_full_errno(errno == ENOENT ? LOG_DEBUG : LOG_ERR, errno, "Failed to open %s: %m", p);
+
+ if (fstatvfs(dirfd(d), &ss) < 0)
+ return log_error_errno(errno, "Failed to fstatvfs(%s): %m", p);
+
+ FOREACH_DIRENT_ALL(de, d, break) {
+ struct stat st;
+
+ if (!endswith(de->d_name, ".journal") &&
+ !endswith(de->d_name, ".journal~"))
+ continue;
+
+ if (fstatat(dirfd(d), de->d_name, &st, AT_SYMLINK_NOFOLLOW) < 0) {
+ log_debug_errno(errno, "Failed to stat %s/%s, ignoring: %m", p, de->d_name);
+ continue;
+ }
+
+ if (!S_ISREG(st.st_mode))
+ continue;
+
+ sum += (uint64_t) st.st_blocks * 512UL;
+ }
+
+ /* If requested, then let's bump the min_use limit to the
+ * current usage on disk. We do this when starting up and
+ * first opening the journal files. This way sudden spikes in
+ * disk usage will not cause journald to vacuum files without
+ * bounds. Note that this means that only a restart of
+ * journald will make it reset this value. */
+
+ if (patch_min_use)
+ metrics->min_use = MAX(metrics->min_use, sum);
+
+ ss_avail = ss.f_bsize * ss.f_bavail;
+ avail = LESS_BY(ss_avail, metrics->keep_free);
+
+ s->cached_space_limit = MIN(MAX(sum + avail, metrics->min_use), metrics->max_use);
+ s->cached_space_available = LESS_BY(s->cached_space_limit, sum);
+ s->cached_space_timestamp = ts;
+
+ if (verbose) {
+ char fb1[FORMAT_BYTES_MAX], fb2[FORMAT_BYTES_MAX], fb3[FORMAT_BYTES_MAX],
+ fb4[FORMAT_BYTES_MAX], fb5[FORMAT_BYTES_MAX], fb6[FORMAT_BYTES_MAX];
+ format_bytes(fb1, sizeof(fb1), sum);
+ format_bytes(fb2, sizeof(fb2), metrics->max_use);
+ format_bytes(fb3, sizeof(fb3), metrics->keep_free);
+ format_bytes(fb4, sizeof(fb4), ss_avail);
+ format_bytes(fb5, sizeof(fb5), s->cached_space_limit);
+ format_bytes(fb6, sizeof(fb6), s->cached_space_available);
+
+ server_driver_message(s, SD_MESSAGE_JOURNAL_USAGE,
+ LOG_MESSAGE("%s (%s) is %s, max %s, %s free.",
+ name, path, fb1, fb5, fb6),
+ "JOURNAL_NAME=%s", name,
+ "JOURNAL_PATH=%s", path,
+ "CURRENT_USE=%"PRIu64, sum,
+ "CURRENT_USE_PRETTY=%s", fb1,
+ "MAX_USE=%"PRIu64, metrics->max_use,
+ "MAX_USE_PRETTY=%s", fb2,
+ "DISK_KEEP_FREE=%"PRIu64, metrics->keep_free,
+ "DISK_KEEP_FREE_PRETTY=%s", fb3,
+ "DISK_AVAILABLE=%"PRIu64, ss_avail,
+ "DISK_AVAILABLE_PRETTY=%s", fb4,
+ "LIMIT=%"PRIu64, s->cached_space_limit,
+ "LIMIT_PRETTY=%s", fb5,
+ "AVAILABLE=%"PRIu64, s->cached_space_available,
+ "AVAILABLE_PRETTY=%s", fb6,
+ NULL);
+ }
+
+ if (available)
+ *available = s->cached_space_available;
+ if (limit)
+ *limit = s->cached_space_limit;
+
+ return 1;
+}
+
+static int determine_space(Server *s, bool verbose, bool patch_min_use, uint64_t *available, uint64_t *limit) {
+ JournalMetrics *metrics;
+ const char *path, *name;
+
+ assert(s);
+
+ if (s->system_journal) {
+ path = "/var/log/journal/";
+ metrics = &s->system_metrics;
+ name = "System journal";
+ } else {
+ path = "/run/log/journal/";
+ metrics = &s->runtime_metrics;
+ name = "Runtime journal";
+ }
+
+ return determine_space_for(s, metrics, path, name, verbose, patch_min_use, available, limit);
+}
+
+static void server_add_acls(JournalFile *f, uid_t uid) {
+#ifdef HAVE_ACL
+ int r;
+#endif
+ assert(f);
+
+#ifdef HAVE_ACL
+ if (uid <= SYSTEM_UID_MAX)
+ return;
+
+ r = add_acls_for_user(f->fd, uid);
+ if (r < 0)
+ log_warning_errno(r, "Failed to set ACL on %s, ignoring: %m", f->path);
+#endif
+}
+
+static int open_journal(
+ Server *s,
+ bool reliably,
+ const char *fname,
+ int flags,
+ bool seal,
+ JournalMetrics *metrics,
+ JournalFile **ret) {
+ int r;
+ JournalFile *f;
+
+ assert(s);
+ assert(fname);
+ assert(ret);
+
+ if (reliably)
+ r = journal_file_open_reliably(fname, flags, 0640, s->compress, seal, metrics, s->mmap, s->deferred_closes, NULL, &f);
+ else
+ r = journal_file_open(-1, fname, flags, 0640, s->compress, seal, metrics, s->mmap, s->deferred_closes, NULL, &f);
+ if (r < 0)
+ return r;
+
+ r = journal_file_enable_post_change_timer(f, s->event, POST_CHANGE_TIMER_INTERVAL_USEC);
+ if (r < 0) {
+ (void) journal_file_close(f);
+ return r;
+ }
+
+ *ret = f;
+ return r;
+}
+
+static JournalFile* find_journal(Server *s, uid_t uid) {
+ _cleanup_free_ char *p = NULL;
+ int r;
+ JournalFile *f;
+ sd_id128_t machine;
+
+ assert(s);
+
+ /* We split up user logs only on /var, not on /run. If the
+ * runtime file is open, we write to it exclusively, in order
+ * to guarantee proper order as soon as we flush /run to
+ * /var and close the runtime file. */
+
+ if (s->runtime_journal)
+ return s->runtime_journal;
+
+ if (uid <= SYSTEM_UID_MAX)
+ return s->system_journal;
+
+ r = sd_id128_get_machine(&machine);
+ if (r < 0)
+ return s->system_journal;
+
+ f = ordered_hashmap_get(s->user_journals, UID_TO_PTR(uid));
+ if (f)
+ return f;
+
+ 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 (ordered_hashmap_size(s->user_journals) >= USER_JOURNALS_MAX) {
+ /* Too many open? Then let's close one */
+ f = ordered_hashmap_steal_first(s->user_journals);
+ assert(f);
+ (void) journal_file_close(f);
+ }
+
+ r = open_journal(s, true, p, O_RDWR|O_CREAT, s->seal, &s->system_metrics, &f);
+ if (r < 0)
+ return s->system_journal;
+
+ server_add_acls(f, uid);
+
+ r = ordered_hashmap_put(s->user_journals, UID_TO_PTR(uid), f);
+ if (r < 0) {
+ (void) journal_file_close(f);
+ return s->system_journal;
+ }
+
+ 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, s->deferred_closes);
+ 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_add_acls(*f, uid);
+
+ return r;
+}
+
+void server_rotate(Server *s) {
+ JournalFile *f;
+ void *k;
+ Iterator i;
+ int r;
+
+ log_debug("Rotating...");
+
+ (void) do_rotate(s, &s->runtime_journal, "runtime", false, 0);
+ (void) do_rotate(s, &s->system_journal, "system", s->seal, 0);
+
+ ORDERED_HASHMAP_FOREACH_KEY(f, k, s->user_journals, i) {
+ r = do_rotate(s, &f, "user", s->seal, PTR_TO_UID(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);
+ }
+
+ /* Perform any deferred closes which aren't still offlining. */
+ SET_FOREACH(f, s->deferred_closes, i)
+ if (!journal_file_is_offlining(f)) {
+ (void) set_remove(s->deferred_closes, f);
+ (void) journal_file_close(f);
+ }
+}
+
+void server_sync(Server *s) {
+ JournalFile *f;
+ Iterator i;
+ int r;
+
+ if (s->system_journal) {
+ r = journal_file_set_offline(s->system_journal, false);
+ if (r < 0)
+ log_warning_errno(r, "Failed to sync system journal, ignoring: %m");
+ }
+
+ ORDERED_HASHMAP_FOREACH(f, s->user_journals, i) {
+ r = journal_file_set_offline(f, false);
+ if (r < 0)
+ log_warning_errno(r, "Failed to sync user journal, ignoring: %m");
+ }
+
+ if (s->sync_event_source) {
+ r = sd_event_source_set_enabled(s->sync_event_source, SD_EVENT_OFF);
+ if (r < 0)
+ log_error_errno(r, "Failed to disable sync timer source: %m");
+ }
+
+ s->sync_scheduled = false;
+}
+
+static void do_vacuum(
+ Server *s,
+ JournalFile *f,
+ JournalMetrics *metrics,
+ const char *path,
+ const char *name,
+ bool verbose,
+ bool patch_min_use) {
+
+ const char *p;
+ uint64_t limit;
+ int r;
+
+ assert(s);
+ assert(metrics);
+ assert(path);
+ assert(name);
+
+ if (!f)
+ return;
+
+ p = strjoina(path, SERVER_MACHINE_ID(s));
+
+ limit = metrics->max_use;
+ (void) determine_space_for(s, metrics, path, name, verbose, patch_min_use, NULL, &limit);
+
+ r = journal_directory_vacuum(p, limit, metrics->n_max_files, s->max_retention_usec, &s->oldest_file_usec, verbose);
+ if (r < 0 && r != -ENOENT)
+ log_warning_errno(r, "Failed to vacuum %s, ignoring: %m", p);
+}
+
+int server_vacuum(Server *s, bool verbose, bool patch_min_use) {
+ assert(s);
+
+ log_debug("Vacuuming...");
+
+ s->oldest_file_usec = 0;
+
+ do_vacuum(s, s->system_journal, &s->system_metrics, "/var/log/journal/", "System journal", verbose, patch_min_use);
+ do_vacuum(s, s->runtime_journal, &s->runtime_metrics, "/run/log/journal/", "Runtime journal", verbose, patch_min_use);
+
+ s->cached_space_limit = 0;
+ s->cached_space_available = 0;
+ s->cached_space_timestamp = 0;
+
+ return 0;
+}
+
+static void server_cache_machine_id(Server *s) {
+ sd_id128_t id;
+ int r;
+
+ assert(s);
+
+ r = sd_id128_get_machine(&id);
+ if (r < 0)
+ return;
+
+ sd_id128_to_string(id, stpcpy(s->machine_id_field, "_MACHINE_ID="));
+}
+
+static void server_cache_boot_id(Server *s) {
+ sd_id128_t id;
+ int r;
+
+ assert(s);
+
+ r = sd_id128_get_boot(&id);
+ if (r < 0)
+ return;
+
+ sd_id128_to_string(id, stpcpy(s->boot_id_field, "_BOOT_ID="));
+}
+
+static void server_cache_hostname(Server *s) {
+ _cleanup_free_ char *t = NULL;
+ char *x;
+
+ assert(s);
+
+ t = gethostname_malloc();
+ if (!t)
+ return;
+
+ x = strappend("_HOSTNAME=", t);
+ if (!x)
+ return;
+
+ free(s->hostname_field);
+ s->hostname_field = x;
+}
+
+static bool shall_try_append_again(JournalFile *f, int r) {
+ switch(r) {
+ case -E2BIG: /* Hit configured limit */
+ case -EFBIG: /* Hit fs limit */
+ case -EDQUOT: /* Quota limit hit */
+ case -ENOSPC: /* Disk full */
+ log_debug("%s: Allocation limit reached, rotating.", f->path);
+ return true;
+ case -EIO: /* I/O error of some kind (mmap) */
+ log_warning("%s: IO error, rotating.", f->path);
+ return true;
+ case -EHOSTDOWN: /* Other machine */
+ log_info("%s: Journal file from other machine, rotating.", f->path);
+ return true;
+ case -EBUSY: /* Unclean shutdown */
+ log_info("%s: Unclean shutdown, rotating.", f->path);
+ return true;
+ case -EPROTONOSUPPORT: /* Unsupported feature */
+ log_info("%s: Unsupported feature, rotating.", f->path);
+ return true;
+ case -EBADMSG: /* Corrupted */
+ case -ENODATA: /* Truncated */
+ case -ESHUTDOWN: /* Already archived */
+ log_warning("%s: Journal file corrupted, rotating.", f->path);
+ return true;
+ case -EIDRM: /* Journal file has been deleted */
+ log_warning("%s: Journal file has been deleted, rotating.", f->path);
+ return true;
+ default:
+ return false;
+ }
+}
+
+static void write_to_journal(Server *s, uid_t uid, struct iovec *iovec, unsigned n, int priority) {
+ JournalFile *f;
+ bool vacuumed = false;
+ int r;
+
+ assert(s);
+ assert(iovec);
+ assert(n > 0);
+
+ f = find_journal(s, uid);
+ if (!f)
+ return;
+
+ if (journal_file_rotate_suggested(f, s->max_file_usec)) {
+ log_debug("%s: Journal header limits reached or header out-of-date, rotating.", f->path);
+ server_rotate(s);
+ server_vacuum(s, false, false);
+ vacuumed = true;
+
+ f = find_journal(s, uid);
+ if (!f)
+ return;
+ }
+
+ r = journal_file_append_entry(f, NULL, iovec, n, &s->seqnum, NULL, NULL);
+ if (r >= 0) {
+ server_schedule_sync(s, priority);
+ return;
+ }
+
+ if (vacuumed || !shall_try_append_again(f, r)) {
+ log_error_errno(r, "Failed to write entry (%d items, %zu bytes), ignoring: %m", n, IOVEC_TOTAL_SIZE(iovec, n));
+ return;
+ }
+
+ server_rotate(s);
+ server_vacuum(s, false, false);
+
+ f = find_journal(s, uid);
+ if (!f)
+ return;
+
+ log_debug("Retrying write.");
+ r = journal_file_append_entry(f, NULL, iovec, n, &s->seqnum, NULL, NULL);
+ if (r < 0)
+ log_error_errno(r, "Failed to write entry (%d items, %zu bytes) despite vacuuming, ignoring: %m", n, IOVEC_TOTAL_SIZE(iovec, n));
+ else
+ server_schedule_sync(s, priority);
+}
+
+static void dispatch_message_real(
+ Server *s,
+ struct iovec *iovec, unsigned n, unsigned m,
+ const struct ucred *ucred,
+ const 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)],
+ uid[sizeof("_UID=") + DECIMAL_STR_MAX(uid_t)],
+ gid[sizeof("_GID=") + DECIMAL_STR_MAX(gid_t)],
+ owner_uid[sizeof("_SYSTEMD_OWNER_UID=") + DECIMAL_STR_MAX(uid_t)],
+ source_time[sizeof("_SOURCE_REALTIME_TIMESTAMP=") + DECIMAL_STR_MAX(usec_t)],
+ o_uid[sizeof("OBJECT_UID=") + DECIMAL_STR_MAX(uid_t)],
+ o_gid[sizeof("OBJECT_GID=") + DECIMAL_STR_MAX(gid_t)],
+ o_owner_uid[sizeof("OBJECT_SYSTEMD_OWNER_UID=") + DECIMAL_STR_MAX(uid_t)];
+ uid_t object_uid;
+ gid_t object_gid;
+ char *x;
+ int r;
+ char *t, *c;
+ uid_t realuid = 0, owner = 0, journal_uid;
+ bool owner_valid = false;
+#ifdef HAVE_AUDIT
+ char audit_session[sizeof("_AUDIT_SESSION=") + DECIMAL_STR_MAX(uint32_t)],
+ audit_loginuid[sizeof("_AUDIT_LOGINUID=") + DECIMAL_STR_MAX(uid_t)],
+ o_audit_session[sizeof("OBJECT_AUDIT_SESSION=") + DECIMAL_STR_MAX(uint32_t)],
+ o_audit_loginuid[sizeof("OBJECT_AUDIT_LOGINUID=") + DECIMAL_STR_MAX(uid_t)];
+
+ uint32_t audit;
+ uid_t loginuid;
+#endif
+
+ assert(s);
+ assert(iovec);
+ assert(n > 0);
+ assert(n + N_IOVEC_META_FIELDS + (object_pid ? N_IOVEC_OBJECT_FIELDS : 0) <= m);
+
+ if (ucred) {
+ realuid = ucred->uid;
+
+ sprintf(pid, "_PID="PID_FMT, ucred->pid);
+ IOVEC_SET_STRING(iovec[n++], pid);
+
+ sprintf(uid, "_UID="UID_FMT, ucred->uid);
+ IOVEC_SET_STRING(iovec[n++], uid);
+
+ sprintf(gid, "_GID="GID_FMT, ucred->gid);
+ IOVEC_SET_STRING(iovec[n++], gid);
+
+ r = get_process_comm(ucred->pid, &t);
+ if (r >= 0) {
+ x = strjoina("_COMM=", t);
+ free(t);
+ IOVEC_SET_STRING(iovec[n++], x);
+ }
+
+ r = get_process_exe(ucred->pid, &t);
+ if (r >= 0) {
+ x = strjoina("_EXE=", t);
+ free(t);
+ IOVEC_SET_STRING(iovec[n++], x);
+ }
+
+ r = get_process_cmdline(ucred->pid, 0, false, &t);
+ if (r >= 0) {
+ x = strjoina("_CMDLINE=", t);
+ free(t);
+ IOVEC_SET_STRING(iovec[n++], x);
+ }
+
+ r = get_process_capeff(ucred->pid, &t);
+ if (r >= 0) {
+ x = strjoina("_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) {
+ 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="UID_FMT, loginuid);
+ IOVEC_SET_STRING(iovec[n++], audit_loginuid);
+ }
+#endif
+
+ r = cg_pid_get_path_shifted(ucred->pid, s->cgroup_root, &c);
+ if (r >= 0) {
+ char *session = NULL;
+
+ x = strjoina("_SYSTEMD_CGROUP=", c);
+ IOVEC_SET_STRING(iovec[n++], x);
+
+ r = cg_path_get_session(c, &t);
+ if (r >= 0) {
+ session = strjoina("_SYSTEMD_SESSION=", t);
+ free(t);
+ IOVEC_SET_STRING(iovec[n++], session);
+ }
+
+ if (cg_path_get_owner_uid(c, &owner) >= 0) {
+ owner_valid = true;
+
+ sprintf(owner_uid, "_SYSTEMD_OWNER_UID="UID_FMT, owner);
+ IOVEC_SET_STRING(iovec[n++], owner_uid);
+ }
+
+ if (cg_path_get_unit(c, &t) >= 0) {
+ x = strjoina("_SYSTEMD_UNIT=", t);
+ free(t);
+ IOVEC_SET_STRING(iovec[n++], x);
+ } else if (unit_id && !session) {
+ x = strjoina("_SYSTEMD_UNIT=", unit_id);
+ IOVEC_SET_STRING(iovec[n++], x);
+ }
+
+ if (cg_path_get_user_unit(c, &t) >= 0) {
+ x = strjoina("_SYSTEMD_USER_UNIT=", t);
+ free(t);
+ IOVEC_SET_STRING(iovec[n++], x);
+ } else if (unit_id && session) {
+ x = strjoina("_SYSTEMD_USER_UNIT=", unit_id);
+ IOVEC_SET_STRING(iovec[n++], x);
+ }
+
+ if (cg_path_get_slice(c, &t) >= 0) {
+ x = strjoina("_SYSTEMD_SLICE=", t);
+ free(t);
+ IOVEC_SET_STRING(iovec[n++], x);
+ }
+
+ free(c);
+ } else if (unit_id) {
+ x = strjoina("_SYSTEMD_UNIT=", unit_id);
+ IOVEC_SET_STRING(iovec[n++], x);
+ }
+
+#ifdef HAVE_SELINUX
+ if (mac_selinux_have()) {
+ if (label) {
+ x = alloca(strlen("_SELINUX_CONTEXT=") + label_len + 1);
+
+ *((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 = strjoina("_SELINUX_CONTEXT=", con);
+
+ freecon(con);
+ IOVEC_SET_STRING(iovec[n++], x);
+ }
+ }
+ }
+#endif
+ }
+ assert(n <= m);
+
+ if (object_pid) {
+ r = get_process_uid(object_pid, &object_uid);
+ if (r >= 0) {
+ 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="GID_FMT, object_gid);
+ IOVEC_SET_STRING(iovec[n++], o_gid);
+ }
+
+ r = get_process_comm(object_pid, &t);
+ if (r >= 0) {
+ x = strjoina("OBJECT_COMM=", t);
+ free(t);
+ IOVEC_SET_STRING(iovec[n++], x);
+ }
+
+ r = get_process_exe(object_pid, &t);
+ if (r >= 0) {
+ x = strjoina("OBJECT_EXE=", t);
+ free(t);
+ IOVEC_SET_STRING(iovec[n++], x);
+ }
+
+ r = get_process_cmdline(object_pid, 0, false, &t);
+ if (r >= 0) {
+ x = strjoina("OBJECT_CMDLINE=", t);
+ free(t);
+ IOVEC_SET_STRING(iovec[n++], x);
+ }
+
+#ifdef HAVE_AUDIT
+ r = audit_session_from_pid(object_pid, &audit);
+ if (r >= 0) {
+ 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="UID_FMT, loginuid);
+ IOVEC_SET_STRING(iovec[n++], o_audit_loginuid);
+ }
+#endif
+
+ r = cg_pid_get_path_shifted(object_pid, s->cgroup_root, &c);
+ if (r >= 0) {
+ x = strjoina("OBJECT_SYSTEMD_CGROUP=", c);
+ IOVEC_SET_STRING(iovec[n++], x);
+
+ r = cg_path_get_session(c, &t);
+ if (r >= 0) {
+ x = strjoina("OBJECT_SYSTEMD_SESSION=", t);
+ free(t);
+ IOVEC_SET_STRING(iovec[n++], x);
+ }
+
+ if (cg_path_get_owner_uid(c, &owner) >= 0) {
+ sprintf(o_owner_uid, "OBJECT_SYSTEMD_OWNER_UID="UID_FMT, owner);
+ IOVEC_SET_STRING(iovec[n++], o_owner_uid);
+ }
+
+ if (cg_path_get_unit(c, &t) >= 0) {
+ x = strjoina("OBJECT_SYSTEMD_UNIT=", t);
+ free(t);
+ IOVEC_SET_STRING(iovec[n++], x);
+ }
+
+ if (cg_path_get_user_unit(c, &t) >= 0) {
+ x = strjoina("OBJECT_SYSTEMD_USER_UNIT=", t);
+ free(t);
+ IOVEC_SET_STRING(iovec[n++], x);
+ }
+
+ free(c);
+ }
+ }
+ assert(n <= m);
+
+ if (tv) {
+ sprintf(source_time, "_SOURCE_REALTIME_TIMESTAMP=%llu", (unsigned long long) timeval_load(tv));
+ IOVEC_SET_STRING(iovec[n++], source_time);
+ }
+
+ /* Note that strictly speaking storing the boot id here is
+ * redundant since the entry includes this in-line
+ * anyway. However, we need this indexed, too. */
+ if (!isempty(s->boot_id_field))
+ IOVEC_SET_STRING(iovec[n++], s->boot_id_field);
+
+ if (!isempty(s->machine_id_field))
+ IOVEC_SET_STRING(iovec[n++], s->machine_id_field);
+
+ if (!isempty(s->hostname_field))
+ IOVEC_SET_STRING(iovec[n++], s->hostname_field);
+
+ assert(n <= m);
+
+ if (s->split_mode == SPLIT_UID && realuid > 0)
+ /* 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. 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;
+
+ write_to_journal(s, journal_uid, iovec, n, priority);
+}
+
+void server_driver_message(Server *s, sd_id128_t message_id, const char *format, ...) {
+ char mid[11 + 32 + 1];
+ struct iovec iovec[N_IOVEC_META_FIELDS + 5 + N_IOVEC_PAYLOAD_FIELDS];
+ unsigned n = 0, m;
+ int r;
+ va_list ap;
+ struct ucred ucred = {};
+
+ assert(s);
+ assert(format);
+
+ assert_cc(3 == LOG_FAC(LOG_DAEMON));
+ IOVEC_SET_STRING(iovec[n++], "SYSLOG_FACILITY=3");
+ IOVEC_SET_STRING(iovec[n++], "SYSLOG_IDENTIFIER=systemd-journald");
+
+ IOVEC_SET_STRING(iovec[n++], "_TRANSPORT=driver");
+ assert_cc(6 == LOG_INFO);
+ IOVEC_SET_STRING(iovec[n++], "PRIORITY=6");
+
+ if (!sd_id128_equal(message_id, SD_ID128_NULL)) {
+ snprintf(mid, sizeof(mid), LOG_MESSAGE_ID(message_id));
+ IOVEC_SET_STRING(iovec[n++], mid);
+ }
+
+ m = n;
+
+ va_start(ap, format);
+ r = log_format_iovec(iovec, ELEMENTSOF(iovec), &n, false, 0, format, ap);
+ /* Error handling below */
+ va_end(ap);
+
+ ucred.pid = getpid();
+ ucred.uid = getuid();
+ ucred.gid = getgid();
+
+ if (r >= 0)
+ dispatch_message_real(s, iovec, n, ELEMENTSOF(iovec), &ucred, NULL, NULL, 0, NULL, LOG_INFO, 0);
+
+ while (m < n)
+ free(iovec[m++].iov_base);
+
+ if (r < 0) {
+ /* We failed to format the message. Emit a warning instead. */
+ char buf[LINE_MAX];
+
+ xsprintf(buf, "MESSAGE=Entry printing failed: %s", strerror(-r));
+
+ n = 3;
+ IOVEC_SET_STRING(iovec[n++], "PRIORITY=4");
+ IOVEC_SET_STRING(iovec[n++], buf);
+ dispatch_message_real(s, iovec, n, ELEMENTSOF(iovec), &ucred, NULL, NULL, 0, NULL, LOG_INFO, 0);
+ }
+}
+
+void server_dispatch_message(
+ Server *s,
+ struct iovec *iovec, unsigned n, unsigned m,
+ const struct ucred *ucred,
+ const struct timeval *tv,
+ const char *label, size_t label_len,
+ const char *unit_id,
+ int priority,
+ pid_t object_pid) {
+
+ int rl, r;
+ _cleanup_free_ char *path = NULL;
+ uint64_t available = 0;
+ char *c;
+
+ assert(s);
+ assert(iovec || n == 0);
+
+ if (n == 0)
+ return;
+
+ 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;
+
+ r = cg_pid_get_path_shifted(ucred->pid, s->cgroup_root, &path);
+ if (r < 0)
+ goto finish;
+
+ /* example: /user/lennart/3/foobar
+ * /system/dbus.service/foobar
+ *
+ * So let's cut of everything past the third /, since that is
+ * where user directories start */
+
+ c = strchr(path, '/');
+ if (c) {
+ c = strchr(c+1, '/');
+ if (c) {
+ c = strchr(c+1, '/');
+ if (c)
+ *c = 0;
+ }
+ }
+
+ (void) determine_space(s, false, false, &available, NULL);
+ rl = journal_rate_limit_test(s->rate_limit, path, priority & LOG_PRIMASK, available);
+ if (rl == 0)
+ return;
+
+ /* Write a suppression message if we suppressed something */
+ if (rl > 1)
+ server_driver_message(s, SD_MESSAGE_JOURNAL_DROPPED,
+ LOG_MESSAGE("Suppressed %u messages from %s", rl - 1, path),
+ NULL);
+
+finish:
+ dispatch_message_real(s, iovec, n, m, ucred, tv, label, label_len, unit_id, priority, object_pid);
+}
+
+
+static int system_journal_open(Server *s, bool flush_requested) {
+ const char *fn;
+ int r = 0;
+
+ if (!s->system_journal &&
+ (s->storage == STORAGE_PERSISTENT || s->storage == STORAGE_AUTO) &&
+ (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.
