man/systemd.exec: explain how BPF token works

Add a small paragraph explaining how BPF token works, how it's being
created and its relationship between the BPF filesystem.
Move all the relevant documentation in the PrivateBPF= section and let
point all the BPFDelegate* options to that one.

man/systemd-journald.service.xml

@@ -218,6 +218,22 @@ systemd-tmpfiles --create --prefix /var/log/journal</programlisting>
 
         <xi:include href="version-info.xml" xpointer="v228"/></listitem>
       </varlistentry>
+
+      <varlistentry>
+        <term>SIGHUP</term>
+
+        <listitem><para>Upon reception of the <constant>SIGHUP</constant> process signal
+         <command>systemd-journald</command> will reload its configuration values
+         and update the kernel log buffer and journals to reflect the new configuration.
+         If <varname>ReadKmsg=</varname> has changed, the kernel log buffer will be flushed
+         and updated as part of the reload. The active journal (e.g., persistent, volatile)
+         will continue to be used with the updated configuration.
+         However, if the storage mode has changed from persistent to volatile
+         and the current journal in use is the persistent journal, then the active journal will
+         be switched to the volatile journal.</para>
+
+        <xi:include href="version-info.xml" xpointer="v258"/></listitem>
+      </varlistentry>
     </variablelist>
   </refsect1>
 

man/systemd.exec.xml

@@ -2561,10 +2561,16 @@ RestrictNamespaces=~cgroup net</programlisting>
         <listitem><para>Takes a boolean argument. If set, mount a private instance of the BPF filesystem
         on <filename>/sys/fs/bpf/</filename>, effectively hiding the host bpffs which contains information
         about loaded programs and maps. Otherwise, if <varname>ProtectKernelTunables=</varname> is set, the
-        instance from the host is inherited but mounted read-only.</para>
+        instance from the host is inherited but mounted read-only. Defaults to false.</para>
+
+        <para>This can be used together with the bpffs delegate feature to choose what BPF functions are
+        available to the unit's processes. When mounting the BPF filesystem with the fsopen() API, four mount
+        options can be specified to set a list of BPF commands, maps, programs and attachment types that are
+        allowed to be used. Processes needs to get a file descriptor for the bpffs mountpoint and use that to
+        get a token which will enable for that user namespace the BPF functionalities choosen upon bpffs mount.
+        A more detailed explanation of the feature can be found in this
+        <ulink url="https://lwn.net/Articles/947173/">LWN post</ulink>.</para>
 
-        <para>This can be used together with the BPF delegate feature to choose what BPF functions are
-        available to the unit's processes. Defaults to false.</para>
 
         <xi:include href="version-info.xml" xpointer="v258"/></listitem>
       </varlistentry>
@@ -2575,10 +2581,10 @@ RestrictNamespaces=~cgroup net</programlisting>
         <listitem><para>Accepts a list of BPF commands to allow or <literal>any</literal> to allow everything.
         Defaults to none. The accepted values are:
         <xi:include href="bpf-delegate.xml" xpointer="bpf_delegate_cmd"/>
-        Requires <varname>PrivateBPF=</varname> enabled to be effective.
-        This will set the <constant>delegate_cmds</constant> bpffs mount option.
-        A more detailed explanation of the feature can be found in this
-        <ulink url="https://lwn.net/Articles/947173/">LWN post</ulink>.</para>
+        This will set the <constant>delegate_cmds</constant> bpffs mount option.</para>
+
+        <para>Requires <varname>PrivateBPF=yes</varname> to be effective,
+        see <varname>PrivateBPF=</varname> more details.</para>
 
         <xi:include href="version-info.xml" xpointer="v258"/></listitem>
       </varlistentry>
@@ -2589,8 +2595,10 @@ RestrictNamespaces=~cgroup net</programlisting>
         <listitem><para>Accepts a list of BPF maps to allow or <literal>any</literal> to allow everything.
         Defaults to none. The accepted values are:
         <xi:include href="bpf-delegate.xml" xpointer="bpf_delegate_map_type"/>
-        This will set the <constant>delegate_maps</constant> bpffs mount option.
-        See <varname>BPFDelegateCommands=</varname> for dependencies and more details.</para>
+        This will set the <constant>delegate_maps</constant> bpffs mount option.</para>
+
+        <para>Requires <varname>PrivateBPF=yes</varname> to be effective,
+        see <varname>PrivateBPF=</varname> more details.</para>
 
         <xi:include href="version-info.xml" xpointer="v258"/></listitem>
       </varlistentry>
@@ -2601,8 +2609,10 @@ RestrictNamespaces=~cgroup net</programlisting>
         <listitem><para>Accepts a list of BPF programs to allow or <literal>any</literal> to allow everything.
         Defaults to none. The accepted values are:
         <xi:include href="bpf-delegate.xml" xpointer="bpf_delegate_prog_type"/>
-        This will set the <constant>delegate_progs</constant> bpffs mount option.
-        See <varname>BPFDelegateCommands=</varname> for dependencies and more details.</para>
+        This will set the <constant>delegate_progs</constant> bpffs mount option.</para>
+
+        <para>Requires <varname>PrivateBPF=yes</varname> to be effective,
+        see <varname>PrivateBPF=</varname> more details.</para>
 
         <xi:include href="version-info.xml" xpointer="v258"/></listitem>
       </varlistentry>
@@ -2613,8 +2623,10 @@ RestrictNamespaces=~cgroup net</programlisting>
         <listitem><para>Accepts a list of BPF attach points to allow or <literal>any</literal> to allow everything.
         Defaults to none. The accepted values are:
         <xi:include href="bpf-delegate.xml" xpointer="bpf_delegate_attach_type"/>
-        This will set the <constant>delegate_attachs</constant> bpffs mount option.
-        See <varname>BPFDelegateCommands=</varname> for dependencies and more details.</para>
+        This will set the <constant>delegate_attachs</constant> bpffs mount option.</para>
+
+        <para>Requires <varname>PrivateBPF=yes</varname> to be effective,
+        see <varname>PrivateBPF=</varname> more details.</para>
 
         <xi:include href="version-info.xml" xpointer="v258"/></listitem>
       </varlistentry>

man/systemd.system-credentials.xml

@@ -475,6 +475,8 @@
       <varlistentry>
         <term><varname>userdb.user.*</varname></term>
         <term><varname>userdb.group.*</varname></term>
+        <term><varname>userdb.transient.user.*</varname></term>
+        <term><varname>userdb.transient.group.*</varname></term>
 