+ *
+ * If in persistent mode: create /var/log/journal and
+ * the machine path */
+
+ if (s->storage == STORAGE_PERSISTENT)
+ (void) mkdir_p("/var/log/journal/", 0755);
+
+ fn = strjoina("/var/log/journal/", SERVER_MACHINE_ID(s));
+ (void) mkdir(fn, 0755);
+
+ fn = strjoina(fn, "/system.journal");
+ r = open_journal(s, true, fn, O_RDWR|O_CREAT, s->seal, &s->system_metrics, &s->system_journal);
+ if (r >= 0) {
+ server_add_acls(s->system_journal, 0);
+ (void) determine_space_for(s, &s->system_metrics, "/var/log/journal/", "System journal", true, true, NULL, NULL);
+ } else if (r < 0) {
+ if (r != -ENOENT && r != -EROFS)
+ log_warning_errno(r, "Failed to open system journal: %m");
+
+ r = 0;
+ }
+ }
+
+ if (!s->runtime_journal &&
+ (s->storage != STORAGE_NONE)) {
+
+ fn = strjoina("/run/log/journal/", SERVER_MACHINE_ID(s), "/system.journal");
+
+ if (s->system_journal) {
+
+ /* Try to open the runtime journal, but only
+ * if it already exists, so that we can flush
+ * it into the system journal */
+
+ r = open_journal(s, false, fn, O_RDWR, false, &s->runtime_metrics, &s->runtime_journal);
+ if (r < 0) {
+ if (r != -ENOENT)
+ log_warning_errno(r, "Failed to open runtime journal: %m");
+
+ r = 0;
+ }
+
+ } else {
+
+ /* OK, we really need the runtime journal, so create
+ * it if necessary. */
+
+ (void) mkdir("/run/log", 0755);
+ (void) mkdir("/run/log/journal", 0755);
+ (void) mkdir_parents(fn, 0750);
+
+ r = open_journal(s, true, fn, O_RDWR|O_CREAT, false, &s->runtime_metrics, &s->runtime_journal);
+ if (r < 0)
+ return log_error_errno(r, "Failed to open runtime journal: %m");
+ }
+
+ if (s->runtime_journal) {
+ server_add_acls(s->runtime_journal, 0);
+ (void) determine_space_for(s, &s->runtime_metrics, "/run/log/journal/", "Runtime journal", true, true, NULL, NULL);
+ }
+ }
+
+ return r;
+}
+
+int server_flush_to_var(Server *s) {
+ sd_id128_t machine;
+ sd_journal *j = NULL;
+ char ts[FORMAT_TIMESPAN_MAX];
+ usec_t start;
+ unsigned n = 0;
+ int r;
+
+ assert(s);
+
+ if (s->storage != STORAGE_AUTO &&
+ s->storage != STORAGE_PERSISTENT)
+ return 0;
+
+ if (!s->runtime_journal)
+ return 0;
+
+ (void) system_journal_open(s, true);
+
+ if (!s->system_journal)
+ return 0;
+
+ log_debug("Flushing to /var...");
+
+ start = now(CLOCK_MONOTONIC);
+
+ r = sd_id128_get_machine(&machine);
+ if (r < 0)
+ return r;
+
+ r = sd_journal_open(&j, SD_JOURNAL_RUNTIME_ONLY);
+ if (r < 0)
+ return log_error_errno(r, "Failed to read runtime journal: %m");
+
+ sd_journal_set_data_threshold(j, 0);
+
+ SD_JOURNAL_FOREACH(j) {
+ Object *o = NULL;
+ JournalFile *f;
+
+ f = j->current_file;
+ assert(f && f->current_offset > 0);
+
+ n++;
+
+ r = journal_file_move_to_object(f, OBJECT_ENTRY, f->current_offset, &o);
+ if (r < 0) {
+ log_error_errno(r, "Can't read entry: %m");
+ goto finish;
+ }
+
+ r = journal_file_copy_entry(f, s->system_journal, o, f->current_offset, NULL, NULL, NULL);
+ if (r >= 0)
+ continue;
+
+ if (!shall_try_append_again(s->system_journal, r)) {
+ log_error_errno(r, "Can't write entry: %m");
+ goto finish;
+ }
+
+ server_rotate(s);
+ server_vacuum(s, false, false);
+
+ if (!s->system_journal) {
+ log_notice("Didn't flush runtime journal since rotation of system journal wasn't successful.");
+ r = -EIO;
+ goto finish;
+ }
+
+ 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_errno(r, "Can't write entry: %m");
+ goto finish;
+ }
+ }
+
+ r = 0;
+
+finish:
+ journal_file_post_change(s->system_journal);
+
+ s->runtime_journal = journal_file_close(s->runtime_journal);
+
+ if (r >= 0)
+ (void) rm_rf("/run/log/journal", REMOVE_ROOT);
+
+ sd_journal_close(j);
+
+ server_driver_message(s, SD_ID128_NULL,
+ LOG_MESSAGE("Time spent on flushing to /var is %s for %u entries.",
+ format_timespan(ts, sizeof(ts), now(CLOCK_MONOTONIC) - start, 0),
+ n),
+ NULL);
+
+ return r;
+}
+
+int server_process_datagram(sd_event_source *es, int fd, uint32_t revents, void *userdata) {
+ Server *s = userdata;
+ struct ucred *ucred = NULL;
+ struct timeval *tv = NULL;
+ struct cmsghdr *cmsg;
+ char *label = NULL;
+ size_t label_len = 0, m;
+ struct iovec iovec;
+ ssize_t n;
+ int *fds = NULL, v = 0;
+ unsigned n_fds = 0;
+
+ 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. */
+ 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),
+ };
+
+ assert(s);
+ 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);
+ return -EIO;
+ }
+
+ /* 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);
+
+ /* 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 - 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;
+
+ return log_error_errno(errno, "recvmsg() failed: %m");
+ }
+
+ CMSG_FOREACH(cmsg, &msghdr) {
+
+ if (cmsg->cmsg_level == SOL_SOCKET &&
+ cmsg->cmsg_type == SCM_CREDENTIALS &&
+ cmsg->cmsg_len == CMSG_LEN(sizeof(struct ucred)))
+ ucred = (struct ucred*) CMSG_DATA(cmsg);
+ else if (cmsg->cmsg_level == SOL_SOCKET &&
+ cmsg->cmsg_type == SCM_SECURITY) {
+ 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)))
+ tv = (struct timeval*) CMSG_DATA(cmsg);
+ else if (cmsg->cmsg_level == SOL_SOCKET &&
+ cmsg->cmsg_type == SCM_RIGHTS) {
+ fds = (int*) CMSG_DATA(cmsg);
+ n_fds = (cmsg->cmsg_len - CMSG_LEN(0)) / sizeof(int);
+ }
+ }
+
+ /* And a trailing NUL, just in case */
+ s->buffer[n] = 0;
+
+ if (fd == s->syslog_fd) {
+ if (n > 0 && n_fds == 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.");
+
+ } 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) {
+ Server *s = userdata;
+ int r;
+
+ assert(s);
+
+ log_info("Received request to flush runtime journal from PID " PID_FMT, si->ssi_pid);
+
+ server_flush_to_var(s);
+ server_sync(s);
+ server_vacuum(s, false, false);
+
+ r = touch("/run/systemd/journal/flushed");
+ if (r < 0)
+ log_warning_errno(r, "Failed to touch /run/systemd/journal/flushed, ignoring: %m");
+
+ return 0;
+}
+
+static int dispatch_sigusr2(sd_event_source *es, const struct signalfd_siginfo *si, void *userdata) {
+ Server *s = userdata;
+ int r;
+
+ assert(s);
+
+ log_info("Received request to rotate journal from PID " PID_FMT, si->ssi_pid);
+ server_rotate(s);
+ server_vacuum(s, true, true);
+
+ /* Let clients know when the most recent rotation happened. */
+ r = write_timestamp_file_atomic("/run/systemd/journal/rotated", now(CLOCK_MONOTONIC));
+ if (r < 0)
+ log_warning_errno(r, "Failed to write /run/systemd/journal/rotated, ignoring: %m");
+
+ return 0;
+}
+
+static int dispatch_sigterm(sd_event_source *es, const struct signalfd_siginfo *si, void *userdata) {
+ Server *s = userdata;
+
+ assert(s);
+
+ log_received_signal(LOG_INFO, si);
+
+ sd_event_exit(s->event, 0);
+ return 0;
+}
+
+static int dispatch_sigrtmin1(sd_event_source *es, const struct signalfd_siginfo *si, void *userdata) {
+ Server *s = userdata;
+ int r;
+
+ assert(s);
+
+ log_debug("Received request to sync from PID " PID_FMT, si->ssi_pid);
+
+ server_sync(s);
+
+ /* Let clients know when the most recent sync happened. */
+ r = write_timestamp_file_atomic("/run/systemd/journal/synced", now(CLOCK_MONOTONIC));
+ if (r < 0)
+ log_warning_errno(r, "Failed to write /run/systemd/journal/synced, ignoring: %m");
+
+ return 0;
+}
+
+static int setup_signals(Server *s) {
+ int r;
+
+ assert(s);
+
+ assert(sigprocmask_many(SIG_SETMASK, NULL, SIGINT, SIGTERM, SIGUSR1, SIGUSR2, SIGRTMIN+1, -1) >= 0);
+
+ 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, &s->sigusr2_event_source, SIGUSR2, dispatch_sigusr2, s);
+ if (r < 0)
+ return r;
+
+ r = sd_event_add_signal(s->event, &s->sigterm_event_source, SIGTERM, dispatch_sigterm, s);
+ if (r < 0)
+ return r;
+
+ /* Let's process SIGTERM late, so that we flush all queued
+ * messages to disk before we exit */
+ r = sd_event_source_set_priority(s->sigterm_event_source, SD_EVENT_PRIORITY_NORMAL+20);
+ if (r < 0)
+ return r;
+
+ /* When journald is invoked on the terminal (when debugging),
+ * it's useful if C-c is handled equivalent to SIGTERM. */
+ r = sd_event_add_signal(s->event, &s->sigint_event_source, SIGINT, dispatch_sigterm, s);
+ if (r < 0)
+ return r;
+
+ r = sd_event_source_set_priority(s->sigint_event_source, SD_EVENT_PRIORITY_NORMAL+20);
+ if (r < 0)
+ return r;
+
+ /* SIGRTMIN+1 causes an immediate sync. We process this very
+ * late, so that everything else queued at this point is
+ * really written to disk. Clients can watch
+ * /run/systemd/journal/synced with inotify until its mtime
+ * changes to see when a sync happened. */
+ r = sd_event_add_signal(s->event, &s->sigrtmin1_event_source, SIGRTMIN+1, dispatch_sigrtmin1, s);
+ if (r < 0)
+ return r;
+
+ r = sd_event_source_set_priority(s->sigrtmin1_event_source, SD_EVENT_PRIORITY_NORMAL+15);
+ if (r < 0)
+ return r;
+
+ return 0;
+}
+
+static int server_parse_proc_cmdline(Server *s) {
+ _cleanup_free_ char *line = NULL;
+ const char *p;
+ int r;
+
+ r = proc_cmdline(&line);
+ if (r < 0) {
+ log_warning_errno(r, "Failed to read /proc/cmdline, ignoring: %m");
+ return 0;
+ }
+
+ p = line;
+ for (;;) {
+ _cleanup_free_ char *word = NULL;
+
+ r = extract_first_word(&p, &word, NULL, 0);
+ if (r < 0)
+ return log_error_errno(r, "Failed to parse journald syntax \"%s\": %m", line);
+
+ if (r == 0)
+ break;
+
+ if (startswith(word, "systemd.journald.forward_to_syslog=")) {
+ r = parse_boolean(word + 35);
+ if (r < 0)
+ log_warning("Failed to parse forward to syslog switch %s. Ignoring.", word + 35);
+ else
+ s->forward_to_syslog = r;
+ } else if (startswith(word, "systemd.journald.forward_to_kmsg=")) {
+ r = parse_boolean(word + 33);
+ if (r < 0)
+ log_warning("Failed to parse forward to kmsg switch %s. Ignoring.", word + 33);
+ else
+ s->forward_to_kmsg = r;
+ } else if (startswith(word, "systemd.journald.forward_to_console=")) {
+ r = parse_boolean(word + 36);
+ if (r < 0)
+ 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) {
+ assert(s);
+
+ return config_parse_many(PKGSYSCONFDIR "/journald.conf",
+ CONF_PATHS_NULSTR("systemd/journald.conf.d"),
+ "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) {
+ Server *s = userdata;
+
+ assert(s);
+
+ server_sync(s);
+ return 0;
+}
+
+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;
+
+ if (s->sync_interval_usec > 0) {
+ usec_t 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_time(
+ s->event,
+ &s->sync_event_source,
+ CLOCK_MONOTONIC,
+ when, 0,
+ server_dispatch_sync, s);
+ if (r < 0)
+ return r;
+
+ r = sd_event_source_set_priority(s->sync_event_source, SD_EVENT_PRIORITY_IMPORTANT);
+ } else {
+ r = sd_event_source_set_time(s->sync_event_source, when);
+ if (r < 0)
+ return r;
+
+ r = sd_event_source_set_enabled(s->sync_event_source, SD_EVENT_ONESHOT);
+ }
+ if (r < 0)
+ return r;
+
+ s->sync_scheduled = true;
+ }
+
+ return 0;
+}
+
+static int dispatch_hostname_change(sd_event_source *es, int fd, uint32_t revents, void *userdata) {
+ Server *s = userdata;
+
+ assert(s);
+
+ server_cache_hostname(s);
+ return 0;
+}
+
+static int server_open_hostname(Server *s) {
+ int r;
+
+ assert(s);
+
+ s->hostname_fd = open("/proc/sys/kernel/hostname", O_RDONLY|O_CLOEXEC|O_NDELAY|O_NOCTTY);
+ 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_event_source, s->hostname_fd, 0, dispatch_hostname_change, s);
+ if (r < 0) {
+ /* kernels prior to 3.2 don't support polling this file. Ignore
+ * the failure. */
+ if (r == -EPERM) {
+ log_warning_errno(r, "Failed to register hostname fd in event loop, ignoring: %m");
+ 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)
+ return log_error_errno(r, "Failed to adjust priority of host name event source: %m");
+
+ return 0;
+}
+
+static int dispatch_notify_event(sd_event_source *es, int fd, uint32_t revents, void *userdata) {
+ Server *s = userdata;
+ int r;
+
+ assert(s);
+ assert(s->notify_event_source == es);
+ assert(s->notify_fd == fd);
+
+ /* The $NOTIFY_SOCKET is writable again, now send exactly one
+ * message on it. Either it's the wtachdog event, the initial
+ * READY=1 event or an stdout stream event. If there's nothing
+ * to write anymore, turn our event source off. The next time
+ * there's something to send it will be turned on again. */
+
+ if (!s->sent_notify_ready) {
+ static const char p[] =
+ "READY=1\n"
+ "STATUS=Processing requests...";
+ ssize_t l;
+
+ l = send(s->notify_fd, p, strlen(p), MSG_DONTWAIT);
+ if (l < 0) {
+ if (errno == EAGAIN)
+ return 0;
+
+ return log_error_errno(errno, "Failed to send READY=1 notification message: %m");
+ }
+
+ s->sent_notify_ready = true;
+ log_debug("Sent READY=1 notification.");
+
+ } else if (s->send_watchdog) {
+
+ static const char p[] =
+ "WATCHDOG=1";
+
+ ssize_t l;
+
+ l = send(s->notify_fd, p, strlen(p), MSG_DONTWAIT);
+ if (l < 0) {
+ if (errno == EAGAIN)
+ return 0;
+
+ return log_error_errno(errno, "Failed to send WATCHDOG=1 notification message: %m");
+ }
+
+ s->send_watchdog = false;
+ log_debug("Sent WATCHDOG=1 notification.");
+
+ } else if (s->stdout_streams_notify_queue)
+ /* Dispatch one stream notification event */
+ stdout_stream_send_notify(s->stdout_streams_notify_queue);
+
+ /* Leave us enabled if there's still more to to do. */
+ if (s->send_watchdog || s->stdout_streams_notify_queue)
+ return 0;
+
+ /* There was nothing to do anymore, let's turn ourselves off. */
+ r = sd_event_source_set_enabled(es, SD_EVENT_OFF);
+ if (r < 0)
+ return log_error_errno(r, "Failed to turn off notify event source: %m");
+
+ return 0;
+}
+
+static int dispatch_watchdog(sd_event_source *es, uint64_t usec, void *userdata) {
+ Server *s = userdata;
+ int r;
+
+ assert(s);
+
+ s->send_watchdog = true;
+
+ r = sd_event_source_set_enabled(s->notify_event_source, SD_EVENT_ON);
+ if (r < 0)
+ log_warning_errno(r, "Failed to turn on notify event source: %m");
+
+ r = sd_event_source_set_time(s->watchdog_event_source, usec + s->watchdog_usec / 2);
+ if (r < 0)
+ return log_error_errno(r, "Failed to restart watchdog event source: %m");
+
+ r = sd_event_source_set_enabled(s->watchdog_event_source, SD_EVENT_ON);
+ if (r < 0)
+ return log_error_errno(r, "Failed to enable watchdog event source: %m");
+
+ return 0;
+}
+
+static int server_connect_notify(Server *s) {
+ union sockaddr_union sa = {
+ .un.sun_family = AF_UNIX,
+ };
+ const char *e;
+ int r;
+
+ assert(s);
+ assert(s->notify_fd < 0);
+ assert(!s->notify_event_source);
+
+ /*
+ So here's the problem: we'd like to send notification
+ messages to PID 1, but we cannot do that via sd_notify(),
+ since that's synchronous, and we might end up blocking on
+ it. Specifically: given that PID 1 might block on
+ dbus-daemon during IPC, and dbus-daemon is logging to us,
+ and might hence block on us, we might end up in a deadlock
+ if we block on sending PID 1 notification messages — by
+ generating a full blocking circle. To avoid this, let's
+ create a non-blocking socket, and connect it to the
+ notification socket, and then wait for POLLOUT before we
+ send anything. This should efficiently avoid any deadlocks,
+ as we'll never block on PID 1, hence PID 1 can safely block
+ on dbus-daemon which can safely block on us again.