         <listitem>
           <para>Configure JSON user and group records. Read by
@@ -499,7 +501,10 @@
           <literal>userdb.user.foobar</literal> the suffix is <literal>foobar</literal>) must match the user
           or group name field from the user or group record.</para>
 
-          <para>Note that the records are created in <filename>/etc/userdb/</filename>
+          <para>Note that the records created for <varname>userdb.user.*</varname> and
+          <varname>userdb.group.*</varname> credentials are created in <filename>/etc/userdb/</filename> and
+          the records created for <varname>userdb.transient.user.*</varname> and
+          <varname>userdb.transient.group.*</varname> are created in <filename>/run/userdb/</filename>
           (<filename>/etc/passwd</filename> and <filename>/etc/group</filename> are not modified).</para>
 
           <xi:include href="version-info.xml" xpointer="v258"/>

ruff.toml

@@ -1,6 +1,11 @@
-target-version = "py39"
+target-version = "py37"
 line-length = 109
 lint.select = ["E", "F", "I", "UP"]
 
 [format]
 quote-style = "single"
+
+[per-file-target-version]
+"src/ukify/*.py" = "py39"
+"test/**/integration-test-wrapper.py" = "py39"
+"test/test-udev.py" = "py39"

shell-completion/bash/kernel-install

@@ -21,6 +21,15 @@ _kernel_install() {
     local comps
     local MACHINE_ID
     local cur=${COMP_WORDS[COMP_CWORD]}
+    local prev=${COMP_WORDS[COMP_CWORD-1]}
+    local entry_types="type1 type2 all"
+
+    case "$prev" in
+        --entry-type)
+            COMPREPLY=( $(compgen -W "$entry_types" -- "$cur") )
+            return 0
+            ;;
+    esac
 
     case $COMP_CWORD in
         1)
@@ -42,6 +51,11 @@ _kernel_install() {
             ;;
     esac
 
+    if [[ "${COMP_WORDS[1]}" == "remove" ]] && [[ $cur == -* ]]; then
+        COMPREPLY=( $(compgen -W '--entry-type' -- "$cur") )
+        return 0
+    fi
+
     COMPREPLY=( $(compgen -W '$comps' -- "$cur") )
     return 0
 }

src/basic/errno-util.h

@@ -188,6 +188,12 @@ static inline bool ERRNO_IS_NEG_PRIVILEGE(intmax_t r) {
 }
 _DEFINE_ABS_WRAPPER(PRIVILEGE);
 
+/* Three different errors for writing on a filesystem */
+static inline bool ERRNO_IS_NEG_FS_WRITE_REFUSED(intmax_t r) {
+        return r == -EROFS || ERRNO_IS_NEG_PRIVILEGE(r);
+}
+_DEFINE_ABS_WRAPPER(FS_WRITE_REFUSED);
+
 /* Three different errors for "not enough disk space" */
 static inline bool ERRNO_IS_NEG_DISK_SPACE(intmax_t r) {
         return IN_SET(r,

src/binfmt/binfmt.c

@@ -193,10 +193,10 @@ static int parse_argv(int argc, char *argv[]) {
         return 1;
 }
 
-static int binfmt_mounted_warn(void) {
+static int binfmt_mounted_and_writable_warn(void) {
         int r;
 
-        r = binfmt_mounted();
+        r = binfmt_mounted_and_writable();
         if (r < 0)
                 return log_error_errno(r, "Failed to check if /proc/sys/fs/binfmt_misc is mounted: %m");
         if (r == 0)
@@ -222,7 +222,7 @@ static int run(int argc, char *argv[]) {
                 return disable_binfmt();
 
         if (argc > optind) {
-                r = binfmt_mounted_warn();
+                r = binfmt_mounted_and_writable_warn();
                 if (r <= 0)
                         return r;
 
@@ -239,7 +239,7 @@ static int run(int argc, char *argv[]) {
                 if (arg_cat_flags != CAT_CONFIG_OFF)
                         return cat_config(files);
 
-                r = binfmt_mounted_warn();
+                r = binfmt_mounted_and_writable_warn();
                 if (r <= 0)
                         return r;
 

src/core/cgroup.c

@@ -62,7 +62,7 @@
 /* Returns the log level to use when cgroup attribute writes fail. When an attribute is missing or we have access
  * problems we downgrade to LOG_DEBUG. This is supposed to be nice to container managers and kernels which want to mask
  * out specific attributes from us. */
-#define LOG_LEVEL_CGROUP_WRITE(r) (IN_SET(ABS(r), ENOENT, EROFS, EACCES, EPERM) ? LOG_DEBUG : LOG_WARNING)
+#define LOG_LEVEL_CGROUP_WRITE(r) ((ABS(r) == ENOENT || ERRNO_IS_FS_WRITE_REFUSED(r)) ? LOG_DEBUG : LOG_WARNING)
 
 static void unit_remove_from_cgroup_empty_queue(Unit *u);
 

src/core/main.c

@@ -1271,8 +1271,8 @@ static void bump_file_max_and_nr_open(void) {
          * different, but the operation would fail silently.) */
         r = sysctl_write("fs/file-max", LONG_MAX_STR);
         if (r < 0)
-                log_full_errno(IN_SET(r, -EROFS, -EPERM, -EACCES) ? LOG_DEBUG : LOG_WARNING,
-                               r, "Failed to bump fs.file-max, ignoring: %m");
+                log_full_errno(ERRNO_IS_NEG_FS_WRITE_REFUSED(r) ? LOG_DEBUG : LOG_WARNING, r,
+                               "Failed to bump fs.file-max, ignoring: %m");
 #endif
 
 #if BUMP_PROC_SYS_FS_NR_OPEN
@@ -1293,7 +1293,8 @@ static void bump_file_max_and_nr_open(void) {
                         continue;
                 }
                 if (r < 0) {
-                        log_full_errno(IN_SET(r, -EROFS, -EPERM, -EACCES) ? LOG_DEBUG : LOG_WARNING, r, "Failed to bump fs.nr_open, ignoring: %m");
+                        log_full_errno(ERRNO_IS_NEG_FS_WRITE_REFUSED(r) ? LOG_DEBUG : LOG_WARNING, r,
+                                       "Failed to bump fs.nr_open, ignoring: %m");
                         break;
                 }
 