+
+ Don't think that this issue is real? It is, see:
+ https://github.com/systemd/systemd/issues/1505
+ */
+
+ e = getenv("NOTIFY_SOCKET");
+ if (!e)
+ return 0;
+
+ if ((e[0] != '@' && e[0] != '/') || e[1] == 0) {
+ log_error("NOTIFY_SOCKET set to an invalid value: %s", e);
+ return -EINVAL;
+ }
+
+ if (strlen(e) > sizeof(sa.un.sun_path)) {
+ log_error("NOTIFY_SOCKET path too long: %s", e);
+ return -EINVAL;
+ }
+
+ s->notify_fd = socket(AF_UNIX, SOCK_DGRAM|SOCK_CLOEXEC|SOCK_NONBLOCK, 0);
+ if (s->notify_fd < 0)
+ return log_error_errno(errno, "Failed to create notify socket: %m");
+
+ (void) fd_inc_sndbuf(s->notify_fd, NOTIFY_SNDBUF_SIZE);
+
+ strncpy(sa.un.sun_path, e, sizeof(sa.un.sun_path));
+ if (sa.un.sun_path[0] == '@')
+ sa.un.sun_path[0] = 0;
+
+ r = connect(s->notify_fd, &sa.sa, SOCKADDR_UN_LEN(sa.un));
+ if (r < 0)
+ return log_error_errno(errno, "Failed to connect to notify socket: %m");
+
+ r = sd_event_add_io(s->event, &s->notify_event_source, s->notify_fd, EPOLLOUT, dispatch_notify_event, s);
+ if (r < 0)
+ return log_error_errno(r, "Failed to watch notification socket: %m");
+
+ if (sd_watchdog_enabled(false, &s->watchdog_usec) > 0) {
+ s->send_watchdog = true;
+
+ r = sd_event_add_time(s->event, &s->watchdog_event_source, CLOCK_MONOTONIC, now(CLOCK_MONOTONIC) + s->watchdog_usec/2, s->watchdog_usec/4, dispatch_watchdog, s);
+ if (r < 0)
+ return log_error_errno(r, "Failed to add watchdog time event: %m");
+ }
+
+ /* This should fire pretty soon, which we'll use to send the
+ * READY=1 event. */
+
+ return 0;
+}
+
+int server_init(Server *s) {
+ _cleanup_fdset_free_ FDSet *fds = NULL;
+ int n, r, fd;
+ bool no_sockets;
+
+ assert(s);
+
+ zero(*s);
+ s->syslog_fd = s->native_fd = s->stdout_fd = s->dev_kmsg_fd = s->audit_fd = s->hostname_fd = s->notify_fd = -1;
+ s->compress = true;
+ s->seal = true;
+
+ s->watchdog_usec = USEC_INFINITY;
+
+ s->sync_interval_usec = DEFAULT_SYNC_INTERVAL_USEC;
+ s->sync_scheduled = false;
+
+ s->rate_limit_interval = DEFAULT_RATE_LIMIT_INTERVAL;
+ s->rate_limit_burst = DEFAULT_RATE_LIMIT_BURST;
+
+ 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;
+
+ journal_reset_metrics(&s->system_metrics);
+ journal_reset_metrics(&s->runtime_metrics);
+
+ 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 "USEC_FMT",%u to 0,0",
+ s->rate_limit_interval, s->rate_limit_burst);
+ s->rate_limit_interval = s->rate_limit_burst = 0;
+ }
+
+ (void) mkdir_p("/run/systemd/journal", 0755);
+
+ s->user_journals = ordered_hashmap_new(NULL);
+ if (!s->user_journals)
+ return log_oom();
+
+ s->mmap = mmap_cache_new();
+ if (!s->mmap)
+ return log_oom();
+
+ s->deferred_closes = set_new(NULL);
+ if (!s->deferred_closes)
+ return log_oom();
+
+ r = sd_event_default(&s->event);
+ if (r < 0)
+ return log_error_errno(r, "Failed to create event loop: %m");
+
+ n = sd_listen_fds(true);
+ 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++) {
+
+ if (sd_is_socket_unix(fd, SOCK_DGRAM, -1, "/run/systemd/journal/socket", 0) > 0) {
+
+ if (s->native_fd >= 0) {
+ log_error("Too many native sockets passed.");
+ return -EINVAL;
+ }
+
+ s->native_fd = fd;
+
+ } else if (sd_is_socket_unix(fd, SOCK_STREAM, 1, "/run/systemd/journal/stdout", 0) > 0) {
+
+ if (s->stdout_fd >= 0) {
+ log_error("Too many stdout sockets passed.");
+ return -EINVAL;
+ }
+
+ s->stdout_fd = fd;
+
+ } 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.");
+ return -EINVAL;
+ }
+
+ 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 {
+
+ if (!fds) {
+ fds = fdset_new();
+ if (!fds)
+ return log_oom();
+ }
+
+ r = fdset_put(fds, fd);
+ if (r < 0)
+ return log_oom();
+ }
+ }
+
+ /* Try to restore streams, but don't bother if this fails */
+ (void) server_restore_streams(s, fds);
+
+ if (fdset_size(fds) > 0) {
+ log_warning("%u unknown file descriptors passed, closing.", fdset_size(fds));
+ fds = fdset_free(fds);
+ }
+
+ no_sockets = s->native_fd < 0 && s->stdout_fd < 0 && s->syslog_fd < 0 && s->audit_fd < 0;
+
+ /* always open stdout, syslog, native, and kmsg sockets */
+
+ /* systemd-journald.socket: /run/systemd/journal/stdout */
+ r = server_open_stdout_socket(s);
+ if (r < 0)
+ return r;
+
+ /* systemd-journald-dev-log.socket: /run/systemd/journal/dev-log */
+ r = server_open_syslog_socket(s);
+ if (r < 0)
+ return r;
+
+ /* systemd-journald.socket: /run/systemd/journal/socket */
+ r = server_open_native_socket(s);
+ if (r < 0)
+ return r;
+
+ /* /dev/ksmg */
+ r = server_open_dev_kmsg(s);
+ if (r < 0)
+ return r;
+
+ /* Unless we got *some* sockets and not audit, open audit socket */
+ if (s->audit_fd >= 0 || no_sockets) {
+ r = server_open_audit(s);
+ if (r < 0)
+ return r;
+ }
+
+ r = server_open_kernel_seqnum(s);
+ if (r < 0)
+ return r;
+
+ r = server_open_hostname(s);
+ if (r < 0)
+ return r;
+
+ r = setup_signals(s);
+ if (r < 0)
+ return r;
+
+ s->udev = udev_new();
+ if (!s->udev)
+ return -ENOMEM;
+
+ s->rate_limit = journal_rate_limit_new(s->rate_limit_interval, s->rate_limit_burst);
+ if (!s->rate_limit)
+ return -ENOMEM;
+
+ r = cg_get_root_path(&s->cgroup_root);
+ if (r < 0)
+ return r;
+
+ server_cache_hostname(s);
+ server_cache_boot_id(s);
+ server_cache_machine_id(s);
+
+ (void) server_connect_notify(s);
+
+ return system_journal_open(s, false);
+}
+
+void server_maybe_append_tags(Server *s) {
+#ifdef HAVE_GCRYPT
+ JournalFile *f;
+ Iterator i;
+ usec_t n;
+
+ n = now(CLOCK_REALTIME);
+
+ if (s->system_journal)
+ journal_file_maybe_append_tag(s->system_journal, n);
+
+ ORDERED_HASHMAP_FOREACH(f, s->user_journals, i)
+ journal_file_maybe_append_tag(f, n);
+#endif
+}
+
+void server_done(Server *s) {
+ JournalFile *f;
+ assert(s);
+
+ if (s->deferred_closes) {
+ journal_file_close_set(s->deferred_closes);
+ set_free(s->deferred_closes);
+ }
+
+ while (s->stdout_streams)
+ stdout_stream_free(s->stdout_streams);
+
+ if (s->system_journal)
+ (void) journal_file_close(s->system_journal);
+
+ if (s->runtime_journal)
+ (void) journal_file_close(s->runtime_journal);
+
+ while ((f = ordered_hashmap_steal_first(s->user_journals)))
+ (void) journal_file_close(f);
+
+ 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);
+ sd_event_source_unref(s->sigterm_event_source);
+ sd_event_source_unref(s->sigint_event_source);
+ sd_event_source_unref(s->sigrtmin1_event_source);
+ sd_event_source_unref(s->hostname_event_source);
+ sd_event_source_unref(s->notify_event_source);
+ sd_event_source_unref(s->watchdog_event_source);
+ sd_event_unref(s->event);
+
+ 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);
+ safe_close(s->notify_fd);
+
+ if (s->rate_limit)
+ journal_rate_limit_free(s->rate_limit);
+
+ if (s->kernel_seqnum)
+ munmap(s->kernel_seqnum, sizeof(uint64_t));
+
+ free(s->buffer);
+ free(s->tty_path);
+ free(s->cgroup_root);
+ free(s->hostname_field);
+
+ if (s->mmap)
+ mmap_cache_unref(s->mmap);
+
+ udev_unref(s->udev);
+}
+
+static const char* const storage_table[_STORAGE_MAX] = {
+ [STORAGE_AUTO] = "auto",
+ [STORAGE_VOLATILE] = "volatile",
+ [STORAGE_PERSISTENT] = "persistent",
+ [STORAGE_NONE] = "none"
+};
+
+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_MAX] = {
+ [SPLIT_LOGIN] = "login",
+ [SPLIT_UID] = "uid",
+ [SPLIT_NONE] = "none",
+};
+
+DEFINE_STRING_TABLE_LOOKUP(split_mode, SplitMode);
+DEFINE_CONFIG_PARSE_ENUM(config_parse_split_mode, split_mode, SplitMode, "Failed to parse split mode setting");
diff --git a/src/grp-journal/libjournal-core/journald-server.h b/src/grp-journal/libjournal-core/journald-server.h
new file mode 100644
index 0000000000..5f1af1e50b
--- /dev/null
+++ b/src/grp-journal/libjournal-core/journald-server.h
@@ -0,0 +1,187 @@
+#pragma once
+
+/***
+ 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 Lesser General Public License as published by
+ the Free Software Foundation; either version 2.1 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
+ Lesser General Public License for more details.
+
+ You should have received a copy of the GNU Lesser General Public License
+ along with systemd; If not, see <http://www.gnu.org/licenses/>.
+***/
+
+#include <stdbool.h>
+#include <sys/types.h>
+
+#include <systemd/sd-event.h>
+
+#include "basic/hashmap.h"
+#include "basic/list.h"
+#include "sd-journal/journal-file.h"
+
+typedef struct Server Server;
+
+#include "journald-rate-limit.h"
+#include "journald-stream.h"
+
+typedef enum Storage {
+ STORAGE_AUTO,
+ STORAGE_VOLATILE,
+ STORAGE_PERSISTENT,
+ STORAGE_NONE,
+ _STORAGE_MAX,
+ _STORAGE_INVALID = -1
+} Storage;
+
+typedef enum SplitMode {
+ SPLIT_UID,
+ SPLIT_LOGIN,
+ SPLIT_NONE,
+ _SPLIT_MAX,
+ _SPLIT_INVALID = -1
+} SplitMode;
+
+struct Server {
+ int syslog_fd;
+ int native_fd;
+ int stdout_fd;
+ int dev_kmsg_fd;
+ int audit_fd;
+ int hostname_fd;
+ int notify_fd;
+
+ sd_event *event;
+
+ sd_event_source *syslog_event_source;
+ sd_event_source *native_event_source;
+ sd_event_source *stdout_event_source;
+ sd_event_source *dev_kmsg_event_source;
+ sd_event_source *audit_event_source;
+ sd_event_source *sync_event_source;
+ sd_event_source *sigusr1_event_source;
+ sd_event_source *sigusr2_event_source;
+ sd_event_source *sigterm_event_source;
+ sd_event_source *sigint_event_source;
+ sd_event_source *sigrtmin1_event_source;
+ sd_event_source *hostname_event_source;
+ sd_event_source *notify_event_source;
+ sd_event_source *watchdog_event_source;
+
+ JournalFile *runtime_journal;
+ JournalFile *system_journal;
+ OrderedHashmap *user_journals;
+
+ uint64_t seqnum;
+
+ char *buffer;
+ size_t buffer_size;
+
+ JournalRateLimit *rate_limit;
+ usec_t sync_interval_usec;
+ usec_t rate_limit_interval;
+ unsigned rate_limit_burst;
+
+ JournalMetrics runtime_metrics;
+ JournalMetrics system_metrics;
+
+ bool compress;
+ bool seal;
+
+ bool forward_to_kmsg;
+ bool forward_to_syslog;
+ bool forward_to_console;
+ bool forward_to_wall;
+
+ unsigned n_forward_syslog_missed;
+ usec_t last_warn_forward_syslog_missed;
+
+ uint64_t cached_space_available;
+ uint64_t cached_space_limit;
+ usec_t cached_space_timestamp;
+
+ uint64_t var_available_timestamp;
+
+ usec_t max_retention_usec;
+ usec_t max_file_usec;
+ usec_t oldest_file_usec;
+
+ LIST_HEAD(StdoutStream, stdout_streams);
+ LIST_HEAD(StdoutStream, stdout_streams_notify_queue);
+ unsigned n_stdout_streams;
+
+ char *tty_path;
+
+ int max_level_store;
+ int max_level_syslog;
+ int max_level_kmsg;
+ int max_level_console;
+ int max_level_wall;
+
+ Storage storage;
+ SplitMode split_mode;
+
+ MMapCache *mmap;
+
+ Set *deferred_closes;
+
+ struct udev *udev;
+
+ uint64_t *kernel_seqnum;
+ bool dev_kmsg_readable:1;
+
+ bool send_watchdog:1;
+ bool sent_notify_ready:1;
+ bool sync_scheduled:1;
+
+ char machine_id_field[sizeof("_MACHINE_ID=") + 32];
+ char boot_id_field[sizeof("_BOOT_ID=") + 32];
+ char *hostname_field;
+
+ /* Cached cgroup root, so that we don't have to query that all the time */
+ char *cgroup_root;
+
+ usec_t watchdog_usec;
+};
+
+#define SERVER_MACHINE_ID(s) ((s)->machine_id_field + strlen("_MACHINE_ID="))
+
+#define N_IOVEC_META_FIELDS 20
+#define N_IOVEC_KERNEL_FIELDS 64
+#define N_IOVEC_UDEV_FIELDS 32
+#define N_IOVEC_OBJECT_FIELDS 12
+#define N_IOVEC_PAYLOAD_FIELDS 15
+
+void server_dispatch_message(Server *s, struct iovec *iovec, unsigned n, unsigned m, const struct ucred *ucred, const struct timeval *tv, const char *label, size_t label_len, const char *unit_id, int priority, pid_t object_pid);
+void server_driver_message(Server *s, sd_id128_t message_id, const char *format, ...) _printf_(3,0) _sentinel_;
+
+/* gperf lookup function */
+const struct ConfigPerfItem* journald_gperf_lookup(const char *key, unsigned length);
+
+int config_parse_storage(const char *unit, const char *filename, unsigned line, const char *section, unsigned section_line, const char *lvalue, int ltype, const char *rvalue, void *data, void *userdata);
+
+const char *storage_to_string(Storage s) _const_;
+Storage storage_from_string(const char *s) _pure_;
+
+int config_parse_split_mode(const char *unit, const char *filename, unsigned line, const char *section, unsigned section_line, const char *lvalue, int ltype, const char *rvalue, void *data, void *userdata);
+
+const char *split_mode_to_string(SplitMode s) _const_;
+SplitMode split_mode_from_string(const char *s) _pure_;
+
+int server_init(Server *s);
+void server_done(Server *s);
+void server_sync(Server *s);
+int server_vacuum(Server *s, bool verbose, bool patch_min_use);
+void server_rotate(Server *s);
+int server_schedule_sync(Server *s, int priority);
+int server_flush_to_var(Server *s);
+void server_maybe_append_tags(Server *s);
+int server_process_datagram(sd_event_source *es, int fd, uint32_t revents, void *userdata);
diff --git a/src/grp-journal/libjournal-core/journald-stream.c b/src/grp-journal/libjournal-core/journald-stream.c
new file mode 100644
index 0000000000..4ad16ee41c
--- /dev/null
+++ b/src/grp-journal/libjournal-core/journald-stream.c
@@ -0,0 +1,785 @@
+/***
+ 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 Lesser General Public License as published by
+ the Free Software Foundation; either version 2.1 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
+ Lesser General Public License for more details.
+
+ You should have received a copy of the GNU Lesser General Public License
+ along with systemd; If not, see <http://www.gnu.org/licenses/>.
+***/
+
+#include <stddef.h>
+#include <unistd.h>
+
+#ifdef HAVE_SELINUX
+#include <selinux/selinux.h>
+#endif
+
+#include "sd-daemon.h"
+#include "sd-event.h"
+
+#include "alloc-util.h"
+#include "dirent-util.h"
+#include "escape.h"
+#include "fd-util.h"
+#include "fileio.h"
+#include "io-util.h"
+#include "journald-console.h"
+#include "journald-kmsg.h"
+#include "journald-server.h"
+#include "journald-stream.h"
+#include "journald-syslog.h"
+#include "journald-wall.h"
+#include "mkdir.h"
+#include "parse-util.h"
+#include "selinux-util.h"
+#include "socket-util.h"
+#include "stdio-util.h"
+#include "string-util.h"
+#include "syslog-util.h"
+
+#define STDOUT_STREAMS_MAX 4096
+
+typedef enum StdoutStreamState {
+ STDOUT_STREAM_IDENTIFIER,
+ STDOUT_STREAM_UNIT_ID,
+ STDOUT_STREAM_PRIORITY,
+ STDOUT_STREAM_LEVEL_PREFIX,
+ STDOUT_STREAM_FORWARD_TO_SYSLOG,
+ STDOUT_STREAM_FORWARD_TO_KMSG,
+ STDOUT_STREAM_FORWARD_TO_CONSOLE,
+ STDOUT_STREAM_RUNNING
+} StdoutStreamState;
+
+struct StdoutStream {
+ Server *server;
+ StdoutStreamState state;
+
+ int fd;
+
+ struct ucred ucred;
+ char *label;
+ char *identifier;
+ char *unit_id;
+ int priority;
+ bool level_prefix:1;
+ bool forward_to_syslog:1;
+ bool forward_to_kmsg:1;
+ bool forward_to_console:1;
+
+ bool fdstore:1;
+ bool in_notify_queue:1;
+
+ char buffer[LINE_MAX+1];
+ size_t length;
+
+ sd_event_source *event_source;
+
+ char *state_file;
+
+ LIST_FIELDS(StdoutStream, stdout_stream);
+ LIST_FIELDS(StdoutStream, stdout_stream_notify_queue);
+};
+
+void stdout_stream_free(StdoutStream *s) {
+ if (!s)
+ return;
+
+ if (s->server) {
+ assert(s->server->n_stdout_streams > 0);
+ s->server->n_stdout_streams--;
+ LIST_REMOVE(stdout_stream, s->server->stdout_streams, s);
+
+ if (s->in_notify_queue)
+ LIST_REMOVE(stdout_stream_notify_queue, s->server->stdout_streams_notify_queue, s);
+ }
+
+ if (s->event_source) {
+ sd_event_source_set_enabled(s->event_source, SD_EVENT_OFF);
+ s->event_source = sd_event_source_unref(s->event_source);
+ }
+
+ safe_close(s->fd);
+ free(s->label);
+ free(s->identifier);
+ free(s->unit_id);
+ free(s->state_file);
+
+ free(s);
+}
+
+DEFINE_TRIVIAL_CLEANUP_FUNC(StdoutStream*, stdout_stream_free);
+
+static void stdout_stream_destroy(StdoutStream *s) {
+ if (!s)
+ return;
+
+ if (s->state_file)
+ (void) unlink(s->state_file);
+
+ stdout_stream_free(s);
+}
+
+static int stdout_stream_save(StdoutStream *s) {
+ _cleanup_free_ char *temp_path = NULL;
+ _cleanup_fclose_ FILE *f = NULL;
+ int r;
+
+ assert(s);
+
+ if (s->state != STDOUT_STREAM_RUNNING)
+ return 0;
+
+ if (!s->state_file) {
+ struct stat st;
+
+ r = fstat(s->fd, &st);
+ if (r < 0)
+ return log_warning_errno(errno, "Failed to stat connected stream: %m");
+
+ /* We use device and inode numbers as identifier for the stream */
+ if (asprintf(&s->state_file, "/run/systemd/journal/streams/%lu:%lu", (unsigned long) st.st_dev, (unsigned long) st.st_ino) < 0)
+ return log_oom();
+ }
+
+ mkdir_p("/run/systemd/journal/streams", 0755);
+
+ r = fopen_temporary(s->state_file, &f, &temp_path);
+ if (r < 0)
+ goto fail;
+
+ fprintf(f,
+ "# This is private data. Do not parse\n"
+ "PRIORITY=%i\n"
+ "LEVEL_PREFIX=%i\n"
+ "FORWARD_TO_SYSLOG=%i\n"
+ "FORWARD_TO_KMSG=%i\n"
+ "FORWARD_TO_CONSOLE=%i\n",
+ s->priority,
+ s->level_prefix,
+ s->forward_to_syslog,
+ s->forward_to_kmsg,
+ s->forward_to_console);
+
+ if (!isempty(s->identifier)) {
+ _cleanup_free_ char *escaped;
+
+ escaped = cescape(s->identifier);
+ if (!escaped) {
+ r = -ENOMEM;
+ goto fail;
+ }
+
+ fprintf(f, "IDENTIFIER=%s\n", escaped);
+ }
+
+ if (!isempty(s->unit_id)) {
+ _cleanup_free_ char *escaped;
+
+ escaped = cescape(s->unit_id);
+ if (!escaped) {
+ r = -ENOMEM;
+ goto fail;
+ }
+
+ fprintf(f, "UNIT=%s\n", escaped);
+ }
+
+ r = fflush_and_check(f);
+ if (r < 0)
+ goto fail;
+
+ if (rename(temp_path, s->state_file) < 0) {
+ r = -errno;
+ goto fail;
+ }
+
+ if (!s->fdstore && !s->in_notify_queue) {
+ LIST_PREPEND(stdout_stream_notify_queue, s->server->stdout_streams_notify_queue, s);
+ s->in_notify_queue = true;
+
+ if (s->server->notify_event_source) {
+ r = sd_event_source_set_enabled(s->server->notify_event_source, SD_EVENT_ON);
+ if (r < 0)
+ log_warning_errno(r, "Failed to enable notify event source: %m");
+ }
+ }
+
+ return 0;
+
+fail:
+ (void) unlink(s->state_file);
+
+ if (temp_path)
+ (void) unlink(temp_path);
+
+ return log_error_errno(r, "Failed to save stream data %s: %m", s->state_file);
+}
+
+static int stdout_stream_log(StdoutStream *s, const char *p) {
+ struct iovec iovec[N_IOVEC_META_FIELDS + 5];
+ int priority;
+ char syslog_priority[] = "PRIORITY=\0";
+ char syslog_facility[sizeof("SYSLOG_FACILITY=")-1 + DECIMAL_STR_MAX(int) + 1];
+ _cleanup_free_ char *message = NULL, *syslog_identifier = NULL;
+ unsigned n = 0;
+ size_t label_len;
+
+ assert(s);
+ assert(p);
+
+ priority = s->priority;
+
+ if (s->level_prefix)
+ syslog_parse_priority(&p, &priority, false);
+
+ if (isempty(p))
+ return 0;
+
+ if (s->forward_to_syslog || s->server->forward_to_syslog)
+ server_forward_syslog(s->server, syslog_fixup_facility(priority), s->identifier, p, &s->ucred, NULL);
+
+ if (s->forward_to_kmsg || s->server->forward_to_kmsg)
+ server_forward_kmsg(s->server, priority, s->identifier, p, &s->ucred);
+
+ if (s->forward_to_console || s->server->forward_to_console)
+ server_forward_console(s->server, priority, s->identifier, p, &s->ucred);
+
+ if (s->server->forward_to_wall)
+ server_forward_wall(s->server, priority, s->identifier, p, &s->ucred);
+
+ IOVEC_SET_STRING(iovec[n++], "_TRANSPORT=stdout");
+
+ syslog_priority[strlen("PRIORITY=")] = '0' + LOG_PRI(priority);
+ IOVEC_SET_STRING(iovec[n++], syslog_priority);
+
+ if (priority & LOG_FACMASK) {
+ xsprintf(syslog_facility, "SYSLOG_FACILITY=%i", LOG_FAC(priority));
+ IOVEC_SET_STRING(iovec[n++], syslog_facility);
+ }
+
+ if (s->identifier) {
+ syslog_identifier = strappend("SYSLOG_IDENTIFIER=", s->identifier);
+ if (syslog_identifier)
+ IOVEC_SET_STRING(iovec[n++], syslog_identifier);
+ }
+
+ message = strappend("MESSAGE=", p);
+ if (message)
+ IOVEC_SET_STRING(iovec[n++], message);
+
+ label_len = s->label ? strlen(s->label) : 0;
+ server_dispatch_message(s->server, iovec, n, ELEMENTSOF(iovec), &s->ucred, NULL, s->label, label_len, s->unit_id, priority, 0);
+ return 0;
+}
+
+static int stdout_stream_line(StdoutStream *s, char *p) {
+ int r;
+ char *orig;
+
+ assert(s);
+ assert(p);
+
+ orig = p;
+ p = strstrip(p);
+
+ switch (s->state) {
+
+ case STDOUT_STREAM_IDENTIFIER:
+ if (isempty(p))
+ s->identifier = NULL;
+ else {
+ s->identifier = strdup(p);
+ if (!s->identifier)
+ return log_oom();
+ }
+
+ s->state = STDOUT_STREAM_UNIT_ID;
+ return 0;
+
+ case STDOUT_STREAM_UNIT_ID:
+ if (s->ucred.uid == 0) {
+ if (isempty(p))
+ s->unit_id = NULL;
+ else {
+ s->unit_id = strdup(p);
+ if (!s->unit_id)
+ return log_oom();
+ }
+ }
+
+ s->state = STDOUT_STREAM_PRIORITY;
+ return 0;
+
+ case STDOUT_STREAM_PRIORITY:
+ r = safe_atoi(p, &s->priority);
+ if (r < 0 || s->priority < 0 || s->priority > 999) {
+ log_warning("Failed to parse log priority line.");
+ return -EINVAL;
+ }
+
+ s->state = STDOUT_STREAM_LEVEL_PREFIX;
+ return 0;
+
+ case STDOUT_STREAM_LEVEL_PREFIX:
+ r = parse_boolean(p);
+ if (r < 0) {
+ log_warning("Failed to parse level prefix line.");
+ return -EINVAL;
+ }
+
+ s->level_prefix = !!r;
+ s->state = STDOUT_STREAM_FORWARD_TO_SYSLOG;
+ return 0;
+
+ case STDOUT_STREAM_FORWARD_TO_SYSLOG:
+ r = parse_boolean(p);
+ if (r < 0) {
+ log_warning("Failed to parse forward to syslog line.");
+ return -EINVAL;
+ }
+
+ s->forward_to_syslog = !!r;
+ s->state = STDOUT_STREAM_FORWARD_TO_KMSG;
+ return 0;
+
+ case STDOUT_STREAM_FORWARD_TO_KMSG:
+ r = parse_boolean(p);
+ if (r < 0) {
+ log_warning("Failed to parse copy to kmsg line.");
+ return -EINVAL;
+ }
+
+ s->forward_to_kmsg = !!r;
+ s->state = STDOUT_STREAM_FORWARD_TO_CONSOLE;
+ return 0;
+
+ case STDOUT_STREAM_FORWARD_TO_CONSOLE:
+ r = parse_boolean(p);
+ if (r < 0) {
+ log_warning("Failed to parse copy to console line.");
+ return -EINVAL;
+ }
+
+ s->forward_to_console = !!r;
+ s->state = STDOUT_STREAM_RUNNING;
+
+ /* Try to save the stream, so that journald can be restarted and we can recover */
+ (void) stdout_stream_save(s);
+ return 0;
+
+ case STDOUT_STREAM_RUNNING:
+ return stdout_stream_log(s, orig);
+ }
+
+ assert_not_reached("Unknown stream state");
+}
+
+static int stdout_stream_scan(StdoutStream *s, bool force_flush) {
+ char *p;
+ size_t remaining;
+ int r;
+
+ assert(s);
+
+ p = s->buffer;
+ remaining = s->length;
+ for (;;) {
+ char *end;
+ size_t skip;
+
+ end = memchr(p, '\n', remaining);
+ if (end)
+ skip = end - p + 1;
+ else if (remaining >= sizeof(s->buffer) - 1) {
+ end = p + sizeof(s->buffer) - 1;
+ skip = remaining;
+ } else
+ break;
+
+ *end = 0;
+
+ r = stdout_stream_line(s, p);
+ if (r < 0)
+ return r;
+
+ remaining -= skip;
+ p += skip;
+ }
+
+ if (force_flush && remaining > 0) {
+ p[remaining] = 0;
+ r = stdout_stream_line(s, p);
+ if (r < 0)
+ return r;
+
+ p += remaining;
+ remaining = 0;
+ }
+
+ if (p > s->buffer) {
+ memmove(s->buffer, p, remaining);
+ s->length = remaining;
+ }
+
+ return 0;
+}
+
+static int stdout_stream_process(sd_event_source *es, int fd, uint32_t revents, void *userdata) {
+ StdoutStream *s = userdata;
+ ssize_t l;
+ int r;
+
+ assert(s);
+
+ if ((revents|EPOLLIN|EPOLLHUP) != (EPOLLIN|EPOLLHUP)) {
+ log_error("Got invalid event from epoll for stdout stream: %"PRIx32, revents);
+ goto terminate;
+ }
+
+ l = read(s->fd, s->buffer+s->length, sizeof(s->buffer)-1-s->length);
+ if (l < 0) {
+
+ if (errno == EAGAIN)
+ return 0;
+
+ log_warning_errno(errno, "Failed to read from stream: %m");
+ goto terminate;
+ }
+
+ if (l == 0) {
+ stdout_stream_scan(s, true);
+ goto terminate;
+ }
+
+ s->length += l;
+ r = stdout_stream_scan(s, false);
+ if (r < 0)
+ goto terminate;
+
+ return 1;
+
+terminate:
+ stdout_stream_destroy(s);
+ return 0;
+}
+
+static int stdout_stream_install(Server *s, int fd, StdoutStream **ret) {
+ _cleanup_(stdout_stream_freep) StdoutStream *stream = NULL;
+ int r;
+
+ assert(s);
+ assert(fd >= 0);
+
+ stream = new0(StdoutStream, 1);
+ if (!stream)
+ return log_oom();
+
+ stream->fd = -1;
+ stream->priority = LOG_INFO;
+
+ r = getpeercred(fd, &stream->ucred);
+ if (r < 0)
+ return log_error_errno(r, "Failed to determine peer credentials: %m");
+
+ if (mac_selinux_have()) {
+ r = getpeersec(fd, &stream->label);
+ if (r < 0 && r != -EOPNOTSUPP)
+ (void) log_warning_errno(r, "Failed to determine peer security context: %m");
+ }
+
+ (void) shutdown(fd, SHUT_WR);
+
+ r = sd_event_add_io(s->event, &stream->event_source, fd, EPOLLIN, stdout_stream_process, stream);
+ if (r < 0)
+ return log_error_errno(r, "Failed to add stream to event loop: %m");
+
+ r = sd_event_source_set_priority(stream->event_source, SD_EVENT_PRIORITY_NORMAL+5);
+ if (r < 0)
+ return log_error_errno(r, "Failed to adjust stdout event source priority: %m");
+
+ stream->fd = fd;
+
+ stream->server = s;
+ LIST_PREPEND(stdout_stream, s->stdout_streams, stream);
+ s->n_stdout_streams++;
+
+ if (ret)
+ *ret = stream;
+
+ stream = NULL;
+
+ return 0;
+}
+
+static int stdout_stream_new(sd_event_source *es, int listen_fd, uint32_t revents, void *userdata) {
+ _cleanup_close_ int fd = -1;
+ Server *s = userdata;
+ int r;
+
+ assert(s);
+
+ if (revents != EPOLLIN) {
+ log_error("Got invalid event from epoll for stdout server fd: %"PRIx32, revents);
+ return -EIO;
+ }
+
+ fd = accept4(s->stdout_fd, NULL, NULL, SOCK_NONBLOCK|SOCK_CLOEXEC);
+ if (fd < 0) {
+ if (errno == EAGAIN)
+ return 0;
+
+ return log_error_errno(errno, "Failed to accept stdout connection: %m");
+ }
+
+ if (s->n_stdout_streams >= STDOUT_STREAMS_MAX) {
+ log_warning("Too many stdout streams, refusing connection.");
+ return 0;
+ }
+
+ r = stdout_stream_install(s, fd, NULL);
+ if (r < 0)
+ return r;
+
+ fd = -1;
+ return 0;
+}
+
+static int stdout_stream_load(StdoutStream *stream, const char *fname) {
+ _cleanup_free_ char
+ *priority = NULL,
+ *level_prefix = NULL,
+ *forward_to_syslog = NULL,
+ *forward_to_kmsg = NULL,
+ *forward_to_console = NULL;
+ int r;
+
+ assert(stream);
+ assert(fname);
+
+ if (!stream->state_file) {
+ stream->state_file = strappend("/run/systemd/journal/streams/", fname);
+ if (!stream->state_file)
+ return log_oom();
+ }
+
+ r = parse_env_file(stream->state_file, NEWLINE,
+ "PRIORITY", &priority,
+ "LEVEL_PREFIX", &level_prefix,
+ "FORWARD_TO_SYSLOG", &forward_to_syslog,
+ "FORWARD_TO_KMSG", &forward_to_kmsg,
+ "FORWARD_TO_CONSOLE", &forward_to_console,
+ "IDENTIFIER", &stream->identifier,
+ "UNIT", &stream->unit_id,
+ NULL);
+ if (r < 0)
+ return log_error_errno(r, "Failed to read: %s", stream->state_file);
+
+ if (priority) {
+ int p;
+
+ p = log_level_from_string(priority);
+ if (p >= 0)
+ stream->priority = p;
+ }
+
+ if (level_prefix) {
+ r = parse_boolean(level_prefix);
+ if (r >= 0)
+ stream->level_prefix = r;
+ }
+
+ if (forward_to_syslog) {
+ r = parse_boolean(forward_to_syslog);
+ if (r >= 0)
+ stream->forward_to_syslog = r;
+ }
+
+ if (forward_to_kmsg) {
+ r = parse_boolean(forward_to_kmsg);
+ if (r >= 0)
+ stream->forward_to_kmsg = r;
+ }
+
+ if (forward_to_console) {
+ r = parse_boolean(forward_to_console);
+ if (r >= 0)
+ stream->forward_to_console = r;
+ }
+
+ return 0;
+}
+
+static int stdout_stream_restore(Server *s, const char *fname, int fd) {
+ StdoutStream *stream;
+ int r;
+
+ assert(s);
+ assert(fname);
+ assert(fd >= 0);
+
+ if (s->n_stdout_streams >= STDOUT_STREAMS_MAX) {
+ log_warning("Too many stdout streams, refusing restoring of stream.");
+ return -ENOBUFS;
+ }
+
+ r = stdout_stream_install(s, fd, &stream);
+ if (r < 0)
+ return r;
+
+ stream->state = STDOUT_STREAM_RUNNING;
+ stream->fdstore = true;
+
+ /* Ignore all parsing errors */
+ (void) stdout_stream_load(stream, fname);
+
+ return 0;
+}
+
+int server_restore_streams(Server *s, FDSet *fds) {
+ _cleanup_closedir_ DIR *d = NULL;
+ struct dirent *de;
+ int r;
+
+ d = opendir("/run/systemd/journal/streams");
+ if (!d) {
+ if (errno == ENOENT)
+ return 0;
+
+ return log_warning_errno(errno, "Failed to enumerate /run/systemd/journal/streams: %m");
+ }
+
+ FOREACH_DIRENT(de, d, goto fail) {
+ unsigned long st_dev, st_ino;
+ bool found = false;
+ Iterator i;
+ int fd;
+
+ if (sscanf(de->d_name, "%lu:%lu", &st_dev, &st_ino) != 2)
+ continue;
+
+ FDSET_FOREACH(fd, fds, i) {
+ struct stat st;
+
+ if (fstat(fd, &st) < 0)
+ return log_error_errno(errno, "Failed to stat %s: %m", de->d_name);
+
+ if (S_ISSOCK(st.st_mode) && st.st_dev == st_dev && st.st_ino == st_ino) {
+ found = true;
+ break;
+ }
+ }
+
+ if (!found) {
+ /* No file descriptor? Then let's delete the state file */
+ log_debug("Cannot restore stream file %s", de->d_name);
+ unlinkat(dirfd(d), de->d_name, 0);
+ continue;
+ }
+
+ fdset_remove(fds, fd);
+
+ r = stdout_stream_restore(s, de->d_name, fd);
+ if (r < 0)
+ safe_close(fd);
+ }
+
+ return 0;
+
+fail:
+ return log_error_errno(errno, "Failed to read streams directory: %m");
+}
+
+int server_open_stdout_socket(Server *s) {
+ static const union sockaddr_union sa = {
+ .un.sun_family = AF_UNIX,
+ .un.sun_path = "/run/systemd/journal/stdout",
+ };
+ int r;
+
+ assert(s);
+
+ if (s->stdout_fd < 0) {
+ s->stdout_fd = socket(AF_UNIX, SOCK_STREAM|SOCK_CLOEXEC|SOCK_NONBLOCK, 0);
+ if (s->stdout_fd < 0)
+ return log_error_errno(errno, "socket() failed: %m");
+
+ (void) unlink(sa.un.sun_path);
+
+ r = bind(s->stdout_fd, &sa.sa, SOCKADDR_UN_LEN(sa.un));
+ if (r < 0)
+ return log_error_errno(errno, "bind(%s) failed: %m", sa.un.sun_path);
+
+ (void) chmod(sa.un.sun_path, 0666);
+
+ if (listen(s->stdout_fd, SOMAXCONN) < 0)
+ return log_error_errno(errno, "listen(%s) failed: %m", sa.un.sun_path);
+ } else
+ fd_nonblock(s->stdout_fd, 1);
+
+ r = sd_event_add_io(s->event, &s->stdout_event_source, s->stdout_fd, EPOLLIN, stdout_stream_new, s);
+ if (r < 0)
+ return log_error_errno(r, "Failed to add stdout server fd to event source: %m");
+
+ r = sd_event_source_set_priority(s->stdout_event_source, SD_EVENT_PRIORITY_NORMAL+5);
+ if (r < 0)
+ return log_error_errno(r, "Failed to adjust priority of stdout server event source: %m");
+
+ return 0;
+}
+
+void stdout_stream_send_notify(StdoutStream *s) {
+ struct iovec iovec = {
+ .iov_base = (char*) "FDSTORE=1",
+ .iov_len = strlen("FDSTORE=1"),
+ };
+ struct msghdr msghdr = {
+ .msg_iov = &iovec,
+ .msg_iovlen = 1,
+ };
+ struct cmsghdr *cmsg;
+ ssize_t l;
+
+ assert(s);
+ assert(!s->fdstore);
+ assert(s->in_notify_queue);
+ assert(s->server);
+ assert(s->server->notify_fd >= 0);
+
+ /* Store the connection fd in PID 1, so that we get it passed
+ * in again on next start */
+
+ msghdr.msg_controllen = CMSG_SPACE(sizeof(int));
+ msghdr.msg_control = alloca0(msghdr.msg_controllen);
+
+ cmsg = CMSG_FIRSTHDR(&msghdr);
+ cmsg->cmsg_level = SOL_SOCKET;
+ cmsg->cmsg_type = SCM_RIGHTS;
+ cmsg->cmsg_len = CMSG_LEN(sizeof(int));
+
+ memcpy(CMSG_DATA(cmsg), &s->fd, sizeof(int));
+
+ l = sendmsg(s->server->notify_fd, &msghdr, MSG_DONTWAIT|MSG_NOSIGNAL);
+ if (l < 0) {
+ if (errno == EAGAIN)
+ return;
+
+ log_error_errno(errno, "Failed to send stream file descriptor to service manager: %m");
+ } else {
+ log_debug("Successfully sent stream file descriptor to service manager.");
+ s->fdstore = 1;
+ }
+
+ LIST_REMOVE(stdout_stream_notify_queue, s->server->stdout_streams_notify_queue, s);
+ s->in_notify_queue = false;
+
+}
diff --git a/src/grp-journal/libjournal-core/journald-stream.h b/src/grp-journal/libjournal-core/journald-stream.h
new file mode 100644
index 0000000000..c971d0a00d
--- /dev/null
+++ b/src/grp-journal/libjournal-core/journald-stream.h
@@ -0,0 +1,32 @@
+#pragma once
+
+/***
+ 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 Lesser General Public License as published by
+ the Free Software Foundation; either version 2.1 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
+ Lesser General Public License for more details.
+
+ You should have received a copy of the GNU Lesser General Public License
+ along with systemd; If not, see <http://www.gnu.org/licenses/>.
+***/
+
+#include "basic/fdset.h"
+
+typedef struct StdoutStream StdoutStream;
+
+#include "journald-server.h"
+
+int server_open_stdout_socket(Server *s);
+int server_restore_streams(Server *s, FDSet *fds);
+
+void stdout_stream_free(StdoutStream *s);
+void stdout_stream_send_notify(StdoutStream *s);
diff --git a/src/grp-journal/libjournal-core/journald-syslog.c b/src/grp-journal/libjournal-core/journald-syslog.c
new file mode 100644
index 0000000000..bc793235d4
--- /dev/null
+++ b/src/grp-journal/libjournal-core/journald-syslog.c
@@ -0,0 +1,455 @@
+/***
+ 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 Lesser General Public License as published by
+ the Free Software Foundation; either version 2.1 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
+ Lesser General Public License for more details.
+
+ You should have received a copy of the GNU Lesser General Public License
+ along with systemd; If not, see <http://www.gnu.org/licenses/>.