@@ -1534,7 +1535,7 @@ static int bump_unix_max_dgram_qlen(void) {
 
         r = sysctl_write("net/unix/max_dgram_qlen", STRINGIFY(DEFAULT_UNIX_MAX_DGRAM_QLEN));
         if (r < 0)
-                return log_full_errno(IN_SET(r, -EROFS, -EPERM, -EACCES) ? LOG_DEBUG : LOG_WARNING, r,
+                return log_full_errno(ERRNO_IS_NEG_FS_WRITE_REFUSED(r) ? LOG_DEBUG : LOG_WARNING, r,
                                       "Failed to bump AF_UNIX datagram queue length, ignoring: %m");
 
         return 1;

src/core/mount.c

@@ -1269,7 +1269,7 @@ static void mount_enter_mounting(Mount *m) {
                  * when the path is on NFS. See issue #24120. All such errors will be logged in the debug level. */
                 if (r < 0 && r != -EEXIST)
                         log_unit_full_errno(UNIT(m),
-                                            (r == -EROFS || ERRNO_IS_PRIVILEGE(r)) ? LOG_DEBUG : LOG_WARNING,
+                                            ERRNO_IS_NEG_FS_WRITE_REFUSED(r) ? LOG_DEBUG : LOG_WARNING,
                                             r, "Failed to make bind mount source '%s', ignoring: %m", p->what);
         }
 

src/journal/fuzz-journald-audit.c

@@ -9,7 +9,7 @@ int LLVMFuzzerTestOneInput(const uint8_t *data, size_t size) {
 
         fuzz_setup_logging();
 
-        assert_se(manager_new(&m) >= 0);
+        assert_se(manager_new(&m, /* namespace= */ NULL) >= 0);
         dummy_manager_init(m, data, size);
         process_audit_string(m, 0, m->buffer, size);
 

src/journal/fuzz-journald-kmsg.c

@@ -12,7 +12,7 @@ int LLVMFuzzerTestOneInput(const uint8_t *data, size_t size) {
 
         fuzz_setup_logging();
 
-        assert_se(manager_new(&m) >= 0);
+        assert_se(manager_new(&m, /* namespace= */ NULL) >= 0);
         dummy_manager_init(m, data, size);
         dev_kmsg_record(m, m->buffer, size);
 

src/journal/fuzz-journald-native-fd.c

@@ -21,7 +21,7 @@ int LLVMFuzzerTestOneInput(const uint8_t *data, size_t size) {
 
         fuzz_setup_logging();
 
-        assert_se(manager_new(&m) >= 0);
+        assert_se(manager_new(&m, /* namespace= */ NULL) >= 0);
         dummy_manager_init(m, NULL, 0);
 
         sealed_fd = memfd_new_and_seal(NULL, data, size);

src/journal/fuzz-journald-stream.c

@@ -23,7 +23,7 @@ int LLVMFuzzerTestOneInput(const uint8_t *data, size_t size) {
         fuzz_setup_logging();
 
         assert_se(socketpair(AF_UNIX, SOCK_STREAM|SOCK_CLOEXEC|SOCK_NONBLOCK, 0, stream_fds) >= 0);
-        assert_se(manager_new(&m) >= 0);
+        assert_se(manager_new(&m, /* namespace= */ NULL) >= 0);
         dummy_manager_init(m, NULL, 0);
 
         assert_se(stdout_stream_install(m, stream_fds[0], &stream) >= 0);

src/journal/fuzz-journald.c

@@ -9,7 +9,7 @@
 void dummy_manager_init(Manager *m, const uint8_t *buffer, size_t size) {
         assert(m);
 
-        m->storage = STORAGE_NONE;
+        m->config.storage = STORAGE_NONE;
         assert_se(sd_event_default(&m->event) >= 0);
 
         if (buffer) {
@@ -33,7 +33,7 @@ void fuzz_journald_processing_function(
         if (size == 0)
                 return;
 
-        assert_se(manager_new(&m) >= 0);
+        assert_se(manager_new(&m, /* namespace= */ NULL) >= 0);
         dummy_manager_init(m, data, size);
         (*f)(m, m->buffer, size, ucred, tv, label, label_len);
 }

src/journal/journald-console.c

@@ -52,7 +52,7 @@ void manager_forward_console(
         assert(m);
         assert(message);
 
-        if (LOG_PRI(priority) > m->max_level_console)
+        if (LOG_PRI(priority) > m->config.max_level_console)
                 return;
 