+***/
+
+#include <stddef.h>
+#include <sys/epoll.h>
+#include <unistd.h>
+
+#include <systemd/sd-messages.h>
+
+#include "basic/alloc-util.h"
+#include "basic/fd-util.h"
+#include "basic/formats-util.h"
+#include "basic/io-util.h"
+#include "basic/process-util.h"
+#include "basic/selinux-util.h"
+#include "basic/socket-util.h"
+#include "basic/stdio-util.h"
+#include "basic/string-util.h"
+#include "basic/syslog-util.h"
+
+#include "journald-console.h"
+#include "journald-kmsg.h"
+#include "journald-server.h"
+#include "journald-syslog.h"
+#include "journald-wall.h"
+
+/* Warn once every 30s if we missed syslog message */
+#define WARN_FORWARD_SYSLOG_MISSED_USEC (30 * USEC_PER_SEC)
+
+static void forward_syslog_iovec(Server *s, const struct iovec *iovec, unsigned n_iovec, const struct ucred *ucred, const struct timeval *tv) {
+
+ static const union sockaddr_union sa = {
+ .un.sun_family = AF_UNIX,
+ .un.sun_path = "/run/systemd/journal/syslog",
+ };
+ struct msghdr msghdr = {
+ .msg_iov = (struct iovec *) iovec,
+ .msg_iovlen = n_iovec,
+ .msg_name = (struct sockaddr*) &sa.sa,
+ .msg_namelen = SOCKADDR_UN_LEN(sa.un),
+ };
+ struct cmsghdr *cmsg;
+ union {
+ struct cmsghdr cmsghdr;
+ uint8_t buf[CMSG_SPACE(sizeof(struct ucred))];
+ } control;
+
+ assert(s);
+ assert(iovec);
+ assert(n_iovec > 0);
+
+ if (ucred) {
+ zero(control);
+ msghdr.msg_control = &control;
+ msghdr.msg_controllen = sizeof(control);
+
+ cmsg = CMSG_FIRSTHDR(&msghdr);
+ cmsg->cmsg_level = SOL_SOCKET;
+ cmsg->cmsg_type = SCM_CREDENTIALS;
+ cmsg->cmsg_len = CMSG_LEN(sizeof(struct ucred));
+ memcpy(CMSG_DATA(cmsg), ucred, sizeof(struct ucred));
+ msghdr.msg_controllen = cmsg->cmsg_len;
+ }
+
+ /* Forward the syslog message we received via /dev/log to
+ * /run/systemd/syslog. Unfortunately we currently can't set
+ * the SO_TIMESTAMP auxiliary data, and hence we don't. */
+
+ if (sendmsg(s->syslog_fd, &msghdr, MSG_NOSIGNAL) >= 0)
+ return;
+
+ /* The socket is full? I guess the syslog implementation is
+ * too slow, and we shouldn't wait for that... */
+ if (errno == EAGAIN) {
+ s->n_forward_syslog_missed++;
+ return;
+ }
+
+ if (ucred && (errno == ESRCH || errno == EPERM)) {
+ struct ucred u;
+
+ /* Hmm, presumably the sender process vanished
+ * by now, or we don't have CAP_SYS_AMDIN, so
+ * let's fix it as good as we can, and retry */
+
+ u = *ucred;
+ u.pid = getpid();
+ memcpy(CMSG_DATA(cmsg), &u, sizeof(struct ucred));
+
+ if (sendmsg(s->syslog_fd, &msghdr, MSG_NOSIGNAL) >= 0)
+ return;
+
+ if (errno == EAGAIN) {
+ s->n_forward_syslog_missed++;
+ return;
+ }
+ }
+
+ if (errno != ENOENT)
+ log_debug_errno(errno, "Failed to forward syslog message: %m");
+}
+
+static void forward_syslog_raw(Server *s, int priority, const char *buffer, const struct ucred *ucred, const struct timeval *tv) {
+ struct iovec iovec;
+
+ assert(s);
+ assert(buffer);
+
+ if (LOG_PRI(priority) > s->max_level_syslog)
+ return;
+
+ IOVEC_SET_STRING(iovec, buffer);
+ forward_syslog_iovec(s, &iovec, 1, ucred, tv);
+}
+
+void server_forward_syslog(Server *s, int priority, const char *identifier, const char *message, const struct ucred *ucred, const struct timeval *tv) {
+ struct iovec iovec[5];
+ char header_priority[DECIMAL_STR_MAX(priority) + 3], header_time[64],
+ header_pid[sizeof("[]: ")-1 + DECIMAL_STR_MAX(pid_t) + 1];
+ int n = 0;
+ time_t t;
+ struct tm *tm;
+ char *ident_buf = NULL;
+
+ assert(s);
+ assert(priority >= 0);
+ assert(priority <= 999);
+ assert(message);
+
+ if (LOG_PRI(priority) > s->max_level_syslog)
+ return;
+
+ /* First: priority field */
+ xsprintf(header_priority, "<%i>", priority);
+ IOVEC_SET_STRING(iovec[n++], header_priority);
+
+ /* Second: timestamp */
+ t = tv ? tv->tv_sec : ((time_t) (now(CLOCK_REALTIME) / USEC_PER_SEC));
+ tm = localtime(&t);
+ if (!tm)
+ return;
+ if (strftime(header_time, sizeof(header_time), "%h %e %T ", tm) <= 0)
+ return;
+ IOVEC_SET_STRING(iovec[n++], header_time);
+
+ /* Third: identifier and PID */
+ if (ucred) {
+ if (!identifier) {
+ get_process_comm(ucred->pid, &ident_buf);
+ identifier = ident_buf;
+ }
+
+ xsprintf(header_pid, "["PID_FMT"]: ", ucred->pid);
+
+ if (identifier)
+ IOVEC_SET_STRING(iovec[n++], identifier);
+
+ IOVEC_SET_STRING(iovec[n++], header_pid);
+ } else if (identifier) {
+ IOVEC_SET_STRING(iovec[n++], identifier);
+ IOVEC_SET_STRING(iovec[n++], ": ");
+ }
+
+ /* Fourth: message */
+ IOVEC_SET_STRING(iovec[n++], message);
+
+ forward_syslog_iovec(s, iovec, n, ucred, tv);
+
+ free(ident_buf);
+}
+
+int syslog_fixup_facility(int priority) {
+
+ if ((priority & LOG_FACMASK) == 0)
+ return (priority & LOG_PRIMASK) | LOG_USER;
+
+ return priority;
+}
+
+size_t syslog_parse_identifier(const char **buf, char **identifier, char **pid) {
+ const char *p;
+ char *t;
+ size_t l, e;
+
+ assert(buf);
+ assert(identifier);
+ assert(pid);
+
+ p = *buf;
+
+ p += strspn(p, WHITESPACE);
+ l = strcspn(p, WHITESPACE);
+
+ if (l <= 0 ||
+ p[l-1] != ':')
+ return 0;
+
+ e = l;
+ l--;
+
+ if (p[l-1] == ']') {
+ size_t k = l-1;
+
+ for (;;) {
+
+ if (p[k] == '[') {
+ t = strndup(p+k+1, l-k-2);
+ if (t)
+ *pid = t;
+
+ l = k;
+ break;
+ }
+
+ if (k == 0)
+ break;
+
+ k--;
+ }
+ }
+
+ t = strndup(p, l);
+ if (t)
+ *identifier = t;
+
+ if (strchr(WHITESPACE, p[e]))
+ e++;
+ *buf = p + e;
+ return e;
+}
+
+static void syslog_skip_date(char **buf) {
+ enum {
+ LETTER,
+ SPACE,
+ NUMBER,
+ SPACE_OR_NUMBER,
+ COLON
+ } sequence[] = {
+ LETTER, LETTER, LETTER,
+ SPACE,
+ SPACE_OR_NUMBER, NUMBER,
+ SPACE,
+ SPACE_OR_NUMBER, NUMBER,
+ COLON,
+ SPACE_OR_NUMBER, NUMBER,
+ COLON,
+ SPACE_OR_NUMBER, NUMBER,
+ SPACE
+ };
+
+ char *p;
+ unsigned i;
+
+ assert(buf);
+ assert(*buf);
+
+ p = *buf;
+
+ for (i = 0; i < ELEMENTSOF(sequence); i++, p++) {
+
+ if (!*p)
+ return;
+
+ switch (sequence[i]) {
+
+ case SPACE:
+ if (*p != ' ')
+ return;
+ break;
+
+ case SPACE_OR_NUMBER:
+ if (*p == ' ')
+ break;
+
+ /* fall through */
+
+ case NUMBER:
+ if (*p < '0' || *p > '9')
+ return;
+
+ break;
+
+ case LETTER:
+ if (!(*p >= 'A' && *p <= 'Z') &&
+ !(*p >= 'a' && *p <= 'z'))
+ return;
+
+ break;
+
+ case COLON:
+ if (*p != ':')
+ return;
+ break;
+
+ }
+ }
+
+ *buf = p;
+}
+
+void server_process_syslog_message(
+ Server *s,
+ const char *buf,
+ const struct ucred *ucred,
+ const struct timeval *tv,
+ const char *label,
+ size_t label_len) {
+
+ char syslog_priority[sizeof("PRIORITY=") + DECIMAL_STR_MAX(int)],
+ syslog_facility[sizeof("SYSLOG_FACILITY=") + DECIMAL_STR_MAX(int)];
+ const char *message = NULL, *syslog_identifier = NULL, *syslog_pid = NULL;
+ struct iovec iovec[N_IOVEC_META_FIELDS + 6];
+ unsigned n = 0;
+ int priority = LOG_USER | LOG_INFO;
+ _cleanup_free_ char *identifier = NULL, *pid = NULL;
+ const char *orig;
+
+ assert(s);
+ assert(buf);
+
+ orig = buf;
+ syslog_parse_priority(&buf, &priority, true);
+
+ if (s->forward_to_syslog)
+ forward_syslog_raw(s, priority, orig, ucred, tv);
+
+ syslog_skip_date((char**) &buf);
+ syslog_parse_identifier(&buf, &identifier, &pid);
+
+ if (s->forward_to_kmsg)
+ server_forward_kmsg(s, priority, identifier, buf, ucred);
+
+ if (s->forward_to_console)
+ server_forward_console(s, priority, identifier, buf, ucred);
+
+ if (s->forward_to_wall)
+ server_forward_wall(s, priority, identifier, buf, ucred);
+
+ IOVEC_SET_STRING(iovec[n++], "_TRANSPORT=syslog");
+
+ xsprintf(syslog_priority, "PRIORITY=%i", priority & LOG_PRIMASK);
+ IOVEC_SET_STRING(iovec[n++], syslog_priority);
+
+ if (priority & LOG_FACMASK) {
+ xsprintf(syslog_facility, "SYSLOG_FACILITY=%i", LOG_FAC(priority));
+ IOVEC_SET_STRING(iovec[n++], syslog_facility);
+ }
+
+ if (identifier) {
+ syslog_identifier = strjoina("SYSLOG_IDENTIFIER=", identifier);
+ IOVEC_SET_STRING(iovec[n++], syslog_identifier);
+ }
+
+ if (pid) {
+ syslog_pid = strjoina("SYSLOG_PID=", pid);
+ IOVEC_SET_STRING(iovec[n++], syslog_pid);
+ }
+
+ message = strjoina("MESSAGE=", buf);
+ if (message)
+ IOVEC_SET_STRING(iovec[n++], message);
+
+ server_dispatch_message(s, iovec, n, ELEMENTSOF(iovec), ucred, tv, label, label_len, NULL, priority, 0);
+}
+
+int server_open_syslog_socket(Server *s) {
+
+ static const union sockaddr_union sa = {
+ .un.sun_family = AF_UNIX,
+ .un.sun_path = "/run/systemd/journal/dev-log",
+ };
+ static const int one = 1;
+ int r;
+
+ assert(s);
+
+ if (s->syslog_fd < 0) {
+ s->syslog_fd = socket(AF_UNIX, SOCK_DGRAM|SOCK_CLOEXEC|SOCK_NONBLOCK, 0);
+ if (s->syslog_fd < 0)
+ return log_error_errno(errno, "socket() failed: %m");
+
+ (void) unlink(sa.un.sun_path);
+
+ r = bind(s->syslog_fd, &sa.sa, SOCKADDR_UN_LEN(sa.un));
+ if (r < 0)
+ return log_error_errno(errno, "bind(%s) failed: %m", sa.un.sun_path);
+
+ (void) chmod(sa.un.sun_path, 0666);
+ } else
+ fd_nonblock(s->syslog_fd, 1);
+
+ r = setsockopt(s->syslog_fd, SOL_SOCKET, SO_PASSCRED, &one, sizeof(one));
+ if (r < 0)
+ return log_error_errno(errno, "SO_PASSCRED failed: %m");
+
+#ifdef HAVE_SELINUX
+ if (mac_selinux_have()) {
+ r = setsockopt(s->syslog_fd, SOL_SOCKET, SO_PASSSEC, &one, sizeof(one));
+ if (r < 0)
+ log_warning_errno(errno, "SO_PASSSEC failed: %m");
+ }
+#endif
+
+ r = setsockopt(s->syslog_fd, SOL_SOCKET, SO_TIMESTAMP, &one, sizeof(one));
+ if (r < 0)
+ return log_error_errno(errno, "SO_TIMESTAMP failed: %m");
+
+ r = sd_event_add_io(s->event, &s->syslog_event_source, s->syslog_fd, EPOLLIN, server_process_datagram, s);
+ if (r < 0)
+ return log_error_errno(r, "Failed to add syslog server fd to event loop: %m");
+
+ r = sd_event_source_set_priority(s->syslog_event_source, SD_EVENT_PRIORITY_NORMAL+5);
+ if (r < 0)
+ return log_error_errno(r, "Failed to adjust syslog event source priority: %m");
+
+ return 0;
+}
+
+void server_maybe_warn_forward_syslog_missed(Server *s) {
+ usec_t n;
+
+ assert(s);
+
+ if (s->n_forward_syslog_missed <= 0)
+ return;
+
+ n = now(CLOCK_MONOTONIC);
+ if (s->last_warn_forward_syslog_missed + WARN_FORWARD_SYSLOG_MISSED_USEC > n)
+ return;
+
+ server_driver_message(s, SD_MESSAGE_FORWARD_SYSLOG_MISSED,
+ LOG_MESSAGE("Forwarding to syslog missed %u messages.",
+ s->n_forward_syslog_missed),
+ NULL);
+
+ s->n_forward_syslog_missed = 0;
+ s->last_warn_forward_syslog_missed = n;
+}
diff --git a/src/grp-journal/libjournal-core/journald-syslog.h b/src/grp-journal/libjournal-core/journald-syslog.h
new file mode 100644
index 0000000000..46ad715314
--- /dev/null
+++ b/src/grp-journal/libjournal-core/journald-syslog.h
@@ -0,0 +1,33 @@
+#pragma once
+
+/***
+ 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 Lesser General Public License as published by
+ the Free Software Foundation; either version 2.1 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
+ Lesser General Public License for more details.
+
+ You should have received a copy of the GNU Lesser General Public License
+ along with systemd; If not, see <http://www.gnu.org/licenses/>.
+***/
+
+#include "journald-server.h"
+
+int syslog_fixup_facility(int priority) _const_;
+
+size_t syslog_parse_identifier(const char **buf, char **identifier, char **pid);
+
+void server_forward_syslog(Server *s, int priority, const char *identifier, const char *message, const struct ucred *ucred, const struct timeval *tv);
+
+void server_process_syslog_message(Server *s, const char *buf, const struct ucred *ucred, const struct timeval *tv, const char *label, size_t label_len);
+int server_open_syslog_socket(Server *s);
+
+void server_maybe_warn_forward_syslog_missed(Server *s);
diff --git a/src/grp-journal/libjournal-core/journald-wall.c b/src/grp-journal/libjournal-core/journald-wall.c
new file mode 100644
index 0000000000..0c816de55f
--- /dev/null
+++ b/src/grp-journal/libjournal-core/journald-wall.c
@@ -0,0 +1,72 @@
+/***
+ This file is part of systemd.
+
+ Copyright 2014 Sebastian Thorarensen
+
+ systemd is free software; you can redistribute it and/or modify it
+ under the terms of the GNU Lesser General Public License as published by
+ the Free Software Foundation; either version 2.1 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
+ Lesser General Public License for more details.
+
+ You should have received a copy of the GNU Lesser General Public License
+ along with systemd; If not, see <http://www.gnu.org/licenses/>.
+***/
+
+#include "basic/alloc-util.h"
+#include "basic/formats-util.h"
+#include "basic/process-util.h"
+#include "basic/string-util.h"
+#include "shared/utmp-wtmp.h"
+
+#include "journald-server.h"
+#include "journald-wall.h"
+
+void server_forward_wall(
+ Server *s,
+ int priority,
+ const char *identifier,
+ const char *message,
+ const struct ucred *ucred) {
+
+ _cleanup_free_ char *ident_buf = NULL, *l_buf = NULL;
+ const char *l;
+ int r;
+
+ assert(s);
+ assert(message);
+
+ if (LOG_PRI(priority) > s->max_level_wall)
+ return;
+
+ if (ucred) {
+ if (!identifier) {
+ get_process_comm(ucred->pid, &ident_buf);
+ identifier = ident_buf;
+ }
+
+ if (asprintf(&l_buf, "%s["PID_FMT"]: %s", strempty(identifier), ucred->pid, message) < 0) {
+ log_oom();
+ return;
+ }
+
+ l = l_buf;
+
+ } else if (identifier) {
+
+ l = l_buf = strjoin(identifier, ": ", message, NULL);
+ if (!l_buf) {
+ log_oom();
+ return;
+ }
+ } else
+ l = message;
+
+ r = utmp_wall(l, "systemd-journald", NULL, NULL, NULL);
+ if (r < 0)
+ log_debug_errno(r, "Failed to send wall message: %m");
+}
diff --git a/src/grp-journal/libjournal-core/journald-wall.h b/src/grp-journal/libjournal-core/journald-wall.h
new file mode 100644
index 0000000000..ebc2b89fa8
--- /dev/null
+++ b/src/grp-journal/libjournal-core/journald-wall.h
@@ -0,0 +1,24 @@
+#pragma once
+
+/***
+ This file is part of systemd.
+
+ Copyright 2014 Sebastian Thorarensen
+
+ systemd is free software; you can redistribute it and/or modify it
+ under the terms of the GNU Lesser General Public License as published by
+ the Free Software Foundation; either version 2.1 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
+ Lesser General Public License for more details.
+
+ You should have received a copy of the GNU Lesser General Public License
+ along with systemd; If not, see <http://www.gnu.org/licenses/>.
+***/
+
+#include "journald-server.h"
+
+void server_forward_wall(Server *s, int priority, const char *identifier, const char *message, const struct ucred *ucred);
diff --git a/src/grp-journal/libjournal-core/test-audit-type.c b/src/grp-journal/libjournal-core/test-audit-type.c
new file mode 100644
index 0000000000..812a3953ea
--- /dev/null
+++ b/src/grp-journal/libjournal-core/test-audit-type.c
@@ -0,0 +1,43 @@
+/***
+ This file is part of systemd.
+
+ Copyright 2015 Zbigniew Jędrzejewski-Szmek
+
+ systemd is free software; you can redistribute it and/or modify it
+ under the terms of the GNU Lesser General Public License as published by
+ the Free Software Foundation; either version 2.1 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
+ Lesser General Public License for more details.
+
+ You should have received a copy of the GNU Lesser General Public License
+ along with systemd; If not, see <http://www.gnu.org/licenses/>.
+***/
+
+#include <stdio.h>
+
+#include <linux/audit.h>
+
+#include "sd-journal/audit-type.h"
+
+static void print_audit_label(int i) {
+ const char *name;
+
+ name = audit_type_name_alloca(i);
+ /* This is a separate function only because of alloca */
+ printf("%i → %s → %s\n", i, audit_type_to_string(i), name);
+}
+
+static void test_audit_type(void) {
+ int i;
+
+ for (i = 0; i <= AUDIT_KERNEL; i++)
+ print_audit_label(i);
+}
+
+int main(int argc, char **argv) {
+ test_audit_type();
+}
diff --git a/src/grp-journal/libjournal-core/test-catalog.c b/src/grp-journal/libjournal-core/test-catalog.c
new file mode 100644
index 0000000000..73ffb16104
--- /dev/null
+++ b/src/grp-journal/libjournal-core/test-catalog.c
@@ -0,0 +1,264 @@
+/***
+ This file is part of systemd.
+
+ Copyright 2012 Lennart Poettering
+ Copyright 2013 Zbigniew Jędrzejewski-Szmek
+
+ systemd is free software; you can redistribute it and/or modify it
+ under the terms of the GNU Lesser General Public License as published by
+ the Free Software Foundation; either version 2.1 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
+ Lesser General Public License for more details.
+
+ You should have received a copy of the GNU Lesser General Public License
+ along with systemd; If not, see <http://www.gnu.org/licenses/>.
+***/
+
+#include <errno.h>
+#include <fcntl.h>
+#include <locale.h>
+#include <unistd.h>
+
+#include <systemd/sd-messages.h>
+
+#include "basic/alloc-util.h"
+#include "basic/fd-util.h"
+#include "basic/fileio.h"
+#include "basic/log.h"
+#include "basic/macro.h"
+#include "basic/string-util.h"
+#include "basic/util.h"
+#include "sd-journal/catalog.h"
+
+static const char *catalog_dirs[] = {
+ CATALOG_DIR,
+ NULL,
+};
+
+static const char *no_catalog_dirs[] = {
+ "/bin/hopefully/with/no/catalog",
+ NULL
+};
+
+static Hashmap * test_import(const char* contents, ssize_t size, int code) {
+ int r;
+ char name[] = "/tmp/test-catalog.XXXXXX";
+ _cleanup_close_ int fd;
+ Hashmap *h;
+
+ if (size < 0)
+ size = strlen(contents);
+
+ assert_se(h = hashmap_new(&catalog_hash_ops));
+
+ fd = mkostemp_safe(name, O_RDWR|O_CLOEXEC);
+ assert_se(fd >= 0);
+ assert_se(write(fd, contents, size) == size);
+
+ r = catalog_import_file(h, name);
+ assert_se(r == code);
+
+ unlink(name);
+
+ return h;
+}
+
+static void test_catalog_import_invalid(void) {
+ _cleanup_hashmap_free_free_free_ Hashmap *h = NULL;
+
+ h = test_import("xxx", -1, -EINVAL);
+ assert_se(hashmap_isempty(h));
+}
+
+static void test_catalog_import_badid(void) {
+ _cleanup_hashmap_free_free_free_ Hashmap *h = NULL;
+ const char *input =
+"-- 0027229ca0644181a76c4e92458afaff dededededededededededededededede\n" \
+"Subject: message\n" \
+"\n" \
+"payload\n";
+ h = test_import(input, -1, -EINVAL);
+}
+
+static void test_catalog_import_one(void) {
+ _cleanup_hashmap_free_free_free_ Hashmap *h = NULL;
+ char *payload;
+ Iterator j;
+
+ const char *input =
+"-- 0027229ca0644181a76c4e92458afaff dededededededededededededededed\n" \
+"Subject: message\n" \
+"\n" \
+"payload\n";
+ const char *expect =
+"Subject: message\n" \
+"\n" \
+"payload\n";
+
+ h = test_import(input, -1, 0);
+ assert_se(hashmap_size(h) == 1);
+
+ HASHMAP_FOREACH(payload, h, j) {
+ printf("expect: %s\n", expect);
+ printf("actual: %s\n", payload);
+ assert_se(streq(expect, payload));
+ }
+}
+
+static void test_catalog_import_merge(void) {
+ _cleanup_hashmap_free_free_free_ Hashmap *h = NULL;
+ char *payload;
+ Iterator j;
+
+ const char *input =
+"-- 0027229ca0644181a76c4e92458afaff dededededededededededededededed\n" \
+"Subject: message\n" \
+"Defined-By: me\n" \
+"\n" \
+"payload\n" \
+"\n" \
+"-- 0027229ca0644181a76c4e92458afaff dededededededededededededededed\n" \
+"Subject: override subject\n" \
+"X-Header: hello\n" \
+"\n" \
+"override payload\n";
+
+ const char *combined =
+"Subject: override subject\n" \
+"X-Header: hello\n" \
+"Subject: message\n" \
+"Defined-By: me\n" \
+"\n" \
+"override payload\n";
+
+ h = test_import(input, -1, 0);
+ assert_se(hashmap_size(h) == 1);
+
+ HASHMAP_FOREACH(payload, h, j) {
+ assert_se(streq(combined, payload));
+ }
+}
+
+static void test_catalog_import_merge_no_body(void) {
+ _cleanup_hashmap_free_free_free_ Hashmap *h = NULL;
+ char *payload;
+ Iterator j;
+
+ const char *input =
+"-- 0027229ca0644181a76c4e92458afaff dededededededededededededededed\n" \
+"Subject: message\n" \
+"Defined-By: me\n" \
+"\n" \
+"payload\n" \
+"\n" \
+"-- 0027229ca0644181a76c4e92458afaff dededededededededededededededed\n" \
+"Subject: override subject\n" \
+"X-Header: hello\n" \
+"\n";
+
+ const char *combined =
+"Subject: override subject\n" \
+"X-Header: hello\n" \
+"Subject: message\n" \
+"Defined-By: me\n" \
+"\n" \
+"payload\n";
+
+ h = test_import(input, -1, 0);
+ assert_se(hashmap_size(h) == 1);
+
+ HASHMAP_FOREACH(payload, h, j) {
+ assert_se(streq(combined, payload));
+ }
+}
+
+static const char* database = NULL;
+
+static void test_catalog_update(void) {
+ static char name[] = "/tmp/test-catalog.XXXXXX";
+ int r;
+
+ r = mkostemp_safe(name, O_RDWR|O_CLOEXEC);
+ assert_se(r >= 0);
+
+ database = name;
+
+ /* Test what happens if there are no files. */
+ r = catalog_update(database, NULL, NULL);
+ assert_se(r >= 0);
+
+ /* Test what happens if there are no files in the directory. */
+ r = catalog_update(database, NULL, no_catalog_dirs);
+ assert_se(r >= 0);
+
+ /* Make sure that we at least have some files loaded or the
+ catalog_list below will fail. */
+ r = catalog_update(database, NULL, catalog_dirs);
+ assert_se(r >= 0);
+}
+
+static void test_catalog_file_lang(void) {
+ _cleanup_free_ char *lang = NULL, *lang2 = NULL, *lang3 = NULL, *lang4 = NULL;
+
+ assert_se(catalog_file_lang("systemd.de_DE.catalog", &lang) == 1);
+ assert_se(streq(lang, "de_DE"));
+
+ assert_se(catalog_file_lang("systemd..catalog", &lang2) == 0);
+ assert_se(lang2 == NULL);
+
+ assert_se(catalog_file_lang("systemd.fr.catalog", &lang2) == 1);
+ assert_se(streq(lang2, "fr"));
+
+ assert_se(catalog_file_lang("systemd.fr.catalog.gz", &lang3) == 0);
+ assert_se(lang3 == NULL);
+
+ assert_se(catalog_file_lang("systemd.01234567890123456789012345678901.catalog", &lang3) == 0);
+ assert_se(lang3 == NULL);
+
+ assert_se(catalog_file_lang("systemd.0123456789012345678901234567890.catalog", &lang3) == 1);
+ assert_se(streq(lang3, "0123456789012345678901234567890"));
+
+ assert_se(catalog_file_lang("/x/y/systemd.catalog", &lang4) == 0);
+ assert_se(lang4 == NULL);
+
+ assert_se(catalog_file_lang("/x/y/systemd.ru_RU.catalog", &lang4) == 1);
+ assert_se(streq(lang4, "ru_RU"));
+}
+
+int main(int argc, char *argv[]) {
+ _cleanup_free_ char *text = NULL;
+ int r;
+
+ setlocale(LC_ALL, "de_DE.UTF-8");
+
+ log_parse_environment();
+ log_open();
+
+ test_catalog_file_lang();
+
+ test_catalog_import_invalid();
+ test_catalog_import_badid();
+ test_catalog_import_one();
+ test_catalog_import_merge();
+ test_catalog_import_merge_no_body();
+
+ test_catalog_update();
+
+ r = catalog_list(stdout, database, true);
+ assert_se(r >= 0);
+
+ r = catalog_list(stdout, database, false);
+ assert_se(r >= 0);
+
+ assert_se(catalog_get(database, SD_MESSAGE_COREDUMP, &text) >= 0);
+ printf(">>>%s<<<\n", text);
+
+ if (database)
+ unlink(database);
+
+ return 0;
+}
diff --git a/src/grp-journal/libjournal-core/test-compress-benchmark.c b/src/grp-journal/libjournal-core/test-compress-benchmark.c
new file mode 100644
index 0000000000..4292e4f8cb
--- /dev/null
+++ b/src/grp-journal/libjournal-core/test-compress-benchmark.c
@@ -0,0 +1,180 @@
+/***
+ This file is part of systemd
+
+ Copyright 2014 Zbigniew Jędrzejewski-Szmek
+
+ systemd is free software; you can redistribute it and/or modify it
+ under the terms of the GNU Lesser General Public License as published by
+ the Free Software Foundation; either version 2.1 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
+ Lesser General Public License for more details.
+
+ You should have received a copy of the GNU Lesser General Public License
+ along with systemd; If not, see <http://www.gnu.org/licenses/>.
+***/
+
+#include "basic/alloc-util.h"
+#include "basic/macro.h"
+#include "basic/parse-util.h"
+#include "basic/random-util.h"
+#include "basic/string-util.h"
+#include "basic/util.h"
+#include "sd-journal/compress.h"
+
+typedef int (compress_t)(const void *src, uint64_t src_size, void *dst,
+ size_t dst_alloc_size, size_t *dst_size);
+typedef int (decompress_t)(const void *src, uint64_t src_size,
+ void **dst, size_t *dst_alloc_size, size_t* dst_size, size_t dst_max);
+
+static usec_t arg_duration = 2 * USEC_PER_SEC;
+static size_t arg_start;
+
+#define MAX_SIZE (1024*1024LU)
+#define PRIME 1048571 /* A prime close enough to one megabyte that mod 4 == 3 */
+
+static size_t _permute(size_t x) {
+ size_t residue;
+
+ if (x >= PRIME)
+ return x;
+
+ residue = x*x % PRIME;
+ if (x <= PRIME / 2)
+ return residue;
+ else
+ return PRIME - residue;
+}
+
+static size_t permute(size_t x) {
+ return _permute((_permute(x) + arg_start) % MAX_SIZE ^ 0xFF345);
+}
+
+static char* make_buf(size_t count, const char *type) {
+ char *buf;
+ size_t i;
+
+ buf = malloc(count);
+ assert_se(buf);
+
+ if (streq(type, "zeros"))
+ memzero(buf, count);
+ else if (streq(type, "simple"))
+ for (i = 0; i < count; i++)
+ buf[i] = 'a' + i % ('z' - 'a' + 1);
+ else if (streq(type, "random")) {
+ size_t step = count / 10;
+
+ random_bytes(buf, step);
+ memzero(buf + 1*step, step);
+ random_bytes(buf + 2*step, step);
+ memzero(buf + 3*step, step);
+ random_bytes(buf + 4*step, step);
+ memzero(buf + 5*step, step);
+ random_bytes(buf + 6*step, step);
+ memzero(buf + 7*step, step);
+ random_bytes(buf + 8*step, step);
+ memzero(buf + 9*step, step);
+ } else
+ assert_not_reached("here");
+
+ return buf;
+}
+
+static void test_compress_decompress(const char* label, const char* type,
+ compress_t compress, decompress_t decompress) {
+ usec_t n, n2 = 0;
+ float dt;
+
+ _cleanup_free_ char *text, *buf;
+ _cleanup_free_ void *buf2 = NULL;
+ size_t buf2_allocated = 0;
+ size_t skipped = 0, compressed = 0, total = 0;
+
+ text = make_buf(MAX_SIZE, type);
+ buf = calloc(MAX_SIZE + 1, 1);
+ assert_se(text && buf);
+
+ n = now(CLOCK_MONOTONIC);
+
+ for (size_t i = 0; i <= MAX_SIZE; i++) {
+ size_t j = 0, k = 0, size;
+ int r;
+
+ size = permute(i);
+ if (size == 0)
+ continue;
+
+ log_debug("%s %zu %zu", type, i, size);
+
+ memzero(buf, MIN(size + 1000, MAX_SIZE));
+
+ r = compress(text, size, buf, size, &j);
+ /* assume compression must be successful except for small or random inputs */
+ assert_se(r == 0 || (size < 2048 && r == -ENOBUFS) || streq(type, "random"));
+
+ /* check for overwrites */
+ assert_se(buf[size] == 0);
+ if (r != 0) {
+ skipped += size;
+ continue;
+ }
+
+ assert_se(j > 0);
+ if (j >= size)
+ log_error("%s \"compressed\" %zu -> %zu", label, size, j);
+
+ r = decompress(buf, j, &buf2, &buf2_allocated, &k, 0);
+ assert_se(r == 0);
+ assert_se(buf2_allocated >= k);
+ assert_se(k == size);
+
+ assert_se(memcmp(text, buf2, size) == 0);
+
+ total += size;
+ compressed += j;
+
+ n2 = now(CLOCK_MONOTONIC);
+ if (n2 - n > arg_duration)
+ break;
+ }
+
+ dt = (n2-n) / 1e6;
+
+ log_info("%s/%s: compressed & decompressed %zu bytes in %.2fs (%.2fMiB/s), "
+ "mean compresion %.2f%%, skipped %zu bytes",
+ label, type, total, dt,
+ total / 1024. / 1024 / dt,
+ 100 - compressed * 100. / total,
+ skipped);
+}
+
+int main(int argc, char *argv[]) {
+ const char *i;
+
+ log_set_max_level(LOG_INFO);
+
+ if (argc >= 2) {
+ unsigned x;
+
+ assert_se(safe_atou(argv[1], &x) >= 0);
+ arg_duration = x * USEC_PER_SEC;
+ }
+ if (argc == 3)
+ (void) safe_atozu(argv[2], &arg_start);
+ else
+ arg_start = getpid();
+
+ NULSTR_FOREACH(i, "zeros\0simple\0random\0") {
+#ifdef HAVE_XZ
+ test_compress_decompress("XZ", i, compress_blob_xz, decompress_blob_xz);
+#endif
+#ifdef HAVE_LZ4
+ test_compress_decompress("LZ4", i, compress_blob_lz4, decompress_blob_lz4);
+#endif
+ }
+ return 0;
+}
diff --git a/src/grp-journal/libjournal-core/test-compress.c b/src/grp-journal/libjournal-core/test-compress.c
new file mode 100644
index 0000000000..68c9a4d76c
--- /dev/null
+++ b/src/grp-journal/libjournal-core/test-compress.c
@@ -0,0 +1,308 @@
+/***
+ This file is part of systemd
+
+ Copyright 2014 Ronny Chevalier
+
+ systemd is free software; you can redistribute it and/or modify it
+ under the terms of the GNU Lesser General Public License as published by
+ the Free Software Foundation; either version 2.1 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
+ Lesser General Public License for more details.
+
+ You should have received a copy of the GNU Lesser General Public License
+ along with systemd; If not, see <http://www.gnu.org/licenses/>.
+***/
+
+#ifdef HAVE_LZ4
+#include <lz4.h>
+#endif
+
+#include "alloc-util.h"
+#include "compress.h"
+#include "fd-util.h"
+#include "fileio.h"
+#include "macro.h"
+#include "random-util.h"
+#include "util.h"
+
+#ifdef HAVE_XZ
+# define XZ_OK 0
+#else
+# define XZ_OK -EPROTONOSUPPORT
+#endif
+
+#ifdef HAVE_LZ4
+# define LZ4_OK 0
+#else
+# define LZ4_OK -EPROTONOSUPPORT
+#endif
+
+typedef int (compress_blob_t)(const void *src, uint64_t src_size,
+ void *dst, size_t dst_alloc_size, size_t *dst_size);
+typedef int (decompress_blob_t)(const void *src, uint64_t src_size,
+ void **dst, size_t *dst_alloc_size,
+ size_t* dst_size, size_t dst_max);
+typedef int (decompress_sw_t)(const void *src, uint64_t src_size,
+ void **buffer, size_t *buffer_size,
+ const void *prefix, size_t prefix_len,
+ uint8_t extra);
+
+typedef int (compress_stream_t)(int fdf, int fdt, uint64_t max_bytes);
+typedef int (decompress_stream_t)(int fdf, int fdt, uint64_t max_size);
+
+static void test_compress_decompress(int compression,
+ compress_blob_t compress,
+ decompress_blob_t decompress,
+ const char *data,
+ size_t data_len,
+ bool may_fail) {
+ char compressed[512];
+ size_t csize, usize = 0;
+ _cleanup_free_ char *decompressed = NULL;
+ int r;
+
+ log_info("/* testing %s %s blob compression/decompression */",
+ object_compressed_to_string(compression), data);
+
+ r = compress(data, data_len, compressed, sizeof(compressed), &csize);
+ if (r == -ENOBUFS) {
+ log_info_errno(r, "compression failed: %m");
+ assert_se(may_fail);
+ } else {
+ assert_se(r == 0);
+ r = decompress(compressed, csize,
+ (void **) &decompressed, &usize, &csize, 0);
+ assert_se(r == 0);
+ assert_se(decompressed);
+ assert_se(memcmp(decompressed, data, data_len) == 0);
+ }
+
+ r = decompress("garbage", 7,
+ (void **) &decompressed, &usize, &csize, 0);
+ assert_se(r < 0);
+
+ /* make sure to have the minimal lz4 compressed size */
+ r = decompress("00000000\1g", 9,
+ (void **) &decompressed, &usize, &csize, 0);
+ assert_se(r < 0);
+
+ r = decompress("\100000000g", 9,
+ (void **) &decompressed, &usize, &csize, 0);
+ assert_se(r < 0);
+
+ memzero(decompressed, usize);
+}
+
+static void test_decompress_startswith(int compression,
+ compress_blob_t compress,
+ decompress_sw_t decompress_sw,
+ const char *data,
+ size_t data_len,
+ bool may_fail) {
+
+ char *compressed;
+ _cleanup_free_ char *compressed1 = NULL, *compressed2 = NULL, *decompressed = NULL;
+ size_t csize, usize = 0, len;
+ int r;
+
+ log_info("/* testing decompress_startswith with %s on %.20s text*/",
+ object_compressed_to_string(compression), data);
+
+#define BUFSIZE_1 512
+#define BUFSIZE_2 20000
+
+ compressed = compressed1 = malloc(BUFSIZE_1);
+ assert_se(compressed1);
+ r = compress(data, data_len, compressed, BUFSIZE_1, &csize);
+ if (r == -ENOBUFS) {
+ log_info_errno(r, "compression failed: %m");
+ assert_se(may_fail);
+
+ compressed = compressed2 = malloc(BUFSIZE_2);
+ assert_se(compressed2);
+ r = compress(data, data_len, compressed, BUFSIZE_2, &csize);
+ assert(r == 0);
+ }
+ assert_se(r == 0);
+
+ len = strlen(data);
+
+ r = decompress_sw(compressed, csize, (void **) &decompressed, &usize, data, len, '\0');
+ assert_se(r > 0);
+ r = decompress_sw(compressed, csize, (void **) &decompressed, &usize, data, len, 'w');
+ assert_se(r == 0);
+ r = decompress_sw(compressed, csize, (void **) &decompressed, &usize, "barbarbar", 9, ' ');
+ assert_se(r == 0);
+ r = decompress_sw(compressed, csize, (void **) &decompressed, &usize, data, len - 1, data[len-1]);
+ assert_se(r > 0);
+ r = decompress_sw(compressed, csize, (void **) &decompressed, &usize, data, len - 1, 'w');
+ assert_se(r == 0);
+ r = decompress_sw(compressed, csize, (void **) &decompressed, &usize, data, len, '\0');
+ assert_se(r > 0);
+}
+
+static void test_compress_stream(int compression,
+ const char* cat,
+ compress_stream_t compress,
+ decompress_stream_t decompress,
+ const char *srcfile) {
+
+ _cleanup_close_ int src = -1, dst = -1, dst2 = -1;
+ char pattern[] = "/tmp/systemd-test.compressed.XXXXXX",
+ pattern2[] = "/tmp/systemd-test.compressed.XXXXXX";
+ int r;
+ _cleanup_free_ char *cmd = NULL, *cmd2;
+ struct stat st = {};
+
+ log_debug("/* testing %s compression */",
+ object_compressed_to_string(compression));
+
+ log_debug("/* create source from %s */", srcfile);
+
+ assert_se((src = open(srcfile, O_RDONLY|O_CLOEXEC)) >= 0);
+
+ log_debug("/* test compression */");
+
+ assert_se((dst = mkostemp_safe(pattern, O_RDWR|O_CLOEXEC)) >= 0);
+
+ assert_se(compress(src, dst, -1) == 0);
+
+ if (cat) {
+ assert_se(asprintf(&cmd, "%s %s | diff %s -", cat, pattern, srcfile) > 0);
+ assert_se(system(cmd) == 0);
+ }
+
+ log_debug("/* test decompression */");
+
+ assert_se((dst2 = mkostemp_safe(pattern2, O_RDWR|O_CLOEXEC)) >= 0);
+
+ assert_se(stat(srcfile, &st) == 0);
+
+ assert_se(lseek(dst, 0, SEEK_SET) == 0);
+ r = decompress(dst, dst2, st.st_size);
+ assert_se(r == 0);
+
+ assert_se(asprintf(&cmd2, "diff %s %s", srcfile, pattern2) > 0);
+ assert_se(system(cmd2) == 0);
+
+ log_debug("/* test faulty decompression */");
+
+ assert_se(lseek(dst, 1, SEEK_SET) == 1);
+ r = decompress(dst, dst2, st.st_size);
+ assert_se(r == -EBADMSG || r == 0);
+
+ assert_se(lseek(dst, 0, SEEK_SET) == 0);
+ assert_se(lseek(dst2, 0, SEEK_SET) == 0);
+ r = decompress(dst, dst2, st.st_size - 1);
+ assert_se(r == -EFBIG);
+
+ assert_se(unlink(pattern) == 0);
+ assert_se(unlink(pattern2) == 0);
+}
+
+#ifdef HAVE_LZ4
+static void test_lz4_decompress_partial(void) {
+ char buf[20000];
+ size_t buf_size = sizeof(buf), compressed;
+ int r;
+ _cleanup_free_ char *huge = NULL;
+
+#define HUGE_SIZE (4096*1024)
+ huge = malloc(HUGE_SIZE);
+ memset(huge, 'x', HUGE_SIZE);
+ memcpy(huge, "HUGE=", 5);
+
+ r = LZ4_compress_limitedOutput(huge, buf, HUGE_SIZE, buf_size);
+ assert_se(r >= 0);
+ compressed = r;
+ log_info("Compressed %i → %zu", HUGE_SIZE, compressed);
+
+ r = LZ4_decompress_safe(buf, huge, r, HUGE_SIZE);
+ assert_se(r >= 0);
+ log_info("Decompressed → %i", r);
+
+ r = LZ4_decompress_safe_partial(buf, huge,
+ compressed,
+ 12, HUGE_SIZE);
+ assert_se(r >= 0);
+ log_info("Decompressed partial %i/%i → %i", 12, HUGE_SIZE, r);
+
+ /* We expect this to fail, because that's how current lz4 works. If this
+ * call succeeds, then lz4 has been fixed, and we need to change our code.
+ */
+ r = LZ4_decompress_safe_partial(buf, huge,
+ compressed,
+ 12, HUGE_SIZE-1);
+ assert_se(r < 0);
+ log_info("Decompressed partial %i/%i → %i", 12, HUGE_SIZE-1, r);
+}
+#endif
+
+int main(int argc, char *argv[]) {
+ const char text[] =
+ "text\0foofoofoofoo AAAA aaaaaaaaa ghost busters barbarbar FFF"
+ "foofoofoofoo AAAA aaaaaaaaa ghost busters barbarbar FFF";
+
+ char data[512] = "random\0";
+
+ char huge[4096*1024];
+ memset(huge, 'x', sizeof(huge));
+ memcpy(huge, "HUGE=", 5);
+ char_array_0(huge);
+
+ log_set_max_level(LOG_DEBUG);
+
+ random_bytes(data + 7, sizeof(data) - 7);
+
+#ifdef HAVE_XZ
+ test_compress_decompress(OBJECT_COMPRESSED_XZ, compress_blob_xz, decompress_blob_xz,
+ text, sizeof(text), false);
+ test_compress_decompress(OBJECT_COMPRESSED_XZ, compress_blob_xz, decompress_blob_xz,
+ data, sizeof(data), true);
+
+ test_decompress_startswith(OBJECT_COMPRESSED_XZ,
+ compress_blob_xz, decompress_startswith_xz,
+ text, sizeof(text), false);
+ test_decompress_startswith(OBJECT_COMPRESSED_XZ,
+ compress_blob_xz, decompress_startswith_xz,
+ data, sizeof(data), true);
+ test_decompress_startswith(OBJECT_COMPRESSED_XZ,
+ compress_blob_xz, decompress_startswith_xz,
+ huge, sizeof(huge), true);
+
+ test_compress_stream(OBJECT_COMPRESSED_XZ, "xzcat",
+ compress_stream_xz, decompress_stream_xz, argv[0]);
+#else
+ log_info("/* XZ test skipped */");
+#endif
+
+#ifdef HAVE_LZ4
+ test_compress_decompress(OBJECT_COMPRESSED_LZ4, compress_blob_lz4, decompress_blob_lz4,
+ text, sizeof(text), false);
+ test_compress_decompress(OBJECT_COMPRESSED_LZ4, compress_blob_lz4, decompress_blob_lz4,
+ data, sizeof(data), true);
+
+ test_decompress_startswith(OBJECT_COMPRESSED_LZ4,
+ compress_blob_lz4, decompress_startswith_lz4,
+ text, sizeof(text), false);
+ test_decompress_startswith(OBJECT_COMPRESSED_LZ4,
+ compress_blob_lz4, decompress_startswith_lz4,
+ data, sizeof(data), true);
+ test_decompress_startswith(OBJECT_COMPRESSED_LZ4,
+ compress_blob_lz4, decompress_startswith_lz4,
+ huge, sizeof(huge), true);
+
+ test_compress_stream(OBJECT_COMPRESSED_LZ4, "lz4cat",
+ compress_stream_lz4, decompress_stream_lz4, argv[0]);
+
+ test_lz4_decompress_partial();
+#else
+ log_info("/* LZ4 test skipped */");
+#endif
+
+ return 0;
+}
diff --git a/src/grp-journal/libjournal-core/test-journal-enum.c b/src/grp-journal/libjournal-core/test-journal-enum.c
new file mode 100644
index 0000000000..588e12c800
--- /dev/null
+++ b/src/grp-journal/libjournal-core/test-journal-enum.c
@@ -0,0 +1,53 @@
+/***
+ This file is part of systemd.
+
+ Copyright 2012 Lennart Poettering
+
+ systemd is free software; you can redistribute it and/or modify it
+ under the terms of the GNU Lesser General Public License as published by
+ the Free Software Foundation; either version 2.1 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
+ Lesser General Public License for more details.
+
+ You should have received a copy of the GNU Lesser General Public License
+ along with systemd; If not, see <http://www.gnu.org/licenses/>.
+***/
+
+#include <stdio.h>
+
+#include <systemd/sd-journal.h>
+
+#include "basic/log.h"
+#include "basic/macro.h"
+#include "sd-journal/journal-internal.h"
+
+int main(int argc, char *argv[]) {
+ unsigned n = 0;
+ _cleanup_(sd_journal_closep) sd_journal*j = NULL;
+
+ log_set_max_level(LOG_DEBUG);
+
+ assert_se(sd_journal_open(&j, SD_JOURNAL_LOCAL_ONLY) >= 0);
+
+ assert_se(sd_journal_add_match(j, "_TRANSPORT=syslog", 0) >= 0);
+ assert_se(sd_journal_add_match(j, "_UID=0", 0) >= 0);
+
+ SD_JOURNAL_FOREACH_BACKWARDS(j) {
+ const void *d;
+ size_t l;
+
+ assert_se(sd_journal_get_data(j, "MESSAGE", &d, &l) >= 0);
+
+ printf("%.*s\n", (int) l, (char*) d);
+
+ n++;
+ if (n >= 10)
+ break;
+ }
+
+ return 0;
+}
diff --git a/src/grp-journal/libjournal-core/test-journal-flush.c b/src/grp-journal/libjournal-core/test-journal-flush.c
new file mode 100644
index 0000000000..66ae8f83c4
--- /dev/null
+++ b/src/grp-journal/libjournal-core/test-journal-flush.c
@@ -0,0 +1,75 @@
+/***
+ This file is part of systemd.
+
+ Copyright 2013 Lennart Poettering
+
+ systemd is free software; you can redistribute it and/or modify it
+ under the terms of the GNU Lesser General Public License as published by
+ the Free Software Foundation; either version 2.1 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
+ Lesser General Public License for more details.
+
+ You should have received a copy of the GNU Lesser General Public License
+ along with systemd; If not, see <http://www.gnu.org/licenses/>.