         /* First: timestamp */

src/journal/journald-gperf.gperf

@@ -19,7 +19,7 @@ struct ConfigPerfItem;
 %struct-type
 %includes
 %%
-Journal.Storage,            config_parse_storage,           0, offsetof(Manager, storage)
+Journal.Storage,            config_parse_storage,           0, offsetof(Manager, config_by_conf.storage)
 Journal.Compress,           config_parse_compress,          0, offsetof(Manager, compress)
 Journal.Seal,               config_parse_bool,              0, offsetof(Manager, seal)
 Journal.ReadKMsg,           config_parse_bool,              0, offsetof(Manager, read_kmsg)
@@ -39,17 +39,17 @@ Journal.RuntimeKeepFree,    config_parse_iec_uint64,        0, offsetof(Manager,
 Journal.RuntimeMaxFiles,    config_parse_uint64,            0, offsetof(Manager, runtime_storage.metrics.n_max_files)
 Journal.MaxRetentionSec,    config_parse_sec,               0, offsetof(Manager, max_retention_usec)
 Journal.MaxFileSec,         config_parse_sec,               0, offsetof(Manager, max_file_usec)
-Journal.ForwardToSyslog,    config_parse_bool,              0, offsetof(Manager, forward_to_syslog)
-Journal.ForwardToKMsg,      config_parse_bool,              0, offsetof(Manager, forward_to_kmsg)
-Journal.ForwardToConsole,   config_parse_bool,              0, offsetof(Manager, forward_to_console)
-Journal.ForwardToWall,      config_parse_bool,              0, offsetof(Manager, forward_to_wall)
-Journal.ForwardToSocket,    config_parse_forward_to_socket, 0, offsetof(Manager, forward_to_socket)
+Journal.ForwardToSyslog,    config_parse_bool,              0, offsetof(Manager, config_by_conf.forward_to_syslog)
+Journal.ForwardToKMsg,      config_parse_bool,              0, offsetof(Manager, config_by_conf.forward_to_kmsg)
+Journal.ForwardToConsole,   config_parse_bool,              0, offsetof(Manager, config_by_conf.forward_to_console)
+Journal.ForwardToWall,      config_parse_bool,              0, offsetof(Manager, config_by_conf.forward_to_wall)
+Journal.ForwardToSocket,    config_parse_forward_to_socket, 0, offsetof(Manager, config_by_conf.forward_to_socket)
 Journal.TTYPath,            config_parse_path,              0, offsetof(Manager, tty_path)
-Journal.MaxLevelStore,      config_parse_log_level,         0, offsetof(Manager, max_level_store)
-Journal.MaxLevelSyslog,     config_parse_log_level,         0, offsetof(Manager, max_level_syslog)
-Journal.MaxLevelKMsg,       config_parse_log_level,         0, offsetof(Manager, max_level_kmsg)
-Journal.MaxLevelConsole,    config_parse_log_level,         0, offsetof(Manager, max_level_console)
-Journal.MaxLevelWall,       config_parse_log_level,         0, offsetof(Manager, max_level_wall)
-Journal.MaxLevelSocket,     config_parse_log_level,         0, offsetof(Manager, max_level_socket)
+Journal.MaxLevelStore,      config_parse_log_level,         0, offsetof(Manager, config_by_conf.max_level_store)
+Journal.MaxLevelSyslog,     config_parse_log_level,         0, offsetof(Manager, config_by_conf.max_level_syslog)
+Journal.MaxLevelKMsg,       config_parse_log_level,         0, offsetof(Manager, config_by_conf.max_level_kmsg)
+Journal.MaxLevelConsole,    config_parse_log_level,         0, offsetof(Manager, config_by_conf.max_level_console)
+Journal.MaxLevelWall,       config_parse_log_level,         0, offsetof(Manager, config_by_conf.max_level_wall)
+Journal.MaxLevelSocket,     config_parse_log_level,         0, offsetof(Manager, config_by_conf.max_level_socket)
 Journal.SplitMode,          config_parse_split_mode,        0, offsetof(Manager, split_mode)
 Journal.LineMax,            config_parse_line_max,          0, offsetof(Manager, line_max)

src/journal/journald-kmsg.c

@@ -46,7 +46,7 @@ void manager_forward_kmsg(
         assert(priority <= 999);
         assert(message);
 
-        if (_unlikely_(LOG_PRI(priority) > m->max_level_kmsg))
+        if (_unlikely_(LOG_PRI(priority) > m->config.max_level_kmsg))
                 return;
 
         if (_unlikely_(m->dev_kmsg_fd < 0))
@@ -128,7 +128,7 @@ void dev_kmsg_record(Manager *m, char *p, size_t l) {
         if (r < 0 || priority < 0 || priority > 999)
                 return;
 
-        if (m->forward_to_kmsg && LOG_FAC(priority) != LOG_KERN)
+        if (m->config.forward_to_kmsg && LOG_FAC(priority) != LOG_KERN)
                 return;
 
         /* seqnum */
@@ -389,6 +389,10 @@ static int dispatch_dev_kmsg(sd_event_source *es, int fd, uint32_t revents, void
         return manager_read_dev_kmsg(m);
 }
 
+static mode_t manager_kmsg_mode(bool read_kmsg) {
+        return O_CLOEXEC|O_NONBLOCK|O_NOCTTY|(read_kmsg ? O_RDWR : O_WRONLY);
+}
+
 int manager_open_dev_kmsg(Manager *m) {
         int r;
 
@@ -396,7 +400,7 @@ int manager_open_dev_kmsg(Manager *m) {
         assert(m->dev_kmsg_fd < 0);
         assert(!m->dev_kmsg_event_source);
 
-        mode_t mode = O_CLOEXEC|O_NONBLOCK|O_NOCTTY|(m->read_kmsg ? O_RDWR : O_WRONLY);
+        mode_t mode = manager_kmsg_mode(m->read_kmsg);
 
         _cleanup_close_ int fd = open("/dev/kmsg", mode);
         if (fd < 0) {
@@ -441,3 +445,33 @@ int manager_open_kernel_seqnum(Manager *m) {
 
         return 0;
 }
+
+int manager_reload_dev_kmsg(Manager *m) {
+        int r;
+
+        assert(m);
+
+        /* Check if the fd has not yet been initialized. If so, open /dev/kmsg. */
+        if (m->dev_kmsg_fd < 0)
+                return manager_open_dev_kmsg(m);
+
+        mode_t mode = manager_kmsg_mode(m->read_kmsg);
+        int flags = fcntl(m->dev_kmsg_fd, F_GETFL);
+        if (flags < 0)
+                /* Proceed with reload in case the flags have changed. */
+                log_warning_errno(errno, "Failed to get flags for /dev/kmsg, ignoring: %m");
+        else if ((flags & O_ACCMODE_STRICT) == mode)
+                /* Mode is the same. No-op. */
+                return 0;
+
+        /* Flush kmsg. */
+        r = manager_flush_dev_kmsg(m);
+        if (r < 0)
+                log_warning_errno(r, "Failed to flush /dev/kmsg on reload, ignoring: %m");
+
+        /* Set kmsg values to default. */
+        m->dev_kmsg_event_source = sd_event_source_disable_unref(m->dev_kmsg_event_source);
+        m->dev_kmsg_fd = safe_close(m->dev_kmsg_fd);
+
+        return manager_open_dev_kmsg(m);
+}

src/journal/journald-kmsg.h

@@ -5,6 +5,7 @@
 
 int manager_open_dev_kmsg(Manager *m);
 int manager_flush_dev_kmsg(Manager *m);
+int manager_reload_dev_kmsg(Manager *m);
 
 void manager_forward_kmsg(Manager *m, int priority, const char *identifier, const char *message, const struct ucred *ucred);
 

src/journal/journald-manager.c

@@ -100,6 +100,7 @@
 
 static int manager_schedule_sync(Manager *m, int priority);
 static int manager_refresh_idle_timer(Manager *m);
+static int dispatch_reload_signal(sd_event_source *s, const struct signalfd_siginfo *si, void *userdata);
 
 static int manager_determine_path_usage(
                 Manager *m,
@@ -269,6 +270,12 @@ static void manager_add_acls(JournalFile *f, uid_t uid) {
 #endif
 }
 
+static int manager_get_file_flags(Manager *m, bool seal) {
+        return (m->compress.enabled ? JOURNAL_COMPRESS : 0) |
+               (seal ? JOURNAL_SEAL : 0) |
+               JOURNAL_STRICT_ORDER;
+}
+
 static int manager_open_journal(
                 Manager *m,
                 bool reliably,
@@ -286,10 +293,7 @@ static int manager_open_journal(
         assert(fname);
         assert(ret);
 
-        file_flags =
-                (m->compress.enabled ? JOURNAL_COMPRESS : 0) |
-                (seal ? JOURNAL_SEAL : 0) |
-                JOURNAL_STRICT_ORDER;
+        file_flags = manager_get_file_flags(m, seal);
 
         set_clear(m->deferred_closes);
 
@@ -363,7 +367,7 @@ static int manager_system_journal_open(
         int r = 0;
 
         if (!m->system_journal &&
-            IN_SET(m->storage, STORAGE_PERSISTENT, STORAGE_AUTO) &&
+            IN_SET(m->config.storage, STORAGE_PERSISTENT, STORAGE_AUTO) &&
             (flush_requested || manager_flushed_flag_is_set(m)) &&
             !relinquish_requested) {
 
@@ -371,7 +375,7 @@ static int manager_system_journal_open(
                  *
                  * If in persistent mode: create /var/log/journal and the machine path */
 
-                if (m->storage == STORAGE_PERSISTENT)
+                if (m->config.storage == STORAGE_PERSISTENT)
                         (void) mkdir_parents(m->system_storage.path, 0755);
 
                 (void) mkdir(m->system_storage.path, 0755);
@@ -408,7 +412,7 @@ static int manager_system_journal_open(
         }
 
         if (!m->runtime_journal &&
-            (m->storage != STORAGE_NONE)) {
+            (m->config.storage != STORAGE_NONE)) {
 
                 fn = strjoina(m->runtime_storage.path, "/system.journal");
 
@@ -532,7 +536,7 @@ static JournalFile* manager_find_journal(Manager *m, uid_t uid) {
          * persistent journal of any sort.
          *
          * Fixes https://github.com/systemd/systemd/issues/20390 */
-        if (!IN_SET(m->storage, STORAGE_AUTO, STORAGE_PERSISTENT))
+        if (!IN_SET(m->config.storage, STORAGE_AUTO, STORAGE_PERSISTENT))
                 return NULL;
 
         if (!uid_for_system_journal(uid)) {
@@ -1279,12 +1283,12 @@ void manager_dispatch_message(
         if (n == 0)
                 return;
 
-        if (LOG_PRI(priority) > m->max_level_store)
+        if (LOG_PRI(priority) > m->config.max_level_store)
                 return;
 
         /* Stop early in case the information will not be stored
          * in a journal. */
-        if (m->storage == STORAGE_NONE)
+        if (m->config.storage == STORAGE_NONE)
                 return;
 
         if (c && c->unit) {
@@ -1320,7 +1324,7 @@ int manager_flush_to_var(Manager *m, bool require_flag_file) {
 
         assert(m);
 
-        if (!IN_SET(m->storage, STORAGE_AUTO, STORAGE_PERSISTENT))
+        if (!IN_SET(m->config.storage, STORAGE_AUTO, STORAGE_PERSISTENT))
                 return 0;
 
         if (m->namespace) /* Flushing concept does not exist for namespace instances */
@@ -1468,7 +1472,7 @@ finish:
 int manager_relinquish_var(Manager *m) {
         assert(m);
 
-        if (m->storage == STORAGE_NONE)
+        if (m->config.storage == STORAGE_NONE)
                 return 0;
 
         if (m->namespace) /* Concept does not exist for namespaced instances */
@@ -1859,6 +1863,10 @@ static int manager_setup_signals(Manager *m) {
         if (r < 0)
                 return r;
 
+        r = sd_event_add_signal(m->event, NULL, SIGHUP|SD_EVENT_SIGNAL_PROCMASK, dispatch_reload_signal, m);
+        if (r < 0)
+                return r;
+
         return 0;
 }
 
@@ -1872,7 +1880,7 @@ static int parse_proc_cmdline_item(const char *key, const char *value, void *dat
                 if (r < 0)
                         log_warning("Failed to parse forward to syslog switch \"%s\". Ignoring.", value);
                 else
-                        m->forward_to_syslog = r;
+                        m->config_by_cmdline.forward_to_syslog = r;
 
         } else if (proc_cmdline_key_streq(key, "systemd.journald.forward_to_kmsg")) {
 
@@ -1880,7 +1888,7 @@ static int parse_proc_cmdline_item(const char *key, const char *value, void *dat
                 if (r < 0)
                         log_warning("Failed to parse forward to kmsg switch \"%s\". Ignoring.", value);
                 else
-                        m->forward_to_kmsg = r;
+                        m->config_by_cmdline.forward_to_kmsg = r;
 
         } else if (proc_cmdline_key_streq(key, "systemd.journald.forward_to_console")) {
 
@@ -1888,7 +1896,7 @@ static int parse_proc_cmdline_item(const char *key, const char *value, void *dat
                 if (r < 0)
                         log_warning("Failed to parse forward to console switch \"%s\". Ignoring.", value);
                 else
-                        m->forward_to_console = r;
+                        m->config_by_cmdline.forward_to_console = r;
 
         } else if (proc_cmdline_key_streq(key, "systemd.journald.forward_to_wall")) {
 
@@ -1896,7 +1904,7 @@ static int parse_proc_cmdline_item(const char *key, const char *value, void *dat
                 if (r < 0)
                         log_warning("Failed to parse forward to wall switch \"%s\". Ignoring.", value);
                 else
-                        m->forward_to_wall = r;
+                        m->config_by_cmdline.forward_to_wall = r;
 
         } else if (proc_cmdline_key_streq(key, "systemd.journald.max_level_console")) {
 
@@ -1907,7 +1915,7 @@ static int parse_proc_cmdline_item(const char *key, const char *value, void *dat
                 if (r < 0)
                         log_warning("Failed to parse max level console value \"%s\". Ignoring.", value);
                 else
-                        m->max_level_console = r;
+                        m->config_by_cmdline.max_level_console = r;
 
         } else if (proc_cmdline_key_streq(key, "systemd.journald.max_level_store")) {
 
@@ -1918,7 +1926,7 @@ static int parse_proc_cmdline_item(const char *key, const char *value, void *dat
                 if (r < 0)
                         log_warning("Failed to parse max level store value \"%s\". Ignoring.", value);
                 else
-                        m->max_level_store = r;
+                        m->config_by_cmdline.max_level_store = r;
 