+***/
+
+#include <fcntl.h>
+
+#include <systemd/sd-journal.h>
+
+#include "basic/alloc-util.h"
+#include "basic/macro.h"
+#include "basic/string-util.h"
+#include "sd-journal/journal-file.h"
+#include "sd-journal/journal-internal.h"
+
+int main(int argc, char *argv[]) {
+ _cleanup_free_ char *fn = NULL;
+ char dn[] = "/var/tmp/test-journal-flush.XXXXXX";
+ JournalFile *new_journal = NULL;
+ sd_journal *j = NULL;
+ unsigned n = 0;
+ int r;
+
+ assert_se(mkdtemp(dn));
+ fn = strappend(dn, "/test.journal");
+
+ r = journal_file_open(-1, fn, O_CREAT|O_RDWR, 0644, false, false, NULL, NULL, NULL, NULL, &new_journal);
+ assert_se(r >= 0);
+
+ r = sd_journal_open(&j, 0);
+ assert_se(r >= 0);
+
+ sd_journal_set_data_threshold(j, 0);
+
+ SD_JOURNAL_FOREACH(j) {
+ Object *o;
+ JournalFile *f;
+
+ f = j->current_file;
+ assert_se(f && f->current_offset > 0);
+
+ r = journal_file_move_to_object(f, OBJECT_ENTRY, f->current_offset, &o);
+ assert_se(r >= 0);
+
+ r = journal_file_copy_entry(f, new_journal, o, f->current_offset, NULL, NULL, NULL);
+ assert_se(r >= 0);
+
+ n++;
+ if (n > 10000)
+ break;
+ }
+
+ sd_journal_close(j);
+
+ (void) journal_file_close(new_journal);
+
+ unlink(fn);
+ assert_se(rmdir(dn) == 0);
+
+ return 0;
+}
diff --git a/src/grp-journal/libjournal-core/test-journal-init.c b/src/grp-journal/libjournal-core/test-journal-init.c
new file mode 100644
index 0000000000..c5f1d345e9
--- /dev/null
+++ b/src/grp-journal/libjournal-core/test-journal-init.c
@@ -0,0 +1,64 @@
+/***
+ This file is part of systemd.
+
+ Copyright 2013 Zbigniew Jędrzejewski-Szmek
+
+ systemd is free software; you can redistribute it and/or modify it
+ under the terms of the GNU Lesser General Public License as published by
+ the Free Software Foundation; either version 2.1 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
+ Lesser General Public License for more details.
+
+ You should have received a copy of the GNU Lesser General Public License
+ along with systemd; If not, see <http://www.gnu.org/licenses/>.
+***/
+
+#include <systemd/sd-journal.h>
+
+#include "basic/log.h"
+#include "basic/parse-util.h"
+#include "basic/rm-rf.h"
+#include "basic/util.h"
+
+int main(int argc, char *argv[]) {
+ sd_journal *j;
+ int r, i, I = 100;
+ char t[] = "/tmp/journal-stream-XXXXXX";
+
+ log_set_max_level(LOG_DEBUG);
+
+ if (argc >= 2) {
+ r = safe_atoi(argv[1], &I);
+ if (r < 0)
+ log_info("Could not parse loop count argument. Using default.");
+ }
+
+ log_info("Running %d loops", I);
+
+ assert_se(mkdtemp(t));
+
+ for (i = 0; i < I; i++) {
+ r = sd_journal_open(&j, SD_JOURNAL_LOCAL_ONLY);
+ assert_se(r == 0);
+
+ sd_journal_close(j);
+
+ r = sd_journal_open_directory(&j, t, 0);
+ assert_se(r == 0);
+
+ sd_journal_close(j);
+
+ j = NULL;
+ r = sd_journal_open_directory(&j, t, SD_JOURNAL_LOCAL_ONLY);
+ assert_se(r == -EINVAL);
+ assert_se(j == NULL);
+ }
+
+ assert_se(rm_rf(t, REMOVE_ROOT|REMOVE_PHYSICAL) >= 0);
+
+ return 0;
+}
diff --git a/src/grp-journal/libjournal-core/test-journal-interleaving.c b/src/grp-journal/libjournal-core/test-journal-interleaving.c
new file mode 100644
index 0000000000..004394d18a
--- /dev/null
+++ b/src/grp-journal/libjournal-core/test-journal-interleaving.c
@@ -0,0 +1,307 @@
+/***
+ This file is part of systemd.
+
+ Copyright 2013 Marius Vollmer
+ Copyright 2013 Zbigniew Jędrzejewski-Szmek
+
+ systemd is free software; you can redistribute it and/or modify it
+ under the terms of the GNU Lesser General Public License as published by
+ the Free Software Foundation; either version 2.1 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
+ Lesser General Public License for more details.
+
+ You should have received a copy of the GNU Lesser General Public License
+ along with systemd; If not, see <http://www.gnu.org/licenses/>.
+***/
+
+#include <fcntl.h>
+#include <unistd.h>
+
+#include <systemd/sd-journal.h>
+
+#include "basic/alloc-util.h"
+#include "basic/log.h"
+#include "basic/parse-util.h"
+#include "basic/rm-rf.h"
+#include "basic/util.h"
+#include "sd-journal/journal-file.h"
+#include "sd-journal/journal-vacuum.h"
+
+/* This program tests skipping around in a multi-file journal.
+ */
+
+static bool arg_keep = false;
+
+noreturn static void log_assert_errno(const char *text, int eno, const char *file, int line, const char *func) {
+ log_internal(LOG_CRIT, 0, file, line, func,
+ "'%s' failed at %s:%u (%s): %s.",
+ text, file, line, func, strerror(eno));
+ abort();
+}
+
+#define assert_ret(expr) \
+ do { \
+ int _r_ = (expr); \
+ if (_unlikely_(_r_ < 0)) \
+ log_assert_errno(#expr, -_r_, __FILE__, __LINE__, __PRETTY_FUNCTION__); \
+ } while (false)
+
+static JournalFile *test_open(const char *name) {
+ JournalFile *f;
+ assert_ret(journal_file_open(-1, name, O_RDWR|O_CREAT, 0644, true, false, NULL, NULL, NULL, NULL, &f));
+ return f;
+}
+
+static void test_close(JournalFile *f) {
+ (void) journal_file_close (f);
+}
+
+static void append_number(JournalFile *f, int n, uint64_t *seqnum) {
+ char *p;
+ dual_timestamp ts;
+ static dual_timestamp previous_ts = {};
+ struct iovec iovec[1];
+
+ dual_timestamp_get(&ts);
+
+ if (ts.monotonic <= previous_ts.monotonic)
+ ts.monotonic = previous_ts.monotonic + 1;
+
+ if (ts.realtime <= previous_ts.realtime)
+ ts.realtime = previous_ts.realtime + 1;
+
+ previous_ts = ts;
+
+ assert_se(asprintf(&p, "NUMBER=%d", n) >= 0);
+ iovec[0].iov_base = p;
+ iovec[0].iov_len = strlen(p);
+ assert_ret(journal_file_append_entry(f, &ts, iovec, 1, seqnum, NULL, NULL));
+ free(p);
+}
+
+static void test_check_number (sd_journal *j, int n) {
+ const void *d;
+ _cleanup_free_ char *k;
+ size_t l;
+ int x;
+
+ assert_ret(sd_journal_get_data(j, "NUMBER", &d, &l));
+ assert_se(k = strndup(d, l));
+ printf("%s\n", k);
+
+ assert_se(safe_atoi(k + 7, &x) >= 0);
+ assert_se(n == x);
+}
+
+static void test_check_numbers_down (sd_journal *j, int count) {
+ int i;
+
+ for (i = 1; i <= count; i++) {
+ int r;
+ test_check_number(j, i);
+ assert_ret(r = sd_journal_next(j));
+ if (i == count)
+ assert_se(r == 0);
+ else
+ assert_se(r == 1);
+ }
+
+}
+
+static void test_check_numbers_up (sd_journal *j, int count) {
+ for (int i = count; i >= 1; i--) {
+ int r;
+ test_check_number(j, i);
+ assert_ret(r = sd_journal_previous(j));
+ if (i == 1)
+ assert_se(r == 0);
+ else
+ assert_se(r == 1);
+ }
+
+}
+
+static void setup_sequential(void) {
+ JournalFile *one, *two;
+ one = test_open("one.journal");
+ two = test_open("two.journal");
+ append_number(one, 1, NULL);
+ append_number(one, 2, NULL);
+ append_number(two, 3, NULL);
+ append_number(two, 4, NULL);
+ test_close(one);
+ test_close(two);
+}
+
+static void setup_interleaved(void) {
+ JournalFile *one, *two;
+ one = test_open("one.journal");
+ two = test_open("two.journal");
+ append_number(one, 1, NULL);
+ append_number(two, 2, NULL);
+ append_number(one, 3, NULL);
+ append_number(two, 4, NULL);
+ test_close(one);
+ test_close(two);
+}
+
+static void test_skip(void (*setup)(void)) {
+ char t[] = "/tmp/journal-skip-XXXXXX";
+ sd_journal *j;
+ int r;
+
+ assert_se(mkdtemp(t));
+ assert_se(chdir(t) >= 0);
+
+ setup();
+
+ /* Seek to head, iterate down.
+ */
+ assert_ret(sd_journal_open_directory(&j, t, 0));
+ assert_ret(sd_journal_seek_head(j));
+ assert_ret(sd_journal_next(j));
+ test_check_numbers_down(j, 4);
+ sd_journal_close(j);
+
+ /* Seek to tail, iterate up.
+ */
+ assert_ret(sd_journal_open_directory(&j, t, 0));
+ assert_ret(sd_journal_seek_tail(j));
+ assert_ret(sd_journal_previous(j));
+ test_check_numbers_up(j, 4);
+ sd_journal_close(j);
+
+ /* Seek to tail, skip to head, iterate down.
+ */
+ assert_ret(sd_journal_open_directory(&j, t, 0));
+ assert_ret(sd_journal_seek_tail(j));
+ assert_ret(r = sd_journal_previous_skip(j, 4));
+ assert_se(r == 4);
+ test_check_numbers_down(j, 4);
+ sd_journal_close(j);
+
+ /* Seek to head, skip to tail, iterate up.
+ */
+ assert_ret(sd_journal_open_directory(&j, t, 0));
+ assert_ret(sd_journal_seek_head(j));
+ assert_ret(r = sd_journal_next_skip(j, 4));
+ assert_se(r == 4);
+ test_check_numbers_up(j, 4);
+ sd_journal_close(j);
+
+ log_info("Done...");
+
+ if (arg_keep)
+ log_info("Not removing %s", t);
+ else {
+ journal_directory_vacuum(".", 3000000, 0, 0, NULL, true);
+
+ assert_se(rm_rf(t, REMOVE_ROOT|REMOVE_PHYSICAL) >= 0);
+ }
+
+ puts("------------------------------------------------------------");
+}
+
+static void test_sequence_numbers(void) {
+
+ char t[] = "/tmp/journal-seq-XXXXXX";
+ JournalFile *one, *two;
+ uint64_t seqnum = 0;
+ sd_id128_t seqnum_id;
+
+ assert_se(mkdtemp(t));
+ assert_se(chdir(t) >= 0);
+
+ assert_se(journal_file_open(-1, "one.journal", O_RDWR|O_CREAT, 0644,
+ true, false, NULL, NULL, NULL, NULL, &one) == 0);
+
+ append_number(one, 1, &seqnum);
+ printf("seqnum=%"PRIu64"\n", seqnum);
+ assert_se(seqnum == 1);
+ append_number(one, 2, &seqnum);
+ printf("seqnum=%"PRIu64"\n", seqnum);
+ assert_se(seqnum == 2);
+
+ assert_se(one->header->state == STATE_ONLINE);
+ assert_se(!sd_id128_equal(one->header->file_id, one->header->machine_id));
+ assert_se(!sd_id128_equal(one->header->file_id, one->header->boot_id));
+ assert_se(sd_id128_equal(one->header->file_id, one->header->seqnum_id));
+
+ memcpy(&seqnum_id, &one->header->seqnum_id, sizeof(sd_id128_t));
+
+ assert_se(journal_file_open(-1, "two.journal", O_RDWR|O_CREAT, 0644,
+ true, false, NULL, NULL, NULL, one, &two) == 0);
+
+ assert_se(two->header->state == STATE_ONLINE);
+ assert_se(!sd_id128_equal(two->header->file_id, one->header->file_id));
+ assert_se(sd_id128_equal(one->header->machine_id, one->header->machine_id));
+ assert_se(sd_id128_equal(one->header->boot_id, one->header->boot_id));
+ assert_se(sd_id128_equal(one->header->seqnum_id, one->header->seqnum_id));
+
+ append_number(two, 3, &seqnum);
+ printf("seqnum=%"PRIu64"\n", seqnum);
+ assert_se(seqnum == 3);
+ append_number(two, 4, &seqnum);
+ printf("seqnum=%"PRIu64"\n", seqnum);
+ assert_se(seqnum == 4);
+
+ test_close(two);
+
+ append_number(one, 5, &seqnum);
+ printf("seqnum=%"PRIu64"\n", seqnum);
+ assert_se(seqnum == 5);
+
+ append_number(one, 6, &seqnum);
+ printf("seqnum=%"PRIu64"\n", seqnum);
+ assert_se(seqnum == 6);
+
+ test_close(one);
+
+ /* restart server */
+ seqnum = 0;
+
+ assert_se(journal_file_open(-1, "two.journal", O_RDWR, 0,
+ true, false, NULL, NULL, NULL, NULL, &two) == 0);
+
+ assert_se(sd_id128_equal(two->header->seqnum_id, seqnum_id));
+
+ append_number(two, 7, &seqnum);
+ printf("seqnum=%"PRIu64"\n", seqnum);
+ assert_se(seqnum == 5);
+
+ /* So..., here we have the same seqnum in two files with the
+ * same seqnum_id. */
+
+ test_close(two);
+
+ log_info("Done...");
+
+ if (arg_keep)
+ log_info("Not removing %s", t);
+ else {
+ journal_directory_vacuum(".", 3000000, 0, 0, NULL, true);
+
+ assert_se(rm_rf(t, REMOVE_ROOT|REMOVE_PHYSICAL) >= 0);
+ }
+}
+
+int main(int argc, char *argv[]) {
+ log_set_max_level(LOG_DEBUG);
+
+ /* journal_file_open requires a valid machine id */
+ if (access("/etc/machine-id", F_OK) != 0)
+ return EXIT_TEST_SKIP;
+
+ arg_keep = argc > 1;
+
+ test_skip(setup_sequential);
+ test_skip(setup_interleaved);
+
+ test_sequence_numbers();
+
+ return 0;
+}
diff --git a/src/grp-journal/libjournal-core/test-journal-match.c b/src/grp-journal/libjournal-core/test-journal-match.c
new file mode 100644
index 0000000000..967e30c46a
--- /dev/null
+++ b/src/grp-journal/libjournal-core/test-journal-match.c
@@ -0,0 +1,76 @@
+/***
+ This file is part of systemd.
+
+ Copyright 2012 Lennart Poettering
+
+ systemd is free software; you can redistribute it and/or modify it
+ under the terms of the GNU Lesser General Public License as published by
+ the Free Software Foundation; either version 2.1 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
+ Lesser General Public License for more details.
+
+ You should have received a copy of the GNU Lesser General Public License
+ along with systemd; If not, see <http://www.gnu.org/licenses/>.
+***/
+
+#include <stdio.h>
+
+#include <systemd/sd-journal.h>
+
+#include "basic/alloc-util.h"
+#include "basic/log.h"
+#include "basic/string-util.h"
+#include "basic/util.h"
+#include "sd-journal/journal-internal.h"
+
+int main(int argc, char *argv[]) {
+ _cleanup_(sd_journal_closep) sd_journal*j = NULL;
+ _cleanup_free_ char *t;
+
+ log_set_max_level(LOG_DEBUG);
+
+ assert_se(sd_journal_open(&j, 0) >= 0);
+
+ assert_se(sd_journal_add_match(j, "foobar", 0) < 0);
+ assert_se(sd_journal_add_match(j, "foobar=waldo", 0) < 0);
+ assert_se(sd_journal_add_match(j, "", 0) < 0);
+ assert_se(sd_journal_add_match(j, "=", 0) < 0);
+ assert_se(sd_journal_add_match(j, "=xxxxx", 0) < 0);
+ assert_se(sd_journal_add_match(j, "HALLO=WALDO", 0) >= 0);
+ assert_se(sd_journal_add_match(j, "QUUX=mmmm", 0) >= 0);
+ assert_se(sd_journal_add_match(j, "QUUX=xxxxx", 0) >= 0);
+ assert_se(sd_journal_add_match(j, "HALLO=", 0) >= 0);
+ assert_se(sd_journal_add_match(j, "QUUX=xxxxx", 0) >= 0);
+ assert_se(sd_journal_add_match(j, "QUUX=yyyyy", 0) >= 0);
+ assert_se(sd_journal_add_match(j, "PIFF=paff", 0) >= 0);
+
+ assert_se(sd_journal_add_disjunction(j) >= 0);
+
+ assert_se(sd_journal_add_match(j, "ONE=one", 0) >= 0);
+ assert_se(sd_journal_add_match(j, "ONE=two", 0) >= 0);
+ assert_se(sd_journal_add_match(j, "TWO=two", 0) >= 0);
+
+ assert_se(sd_journal_add_conjunction(j) >= 0);
+
+ assert_se(sd_journal_add_match(j, "L4_1=yes", 0) >= 0);
+ assert_se(sd_journal_add_match(j, "L4_1=ok", 0) >= 0);
+ assert_se(sd_journal_add_match(j, "L4_2=yes", 0) >= 0);
+ assert_se(sd_journal_add_match(j, "L4_2=ok", 0) >= 0);
+
+ assert_se(sd_journal_add_disjunction(j) >= 0);
+
+ assert_se(sd_journal_add_match(j, "L3=yes", 0) >= 0);
+ assert_se(sd_journal_add_match(j, "L3=ok", 0) >= 0);
+
+ assert_se(t = journal_make_match_string(j));
+
+ printf("resulting match expression is: %s\n", t);
+
+ assert_se(streq(t, "(((L3=ok OR L3=yes) OR ((L4_2=ok OR L4_2=yes) AND (L4_1=ok OR L4_1=yes))) AND ((TWO=two AND (ONE=two OR ONE=one)) OR (PIFF=paff AND (QUUX=yyyyy OR QUUX=xxxxx OR QUUX=mmmm) AND (HALLO= OR HALLO=WALDO))))"));
+
+ return 0;
+}
diff --git a/src/grp-journal/libjournal-core/test-journal-send.c b/src/grp-journal/libjournal-core/test-journal-send.c
new file mode 100644
index 0000000000..192e12c887
--- /dev/null
+++ b/src/grp-journal/libjournal-core/test-journal-send.c
@@ -0,0 +1,102 @@
+/***
+ 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 Lesser General Public License as published by
+ the Free Software Foundation; either version 2.1 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
+ Lesser General Public License for more details.
+
+ You should have received a copy of the GNU Lesser General Public License
+ along with systemd; If not, see <http://www.gnu.org/licenses/>.
+***/
+
+#include <errno.h>
+#include <stdlib.h>
+#include <unistd.h>
+
+#include <systemd/sd-journal.h>
+
+#include "basic/macro.h"
+
+int main(int argc, char *argv[]) {
+ char huge[4096*1024];
+
+ /* utf-8 and non-utf-8, message-less and message-ful iovecs */
+ struct iovec graph1[] = {
+ {(char*) "GRAPH=graph", strlen("GRAPH=graph")}
+ };
+ struct iovec graph2[] = {
+ {(char*) "GRAPH=graph\n", strlen("GRAPH=graph\n")}
+ };
+ struct iovec message1[] = {
+ {(char*) "MESSAGE=graph", strlen("MESSAGE=graph")}
+ };
+ struct iovec message2[] = {
+ {(char*) "MESSAGE=graph\n", strlen("MESSAGE=graph\n")}
+ };
+
+ assert_se(sd_journal_print(LOG_INFO, "piepapo") == 0);
+
+ assert_se(sd_journal_send("MESSAGE=foobar",
+ "VALUE=%i", 7,
+ NULL) == 0);
+
+ errno = ENOENT;
+ assert_se(sd_journal_perror("Foobar") == 0);
+
+ assert_se(sd_journal_perror("") == 0);
+
+ memset(huge, 'x', sizeof(huge));
+ memcpy(huge, "HUGE=", 5);
+ char_array_0(huge);
+
+ assert_se(sd_journal_send("MESSAGE=Huge field attached",
+ huge,
+ NULL) == 0);
+
+ assert_se(sd_journal_send("MESSAGE=uiui",
+ "VALUE=A",
+ "VALUE=B",
+ "VALUE=C",
+ "SINGLETON=1",
+ "OTHERVALUE=X",
+ "OTHERVALUE=Y",
+ "WITH_BINARY=this is a binary value \a",
+ NULL) == 0);
+
+ syslog(LOG_NOTICE, "Hello World!");
+
+ assert_se(sd_journal_print(LOG_NOTICE, "Hello World") == 0);
+
+ assert_se(sd_journal_send("MESSAGE=Hello World!",
+ "MESSAGE_ID=52fb62f99e2c49d89cfbf9d6de5e3555",
+ "PRIORITY=5",
+ "HOME=%s", getenv("HOME"),
+ "TERM=%s", getenv("TERM"),
+ "PAGE_SIZE=%li", sysconf(_SC_PAGESIZE),
+ "N_CPUS=%li", sysconf(_SC_NPROCESSORS_ONLN),
+ NULL) == 0);
+
+ assert_se(sd_journal_sendv(graph1, 1) == 0);
+ assert_se(sd_journal_sendv(graph2, 1) == 0);
+ assert_se(sd_journal_sendv(message1, 1) == 0);
+ assert_se(sd_journal_sendv(message2, 1) == 0);
+
+ /* test without location fields */
+#undef sd_journal_sendv
+ assert_se(sd_journal_sendv(graph1, 1) == 0);
+ assert_se(sd_journal_sendv(graph2, 1) == 0);
+ assert_se(sd_journal_sendv(message1, 1) == 0);
+ assert_se(sd_journal_sendv(message2, 1) == 0);
+
+ sleep(1);
+
+ return 0;
+}
diff --git a/src/grp-journal/libjournal-core/test-journal-stream.c b/src/grp-journal/libjournal-core/test-journal-stream.c
new file mode 100644
index 0000000000..b71c44ef56
--- /dev/null
+++ b/src/grp-journal/libjournal-core/test-journal-stream.c
@@ -0,0 +1,196 @@
+/***
+ This file is part of systemd.
+
+ Copyright 2012 Lennart Poettering
+
+ systemd is free software; you can redistribute it and/or modify it
+ under the terms of the GNU Lesser General Public License as published by
+ the Free Software Foundation; either version 2.1 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
+ Lesser General Public License for more details.
+
+ You should have received a copy of the GNU Lesser General Public License
+ along with systemd; If not, see <http://www.gnu.org/licenses/>.