         } else if (proc_cmdline_key_streq(key, "systemd.journald.max_level_syslog")) {
 
@@ -1929,7 +1937,7 @@ static int parse_proc_cmdline_item(const char *key, const char *value, void *dat
                 if (r < 0)
                         log_warning("Failed to parse max level syslog value \"%s\". Ignoring.", value);
                 else
-                        m->max_level_syslog = r;
+                        m->config_by_cmdline.max_level_syslog = r;
 
         } else if (proc_cmdline_key_streq(key, "systemd.journald.max_level_kmsg")) {
 
@@ -1940,7 +1948,7 @@ static int parse_proc_cmdline_item(const char *key, const char *value, void *dat
                 if (r < 0)
                         log_warning("Failed to parse max level kmsg value \"%s\". Ignoring.", value);
                 else
-                        m->max_level_kmsg = r;
+                        m->config_by_cmdline.max_level_kmsg = r;
 
         } else if (proc_cmdline_key_streq(key, "systemd.journald.max_level_wall")) {
 
@@ -1951,7 +1959,7 @@ static int parse_proc_cmdline_item(const char *key, const char *value, void *dat
                 if (r < 0)
                         log_warning("Failed to parse max level wall value \"%s\". Ignoring.", value);
                 else
-                        m->max_level_wall = r;
+                        m->config_by_cmdline.max_level_wall = r;
 
         } else if (proc_cmdline_key_streq(key, "systemd.journald.max_level_socket")) {
 
@@ -1962,7 +1970,7 @@ static int parse_proc_cmdline_item(const char *key, const char *value, void *dat
                 if (r < 0)
                         log_warning("Failed to parse max level socket value \"%s\". Ignoring.", value);
                 else
-                        m->max_level_socket = r;
+                        m->config_by_cmdline.max_level_socket = r;
 
         } else if (startswith(key, "systemd.journald"))
                 log_warning("Unknown journald kernel command line option \"%s\". Ignoring.", key);
@@ -2394,7 +2402,7 @@ static void manager_load_credentials(Manager *m) {
         if (r < 0)
                 log_debug_errno(r, "Failed to read credential journal.forward_to_socket, ignoring: %m");
         else {
-                r = socket_address_parse(&m->forward_to_socket, data);
+                r = socket_address_parse(&m->config_by_cred.forward_to_socket, data);
                 if (r < 0)
                         log_debug_errno(r, "Failed to parse socket address '%s' from credential journal.forward_to_socket, ignoring: %m", (char *) data);
         }
@@ -2409,12 +2417,225 @@ static void manager_load_credentials(Manager *m) {
                 if (r < 0)
                         log_debug_errno(r, "Failed to parse storage '%s' from credential journal.storage, ignoring: %m", (char *) data);
                 else
-                        m->storage = r;
+                        m->config_by_cred.storage = r;
         }
 }
 
-int manager_new(Manager **ret) {
+static void manager_set_defaults(Manager *m) {
+        assert(m);
+
+        m->compress.enabled = true;
+        m->compress.threshold_bytes = UINT64_MAX;
+
+        m->seal = true;
+
+        /* By default, only read from /dev/kmsg if are the main namespace */
+        m->read_kmsg = !m->namespace;
+
+        m->set_audit = true;
+
+        m->sync_interval_usec = DEFAULT_SYNC_INTERVAL_USEC;
+
+        m->ratelimit_interval = DEFAULT_RATE_LIMIT_INTERVAL;
+        m->ratelimit_burst = DEFAULT_RATE_LIMIT_BURST;
+
+        m->system_storage.name = "System Journal";
+        journal_reset_metrics(&m->system_storage.metrics);
+
+        m->runtime_storage.name = "Runtime Journal";
+        journal_reset_metrics(&m->runtime_storage.metrics);
+
+        m->max_file_usec = DEFAULT_MAX_FILE_USEC;
+
+        m->config.forward_to_wall = true;
+
+        m->config.max_level_store = LOG_DEBUG;
+        m->config.max_level_syslog = LOG_DEBUG;
+        m->config.max_level_kmsg = LOG_NOTICE;
+        m->config.max_level_console = LOG_INFO;
+        m->config.max_level_wall = LOG_EMERG;
+        m->config.max_level_socket = LOG_DEBUG;
+
+        m->line_max = DEFAULT_LINE_MAX;
+}
+
+static void manager_reset_configs(Manager *m) {
+        assert(m);
+
+        m->config_by_cmdline = JOURNAL_CONFIG_INIT;
+        m->config_by_conf = JOURNAL_CONFIG_INIT;
+        m->config_by_cred = JOURNAL_CONFIG_INIT;
+}
+
+static void manager_adjust_configs(Manager *m) {
+        assert(m);
+
+        if (!!m->ratelimit_interval != !!m->ratelimit_burst) { /* One set to 0 and the other not? */
+                log_debug(
+                                "Setting both rate limit interval and burst from %s/%u to 0/0",
+                                FORMAT_TIMESPAN(m->ratelimit_interval, USEC_PER_SEC),
+                                m->ratelimit_burst);
+                m->ratelimit_interval = m->ratelimit_burst = 0;
+        }
+}
+
+static void manager_merge_forward_to_socket(Manager *m) {
+        assert(m);
+
+        /* Conf file takes precendence over credentials. */
+        if (m->config_by_conf.forward_to_socket.sockaddr.sa.sa_family != AF_UNSPEC)
+                m->config.forward_to_socket = m->config_by_conf.forward_to_socket;
+        else if (m->config_by_cred.forward_to_socket.sockaddr.sa.sa_family != AF_UNSPEC)
+                m->config.forward_to_socket = m->config_by_cred.forward_to_socket;
+        else
+                m->config.forward_to_socket = (SocketAddress) { .sockaddr.sa.sa_family = AF_UNSPEC };
+}
+
+static void manager_merge_storage(Manager *m) {
+        assert(m);
+
+        /* Conf file takes precendence over credentials. */
+        if (m->config_by_conf.storage != _STORAGE_INVALID)
+                m->config.storage = m->config_by_conf.storage;
+        else if (m->config_by_cred.storage != _STORAGE_INVALID)
+                m->config.storage = m->config_by_cred.storage;
+        else
+                m->config.storage = m->namespace ? STORAGE_PERSISTENT : STORAGE_AUTO;
+}
+
+#define MERGE_BOOL(name, default_value)                                                 \
+    (m->config.name = (m->config_by_cmdline.name  ? m->config_by_cmdline.name :         \
+                       m->config_by_conf.name     ? m->config_by_conf.name     :        \
+                       m->config_by_cred.name     ? m->config_by_cred.name     :        \
+                       default_value))
+
+#define MERGE_NON_NEGATIVE(name, default_value)                                         \
+        (m->config.name = (m->config_by_cmdline.name >= 0 ? m->config_by_cmdline.name : \
+                           m->config_by_conf.name >= 0    ? m->config_by_conf.name :    \
+                           m->config_by_cred.name >= 0    ? m->config_by_cred.name :    \
+                           default_value))
+
+static void manager_merge_configs(Manager *m) {
+        assert(m);
+
+        /*
+         * From highest to lowest priority: cmdline, conf, cred
+         */
+        manager_merge_storage(m);
+        manager_merge_forward_to_socket(m);
+
+        MERGE_BOOL(forward_to_syslog, false);
+        MERGE_BOOL(forward_to_kmsg, false);
+        MERGE_BOOL(forward_to_console, false);
+        MERGE_BOOL(forward_to_wall, true);
+
+        MERGE_NON_NEGATIVE(max_level_store, LOG_DEBUG);
+        MERGE_NON_NEGATIVE(max_level_syslog, LOG_DEBUG);
+        MERGE_NON_NEGATIVE(max_level_kmsg, LOG_NOTICE);
+        MERGE_NON_NEGATIVE(max_level_console, LOG_INFO);
+        MERGE_NON_NEGATIVE(max_level_wall, LOG_EMERG);
+        MERGE_NON_NEGATIVE(max_level_socket, LOG_DEBUG);
+}
+
+static void manager_load_config(Manager *m) {
+        assert(m);
+
+        int r;
+
+        manager_set_defaults(m);
+        manager_reset_configs(m);
+
+        manager_load_credentials(m);
+        manager_parse_config_file(m);
+
+        if (!m->namespace) {
+                /* Parse kernel command line, but only if we are not a namespace instance */
+                r = proc_cmdline_parse(parse_proc_cmdline_item, m, PROC_CMDLINE_STRIP_RD_PREFIX);
+                if (r < 0)
+                        log_warning_errno(r, "Failed to parse kernel command line, ignoring: %m");
+        }
+
+        manager_merge_configs(m);
+
+        manager_adjust_configs(m);
+}
+
+static void manager_reload_config(Manager *m) {
+        assert(m);
+
+        manager_set_defaults(m);
+
+        m->config_by_conf = JOURNAL_CONFIG_INIT;
+        manager_parse_config_file(m);
+
+        manager_merge_configs(m);
+        manager_adjust_configs(m);
+}
+
+static int manager_reload_journals(Manager *m) {
+        assert(m);
+
+        int r;
+
+        if (m->system_journal && IN_SET(m->config.storage, STORAGE_PERSISTENT, STORAGE_AUTO)) {
+                /* Current journal can continue being used. Update config values as needed. */
+                r = journal_file_reload(
+                                m->system_journal,
+                                manager_get_file_flags(m, m->seal),
+                                m->compress.threshold_bytes,
+                                &m->system_storage.metrics);
+                if (r < 0)
+                        return log_warning_errno(r, "Failed to reload system journal on reload, ignoring: %m");
+        } else if (m->system_journal && m->config.storage == STORAGE_VOLATILE) {
+                /* Journal needs to be switched from system to runtime. */
+                r = manager_relinquish_var(m);
+                if (r < 0)
+                        return log_warning_errno(r, "Failed to relinquish to runtime journal on reload, ignoring: %m");
+        } else if (m->runtime_journal && IN_SET(m->config.storage, STORAGE_PERSISTENT, STORAGE_AUTO, STORAGE_VOLATILE)) {
+                /* Current journal can continue being used. Update config values as needed.*/
+                r = journal_file_reload(
+                                m->runtime_journal,
+                                manager_get_file_flags(m, /* seal */ false),
+                                m->compress.threshold_bytes,
+                                &m->runtime_storage.metrics);
+                if (r < 0)
+                        return log_warning_errno(r, "Failed to reload runtime journal on reload, ignoring: %m");
+        }
+
+        /* If journal-related configuration, such as SystemMaxUse, SystemMaxFileSize, RuntimeMaxUse, RuntimeMaxFileSize,
+         * were to change, then we can vacuum for the change to take effect. For example, if pre-reload SystemMaxUse=2M,
+         * current usage=1.5M, and the post-reload SystemMaxUse=1M, the vacuum can shrink it to 1M.
+         */
+        manager_vacuum(m, /* verbose */ false);
+
+        return 0;
+}
+
+static int dispatch_reload_signal(sd_event_source *s, const struct signalfd_siginfo *si, void *userdata) {
+        Manager *m = ASSERT_PTR(userdata);
+        int r;
+
+        (void) notify_reloading();
+
+        manager_reload_config(m);
+
+        r = manager_reload_dev_kmsg(m);
+        if (r < 0)
+                return r;
+
+        r = manager_reload_journals(m);
+        if (r < 0)
+                return r;
+
+        log_info("Config file reloaded.");
+        (void) sd_notify(/* unset_environment */ false, NOTIFY_READY_MESSAGE);
+
+        return 0;
+}
+
+int manager_new(Manager **ret, const char *namespace) {
         _cleanup_(manager_freep) Manager *m = NULL;
+        int r;
 
         assert(ret);
 
@@ -2432,37 +2653,10 @@ int manager_new(Manager **ret) {
                 .notify_fd = -EBADF,
                 .forward_socket_fd = -EBADF,
 
-                .compress.enabled = true,
-                .compress.threshold_bytes = UINT64_MAX,
-                .seal = true,
-
-                .set_audit = true,
-
                 .watchdog_usec = USEC_INFINITY,
 
-                .sync_interval_usec = DEFAULT_SYNC_INTERVAL_USEC,
                 .sync_scheduled = false,
 