+***/
+
+#include <fcntl.h>
+#include <unistd.h>
+
+#include <systemd/sd-journal.h>
+
+#include "basic/alloc-util.h"
+#include "basic/log.h"
+#include "basic/macro.h"
+#include "basic/parse-util.h"
+#include "basic/rm-rf.h"
+#include "basic/util.h"
+#include "sd-journal/journal-file.h"
+#include "sd-journal/journal-internal.h"
+
+#define N_ENTRIES 200
+
+static void verify_contents(sd_journal *j, unsigned skip) {
+ unsigned i;
+
+ assert_se(j);
+
+ i = 0;
+ SD_JOURNAL_FOREACH(j) {
+ const void *d;
+ char *k, *c;
+ size_t l;
+ unsigned u = 0;
+
+ assert_se(sd_journal_get_cursor(j, &k) >= 0);
+ printf("cursor: %s\n", k);
+ free(k);
+
+ assert_se(sd_journal_get_data(j, "MAGIC", &d, &l) >= 0);
+ printf("\t%.*s\n", (int) l, (const char*) d);
+
+ assert_se(sd_journal_get_data(j, "NUMBER", &d, &l) >= 0);
+ assert_se(k = strndup(d, l));
+ printf("\t%s\n", k);
+
+ if (skip > 0) {
+ assert_se(safe_atou(k + 7, &u) >= 0);
+ assert_se(i == u);
+ i += skip;
+ }
+
+ free(k);
+
+ assert_se(sd_journal_get_cursor(j, &c) >= 0);
+ assert_se(sd_journal_test_cursor(j, c) > 0);
+ free(c);
+ }
+
+ if (skip > 0)
+ assert_se(i == N_ENTRIES);
+}
+
+int main(int argc, char *argv[]) {
+ JournalFile *one, *two, *three;
+ char t[] = "/tmp/journal-stream-XXXXXX";
+ unsigned i;
+ _cleanup_(sd_journal_closep) sd_journal *j = NULL;
+ char *z;
+ const void *data;
+ size_t l;
+ dual_timestamp previous_ts = DUAL_TIMESTAMP_NULL;
+
+ /* journal_file_open requires a valid machine id */
+ if (access("/etc/machine-id", F_OK) != 0)
+ return EXIT_TEST_SKIP;
+
+ log_set_max_level(LOG_DEBUG);
+
+ assert_se(mkdtemp(t));
+ assert_se(chdir(t) >= 0);
+
+ assert_se(journal_file_open(-1, "one.journal", O_RDWR|O_CREAT, 0666, true, false, NULL, NULL, NULL, NULL, &one) == 0);
+ assert_se(journal_file_open(-1, "two.journal", O_RDWR|O_CREAT, 0666, true, false, NULL, NULL, NULL, NULL, &two) == 0);
+ assert_se(journal_file_open(-1, "three.journal", O_RDWR|O_CREAT, 0666, true, false, NULL, NULL, NULL, NULL, &three) == 0);
+
+ for (i = 0; i < N_ENTRIES; i++) {
+ char *p, *q;
+ dual_timestamp ts;
+ struct iovec iovec[2];
+
+ dual_timestamp_get(&ts);
+
+ if (ts.monotonic <= previous_ts.monotonic)
+ ts.monotonic = previous_ts.monotonic + 1;
+
+ if (ts.realtime <= previous_ts.realtime)
+ ts.realtime = previous_ts.realtime + 1;
+
+ previous_ts = ts;
+
+ assert_se(asprintf(&p, "NUMBER=%u", i) >= 0);
+ iovec[0].iov_base = p;
+ iovec[0].iov_len = strlen(p);
+
+ assert_se(asprintf(&q, "MAGIC=%s", i % 5 == 0 ? "quux" : "waldo") >= 0);
+
+ iovec[1].iov_base = q;
+ iovec[1].iov_len = strlen(q);
+
+ if (i % 10 == 0)
+ assert_se(journal_file_append_entry(three, &ts, iovec, 2, NULL, NULL, NULL) == 0);
+ else {
+ if (i % 3 == 0)
+ assert_se(journal_file_append_entry(two, &ts, iovec, 2, NULL, NULL, NULL) == 0);
+
+ assert_se(journal_file_append_entry(one, &ts, iovec, 2, NULL, NULL, NULL) == 0);
+ }
+
+ free(p);
+ free(q);
+ }
+
+ (void) journal_file_close(one);
+ (void) journal_file_close(two);
+ (void) journal_file_close(three);
+
+ assert_se(sd_journal_open_directory(&j, t, 0) >= 0);
+
+ assert_se(sd_journal_add_match(j, "MAGIC=quux", 0) >= 0);
+ SD_JOURNAL_FOREACH_BACKWARDS(j) {
+ _cleanup_free_ char *c;
+
+ assert_se(sd_journal_get_data(j, "NUMBER", &data, &l) >= 0);
+ printf("\t%.*s\n", (int) l, (const char*) data);
+
+ assert_se(sd_journal_get_cursor(j, &c) >= 0);
+ assert_se(sd_journal_test_cursor(j, c) > 0);
+ }
+
+ SD_JOURNAL_FOREACH(j) {
+ _cleanup_free_ char *c;
+
+ assert_se(sd_journal_get_data(j, "NUMBER", &data, &l) >= 0);
+ printf("\t%.*s\n", (int) l, (const char*) data);
+
+ assert_se(sd_journal_get_cursor(j, &c) >= 0);
+ assert_se(sd_journal_test_cursor(j, c) > 0);
+ }
+
+ sd_journal_flush_matches(j);
+
+ verify_contents(j, 1);
+
+ printf("NEXT TEST\n");
+ assert_se(sd_journal_add_match(j, "MAGIC=quux", 0) >= 0);
+
+ assert_se(z = journal_make_match_string(j));
+ printf("resulting match expression is: %s\n", z);
+ free(z);
+
+ verify_contents(j, 5);
+
+ printf("NEXT TEST\n");
+ sd_journal_flush_matches(j);
+ assert_se(sd_journal_add_match(j, "MAGIC=waldo", 0) >= 0);
+ assert_se(sd_journal_add_match(j, "NUMBER=10", 0) >= 0);
+ assert_se(sd_journal_add_match(j, "NUMBER=11", 0) >= 0);
+ assert_se(sd_journal_add_match(j, "NUMBER=12", 0) >= 0);
+
+ assert_se(z = journal_make_match_string(j));
+ printf("resulting match expression is: %s\n", z);
+ free(z);
+
+ verify_contents(j, 0);
+
+ assert_se(sd_journal_query_unique(j, "NUMBER") >= 0);
+ SD_JOURNAL_FOREACH_UNIQUE(j, data, l)
+ printf("%.*s\n", (int) l, (const char*) data);
+
+ assert_se(rm_rf(t, REMOVE_ROOT|REMOVE_PHYSICAL) >= 0);
+
+ return 0;
+}
diff --git a/src/grp-journal/libjournal-core/test-journal-syslog.c b/src/grp-journal/libjournal-core/test-journal-syslog.c
new file mode 100644
index 0000000000..4756126478
--- /dev/null
+++ b/src/grp-journal/libjournal-core/test-journal-syslog.c
@@ -0,0 +1,45 @@
+/***
+ 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 Lesser General Public License as published by
+ the Free Software Foundation; either version 2.1 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
+ Lesser General Public License for more details.
+
+ You should have received a copy of the GNU Lesser General Public License
+ along with systemd; If not, see <http://www.gnu.org/licenses/>.
+***/
+
+#include "basic/alloc-util.h"
+#include "basic/macro.h"
+#include "basic/string-util.h"
+
+#include "journald-syslog.h"
+
+static void test_syslog_parse_identifier(const char* str,
+ const char *ident, const char*pid, int ret) {
+ const char *buf = str;
+ _cleanup_free_ char *ident2 = NULL, *pid2 = NULL;
+ int ret2;
+
+ ret2 = syslog_parse_identifier(&buf, &ident2, &pid2);
+
+ assert_se(ret == ret2);
+ assert_se(ident == ident2 || streq_ptr(ident, ident2));
+ assert_se(pid == pid2 || streq_ptr(pid, pid2));
+}
+
+int main(void) {
+ test_syslog_parse_identifier("pidu[111]: xxx", "pidu", "111", 11);
+ test_syslog_parse_identifier("pidu: xxx", "pidu", NULL, 6);
+ test_syslog_parse_identifier("pidu xxx", NULL, NULL, 0);
+
+ return 0;
+}
diff --git a/src/grp-journal/libjournal-core/test-journal-verify.c b/src/grp-journal/libjournal-core/test-journal-verify.c
new file mode 100644
index 0000000000..485315c118
--- /dev/null
+++ b/src/grp-journal/libjournal-core/test-journal-verify.c
@@ -0,0 +1,150 @@
+/***
+ This file is part of systemd.
+
+ Copyright 2012 Lennart Poettering
+
+ systemd is free software; you can redistribute it and/or modify it
+ under the terms of the GNU Lesser General Public License as published by
+ the Free Software Foundation; either version 2.1 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
+ Lesser General Public License for more details.
+
+ You should have received a copy of the GNU Lesser General Public License
+ along with systemd; If not, see <http://www.gnu.org/licenses/>.
+***/
+
+#include <fcntl.h>
+#include <stdio.h>
+#include <unistd.h>
+
+#include "basic/fd-util.h"
+#include "basic/log.h"
+#include "basic/rm-rf.h"
+#include "basic/terminal-util.h"
+#include "basic/util.h"
+#include "sd-journal/journal-file.h"
+#include "sd-journal/journal-verify.h"
+
+#define N_ENTRIES 6000
+#define RANDOM_RANGE 77
+
+static void bit_toggle(const char *fn, uint64_t p) {
+ uint8_t b;
+ ssize_t r;
+ int fd;
+
+ fd = open(fn, O_RDWR|O_CLOEXEC);
+ assert_se(fd >= 0);
+
+ r = pread(fd, &b, 1, p/8);
+ assert_se(r == 1);
+
+ b ^= 1 << (p % 8);
+
+ r = pwrite(fd, &b, 1, p/8);
+ assert_se(r == 1);
+
+ safe_close(fd);
+}
+
+static int raw_verify(const char *fn, const char *verification_key) {
+ JournalFile *f;
+ int r;
+
+ r = journal_file_open(-1, fn, O_RDONLY, 0666, true, !!verification_key, NULL, NULL, NULL, NULL, &f);
+ if (r < 0)
+ return r;
+
+ r = journal_file_verify(f, verification_key, NULL, NULL, NULL, false);
+ (void) journal_file_close(f);
+
+ return r;
+}
+
+int main(int argc, char *argv[]) {
+ char t[] = "/tmp/journal-XXXXXX";
+ unsigned n;
+ JournalFile *f;
+ const char *verification_key = argv[1];
+ usec_t from = 0, to = 0, total = 0;
+ char a[FORMAT_TIMESTAMP_MAX];
+ char b[FORMAT_TIMESTAMP_MAX];
+ char c[FORMAT_TIMESPAN_MAX];
+ struct stat st;
+ uint64_t p;
+
+ /* journal_file_open requires a valid machine id */
+ if (access("/etc/machine-id", F_OK) != 0)
+ return EXIT_TEST_SKIP;
+
+ log_set_max_level(LOG_DEBUG);
+
+ assert_se(mkdtemp(t));
+ assert_se(chdir(t) >= 0);
+
+ log_info("Generating...");
+
+ assert_se(journal_file_open(-1, "test.journal", O_RDWR|O_CREAT, 0666, true, !!verification_key, NULL, NULL, NULL, NULL, &f) == 0);
+
+ for (n = 0; n < N_ENTRIES; n++) {
+ struct iovec iovec;
+ struct dual_timestamp ts;
+ char *test;
+
+ dual_timestamp_get(&ts);
+
+ assert_se(asprintf(&test, "RANDOM=%lu", random() % RANDOM_RANGE));
+
+ iovec.iov_base = (void*) test;
+ iovec.iov_len = strlen(test);
+
+ assert_se(journal_file_append_entry(f, &ts, &iovec, 1, NULL, NULL, NULL) == 0);
+
+ free(test);
+ }
+
+ (void) journal_file_close(f);
+
+ log_info("Verifying...");
+
+ assert_se(journal_file_open(-1, "test.journal", O_RDONLY, 0666, true, !!verification_key, NULL, NULL, NULL, NULL, &f) == 0);
+ /* journal_file_print_header(f); */
+ journal_file_dump(f);
+
+ assert_se(journal_file_verify(f, verification_key, &from, &to, &total, true) >= 0);
+
+ if (verification_key && JOURNAL_HEADER_SEALED(f->header))
+ log_info("=> Validated from %s to %s, %s missing",
+ format_timestamp(a, sizeof(a), from),
+ format_timestamp(b, sizeof(b), to),
+ format_timespan(c, sizeof(c), total > to ? total - to : 0, 0));
+
+ (void) journal_file_close(f);
+
+ if (verification_key) {
+ log_info("Toggling bits...");
+
+ assert_se(stat("test.journal", &st) >= 0);
+
+ for (p = 38448*8+0; p < ((uint64_t) st.st_size * 8); p ++) {
+ bit_toggle("test.journal", p);
+
+ log_info("[ %"PRIu64"+%"PRIu64"]", p / 8, p % 8);
+
+ if (raw_verify("test.journal", verification_key) >= 0)
+ log_notice(ANSI_HIGHLIGHT_RED ">>>> %"PRIu64" (bit %"PRIu64") can be toggled without detection." ANSI_NORMAL, p / 8, p % 8);
+
+ bit_toggle("test.journal", p);
+ }
+ }
+
+ log_info("Exiting...");
+
+ assert_se(rm_rf(t, REMOVE_ROOT|REMOVE_PHYSICAL) >= 0);
+
+ return 0;
+}
diff --git a/src/grp-journal/libjournal-core/test-journal.c b/src/grp-journal/libjournal-core/test-journal.c
new file mode 100644
index 0000000000..05de15f0c0
--- /dev/null
+++ b/src/grp-journal/libjournal-core/test-journal.c
@@ -0,0 +1,178 @@
+/***
+ 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 Lesser General Public License as published by
+ the Free Software Foundation; either version 2.1 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
+ Lesser General Public License for more details.
+
+ You should have received a copy of the GNU Lesser General Public License
+ along with systemd; If not, see <http://www.gnu.org/licenses/>.
+***/
+
+#include <fcntl.h>
+#include <unistd.h>
+
+#include "basic/log.h"
+#include "basic/rm-rf.h"
+#include "sd-journal/journal-authenticate.h"
+#include "sd-journal/journal-file.h"
+#include "sd-journal/journal-vacuum.h"
+
+static bool arg_keep = false;
+
+static void test_non_empty(void) {
+ dual_timestamp ts;
+ JournalFile *f;
+ struct iovec iovec;
+ static const char test[] = "TEST1=1", test2[] = "TEST2=2";
+ Object *o;
+ uint64_t p;
+ char t[] = "/tmp/journal-XXXXXX";
+
+ log_set_max_level(LOG_DEBUG);
+
+ assert_se(mkdtemp(t));
+ assert_se(chdir(t) >= 0);
+
+ assert_se(journal_file_open(-1, "test.journal", O_RDWR|O_CREAT, 0666, true, true, NULL, NULL, NULL, NULL, &f) == 0);
+
+ dual_timestamp_get(&ts);
+
+ iovec.iov_base = (void*) test;
+ iovec.iov_len = strlen(test);
+ assert_se(journal_file_append_entry(f, &ts, &iovec, 1, NULL, NULL, NULL) == 0);
+
+ iovec.iov_base = (void*) test2;
+ iovec.iov_len = strlen(test2);
+ assert_se(journal_file_append_entry(f, &ts, &iovec, 1, NULL, NULL, NULL) == 0);
+
+ iovec.iov_base = (void*) test;
+ iovec.iov_len = strlen(test);
+ assert_se(journal_file_append_entry(f, &ts, &iovec, 1, NULL, NULL, NULL) == 0);
+
+#ifdef HAVE_GCRYPT
+ journal_file_append_tag(f);
+#endif
+ journal_file_dump(f);
+
+ assert_se(journal_file_next_entry(f, 0, DIRECTION_DOWN, &o, &p) == 1);
+ assert_se(le64toh(o->entry.seqnum) == 1);
+
+ assert_se(journal_file_next_entry(f, p, DIRECTION_DOWN, &o, &p) == 1);
+ assert_se(le64toh(o->entry.seqnum) == 2);
+
+ assert_se(journal_file_next_entry(f, p, DIRECTION_DOWN, &o, &p) == 1);
+ assert_se(le64toh(o->entry.seqnum) == 3);
+
+ assert_se(journal_file_next_entry(f, p, DIRECTION_DOWN, &o, &p) == 0);
+
+ assert_se(journal_file_next_entry(f, 0, DIRECTION_DOWN, &o, &p) == 1);
+ assert_se(le64toh(o->entry.seqnum) == 1);
+
+ assert_se(journal_file_find_data_object(f, test, strlen(test), NULL, &p) == 1);
+ assert_se(journal_file_next_entry_for_data(f, NULL, 0, p, DIRECTION_DOWN, &o, NULL) == 1);
+ assert_se(le64toh(o->entry.seqnum) == 1);
+
+ assert_se(journal_file_next_entry_for_data(f, NULL, 0, p, DIRECTION_UP, &o, NULL) == 1);
+ assert_se(le64toh(o->entry.seqnum) == 3);
+
+ assert_se(journal_file_find_data_object(f, test2, strlen(test2), NULL, &p) == 1);
+ assert_se(journal_file_next_entry_for_data(f, NULL, 0, p, DIRECTION_UP, &o, NULL) == 1);
+ assert_se(le64toh(o->entry.seqnum) == 2);
+
+ assert_se(journal_file_next_entry_for_data(f, NULL, 0, p, DIRECTION_DOWN, &o, NULL) == 1);
+ assert_se(le64toh(o->entry.seqnum) == 2);
+
+ assert_se(journal_file_find_data_object(f, "quux", 4, NULL, &p) == 0);
+
+ assert_se(journal_file_move_to_entry_by_seqnum(f, 1, DIRECTION_DOWN, &o, NULL) == 1);
+ assert_se(le64toh(o->entry.seqnum) == 1);
+
+ assert_se(journal_file_move_to_entry_by_seqnum(f, 3, DIRECTION_DOWN, &o, NULL) == 1);
+ assert_se(le64toh(o->entry.seqnum) == 3);
+
+ assert_se(journal_file_move_to_entry_by_seqnum(f, 2, DIRECTION_DOWN, &o, NULL) == 1);
+ assert_se(le64toh(o->entry.seqnum) == 2);
+
+ assert_se(journal_file_move_to_entry_by_seqnum(f, 10, DIRECTION_DOWN, &o, NULL) == 0);
+
+ journal_file_rotate(&f, true, true, NULL);
+ journal_file_rotate(&f, true, true, NULL);
+
+ (void) journal_file_close(f);
+
+ log_info("Done...");
+
+ if (arg_keep)
+ log_info("Not removing %s", t);
+ else {
+ journal_directory_vacuum(".", 3000000, 0, 0, NULL, true);
+
+ assert_se(rm_rf(t, REMOVE_ROOT|REMOVE_PHYSICAL) >= 0);
+ }
+
+ puts("------------------------------------------------------------");
+}
+
+static void test_empty(void) {
+ JournalFile *f1, *f2, *f3, *f4;
+ char t[] = "/tmp/journal-XXXXXX";
+
+ log_set_max_level(LOG_DEBUG);
+
+ assert_se(mkdtemp(t));
+ assert_se(chdir(t) >= 0);
+
+ assert_se(journal_file_open(-1, "test.journal", O_RDWR|O_CREAT, 0666, false, false, NULL, NULL, NULL, NULL, &f1) == 0);
+
+ assert_se(journal_file_open(-1, "test-compress.journal", O_RDWR|O_CREAT, 0666, true, false, NULL, NULL, NULL, NULL, &f2) == 0);
+
+ assert_se(journal_file_open(-1, "test-seal.journal", O_RDWR|O_CREAT, 0666, false, true, NULL, NULL, NULL, NULL, &f3) == 0);
+
+ assert_se(journal_file_open(-1, "test-seal-compress.journal", O_RDWR|O_CREAT, 0666, true, true, NULL, NULL, NULL, NULL, &f4) == 0);
+
+ journal_file_print_header(f1);
+ puts("");
+ journal_file_print_header(f2);
+ puts("");
+ journal_file_print_header(f3);
+ puts("");
+ journal_file_print_header(f4);
+ puts("");
+
+ log_info("Done...");
+
+ if (arg_keep)
+ log_info("Not removing %s", t);
+ else {
+ journal_directory_vacuum(".", 3000000, 0, 0, NULL, true);
+
+ assert_se(rm_rf(t, REMOVE_ROOT|REMOVE_PHYSICAL) >= 0);
+ }
+
+ (void) journal_file_close(f1);
+ (void) journal_file_close(f2);
+ (void) journal_file_close(f3);
+ (void) journal_file_close(f4);
+}
+
+int main(int argc, char *argv[]) {
+ arg_keep = argc > 1;
+
+ /* journal_file_open requires a valid machine id */
+ if (access("/etc/machine-id", F_OK) != 0)
+ return EXIT_TEST_SKIP;
+
+ test_non_empty();
+ test_empty();
+
+ return 0;
+}
diff --git a/src/grp-journal/libjournal-core/test-mmap-cache.c b/src/grp-journal/libjournal-core/test-mmap-cache.c
new file mode 100644
index 0000000000..3eb5444480
--- /dev/null
+++ b/src/grp-journal/libjournal-core/test-mmap-cache.c
@@ -0,0 +1,79 @@
+/***
+ This file is part of systemd.
+
+ Copyright 2012 Lennart Poettering
+
+ systemd is free software; you can redistribute it and/or modify it
+ under the terms of the GNU Lesser General Public License as published by
+ the Free Software Foundation; either version 2.1 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
+ Lesser General Public License for more details.
+
+ You should have received a copy of the GNU Lesser General Public License
+ along with systemd; If not, see <http://www.gnu.org/licenses/>.
+***/
+
+#include <fcntl.h>
+#include <stdlib.h>
+#include <sys/mman.h>
+#include <unistd.h>
+
+#include "basic/fd-util.h"
+#include "basic/fileio.h"
+#include "basic/macro.h"
+#include "basic/util.h"
+#include "sd-journal/mmap-cache.h"
+
+int main(int argc, char *argv[]) {
+ int x, y, z, r;
+ char px[] = "/tmp/testmmapXXXXXXX", py[] = "/tmp/testmmapYXXXXXX", pz[] = "/tmp/testmmapZXXXXXX";
+ MMapCache *m;
+ void *p, *q;
+
+ assert_se(m = mmap_cache_new());
+
+ x = mkostemp_safe(px, O_RDWR|O_CLOEXEC);
+ assert_se(x >= 0);
+ unlink(px);
+
+ y = mkostemp_safe(py, O_RDWR|O_CLOEXEC);
+ assert_se(y >= 0);
+ unlink(py);
+
+ z = mkostemp_safe(pz, O_RDWR|O_CLOEXEC);
+ assert_se(z >= 0);
+ unlink(pz);
+
+ r = mmap_cache_get(m, x, PROT_READ, 0, false, 1, 2, NULL, &p);
+ assert_se(r >= 0);
+
+ r = mmap_cache_get(m, x, PROT_READ, 0, false, 2, 2, NULL, &q);
+ assert_se(r >= 0);
+
+ assert_se((uint8_t*) p + 1 == (uint8_t*) q);
+
+ r = mmap_cache_get(m, x, PROT_READ, 1, false, 3, 2, NULL, &q);
+ assert_se(r >= 0);
+
+ assert_se((uint8_t*) p + 2 == (uint8_t*) q);
+
+ r = mmap_cache_get(m, x, PROT_READ, 0, false, 16ULL*1024ULL*1024ULL, 2, NULL, &p);
+ assert_se(r >= 0);
+
+ r = mmap_cache_get(m, x, PROT_READ, 1, false, 16ULL*1024ULL*1024ULL+1, 2, NULL, &q);
+ assert_se(r >= 0);
+
+ assert_se((uint8_t*) p + 1 == (uint8_t*) q);
+
+ mmap_cache_unref(m);
+
+ safe_close(x);
+ safe_close(y);
+ safe_close(z);
+
+ return 0;
+}