-                .ratelimit_interval = DEFAULT_RATE_LIMIT_INTERVAL,
-                .ratelimit_burst = DEFAULT_RATE_LIMIT_BURST,
-
-                .forward_to_wall = true,
-                .forward_to_socket = { .sockaddr.sa.sa_family = AF_UNSPEC },
-
-                .max_file_usec = DEFAULT_MAX_FILE_USEC,
-
-                .max_level_store = LOG_DEBUG,
-                .max_level_syslog = LOG_DEBUG,
-                .max_level_kmsg = LOG_NOTICE,
-                .max_level_console = LOG_INFO,
-                .max_level_wall = LOG_EMERG,
-                .max_level_socket = LOG_DEBUG,
-
-                .line_max = DEFAULT_LINE_MAX,
-
-                .runtime_storage.name = "Runtime Journal",
-                .system_storage.name = "System Journal",
-
                 .kmsg_own_ratelimit = {
                         .interval = DEFAULT_KMSG_OWN_INTERVAL,
                         .burst = DEFAULT_KMSG_OWN_BURST,
@@ -2472,11 +2666,17 @@ int manager_new(Manager **ret) {
                 .sigrtmin18_info.memory_pressure_userdata = m,
         };
 
+        r = manager_set_namespace(m, namespace);
+        if (r < 0)
+                return r;
+
+        manager_load_config(m);
+
         *ret = TAKE_PTR(m);
         return 0;
 }
 
-int manager_init(Manager *m, const char *namespace) {
+int manager_init(Manager *m) {
         const char *native_socket, *syslog_socket, *stdout_socket, *varlink_socket, *e;
         _cleanup_fdset_free_ FDSet *fds = NULL;
         int n, r, varlink_fd = -EBADF;
@@ -2484,33 +2684,6 @@ int manager_init(Manager *m, const char *namespace) {
 
         assert(m);
 
-        r = manager_set_namespace(m, namespace);
-        if (r < 0)
-                return r;
-
-        /* By default, only read from /dev/kmsg if are the main namespace */
-        m->read_kmsg = !m->namespace;
-        m->storage = m->namespace ? STORAGE_PERSISTENT : STORAGE_AUTO;
-
-        journal_reset_metrics(&m->system_storage.metrics);
-        journal_reset_metrics(&m->runtime_storage.metrics);
-
-        manager_load_credentials(m);
-        manager_parse_config_file(m);
-
-        if (!m->namespace) {
-                /* Parse kernel command line, but only if we are not a namespace instance */
-                r = proc_cmdline_parse(parse_proc_cmdline_item, m, PROC_CMDLINE_STRIP_RD_PREFIX);
-                if (r < 0)
-                        log_warning_errno(r, "Failed to parse kernel command line, ignoring: %m");
-        }
-
-        if (!!m->ratelimit_interval != !!m->ratelimit_burst) { /* One set to 0 and the other not? */
-                log_debug("Setting both rate limit interval and burst from "USEC_FMT",%u to 0,0",
-                          m->ratelimit_interval, m->ratelimit_burst);
-                m->ratelimit_interval = m->ratelimit_burst = 0;
-        }
-
         e = getenv("RUNTIME_DIRECTORY");
         if (e)
                 m->runtime_directory = strdup(e);

src/journal/journald-manager.h

@@ -55,6 +55,23 @@ typedef struct SeqnumData {
         uint64_t seqnum;
 } SeqnumData;
 
+typedef struct JournalConfig {
+        SocketAddress forward_to_socket;
+        Storage storage;
+
+        bool forward_to_kmsg;
+        bool forward_to_syslog;
+        bool forward_to_console;
+        bool forward_to_wall;
+
+        int max_level_store;
+        int max_level_syslog;
+        int max_level_kmsg;
+        int max_level_console;
+        int max_level_wall;
+        int max_level_socket;
+} JournalConfig;
+
 typedef struct Manager {
         char *namespace;
 
@@ -111,12 +128,6 @@ typedef struct Manager {
         bool sent_notify_ready;
         bool sync_scheduled;
 
-        bool forward_to_kmsg;
-        bool forward_to_syslog;
-        bool forward_to_console;
-        bool forward_to_wall;
-        SocketAddress forward_to_socket;
-
         unsigned n_forward_syslog_missed;
         usec_t last_warn_forward_syslog_missed;
 
@@ -130,14 +141,6 @@ typedef struct Manager {
 
         char *tty_path;
 
-        int max_level_store;
-        int max_level_syslog;
-        int max_level_kmsg;
-        int max_level_console;
-        int max_level_wall;
-        int max_level_socket;
-
-        Storage storage;
         SplitMode split_mode;
 
         MMapCache *mmap;
@@ -183,6 +186,21 @@ typedef struct Manager {
 
         /* Pending synchronization requests with non-zero rqlen counter */
         LIST_HEAD(SyncReq, sync_req_pending_rqlen);
+
+        /* These structs are used to preserve configurations set by credentials and command line.
+                config - main configuration used by journald manager,
+                config_by_cred - configuration set by credentials,
+                config_by_conf - configuration set by configuration file,
+                config_by_cmdline - configuration set by command line.
+           The priority order of the sub-configurations is:
+                config_by_cmdline > config_by_conf > config_by_cred
+           where A > B means that if the two have the same setting,
+           A's value overrides B's value for that setting.
+         */
+        JournalConfig config;
+        JournalConfig config_by_cred;
+        JournalConfig config_by_conf;
+        JournalConfig config_by_cmdline;
 } Manager;
 
 #define MANAGER_MACHINE_ID(s) ((s)->machine_id_field + STRLEN("_MACHINE_ID="))
@@ -209,6 +227,17 @@ void manager_dispatch_message(Manager *m, struct iovec *iovec, size_t n, size_t
 void manager_driver_message_internal(Manager *m, pid_t object_pid, const char *format, ...) _sentinel_;
 #define manager_driver_message(...) manager_driver_message_internal(__VA_ARGS__, NULL)
 
+#define JOURNAL_CONFIG_INIT                                                                     \
+        (JournalConfig) {                                                                       \
+                .forward_to_socket = (SocketAddress) { .sockaddr.sa.sa_family = AF_UNSPEC },    \
+                .storage = _STORAGE_INVALID,                                                    \
+                .max_level_store = -1,                                                          \
+                .max_level_syslog = -1,                                                         \
+                .max_level_kmsg = -1,                                                           \
+                .max_level_console = -1,                                                        \
+                .max_level_wall = -1,                                                           \
+        }
+
 /* gperf lookup function */
 const struct ConfigPerfItem* journald_gperf_lookup(const char *key, GPERF_LEN_TYPE length);
 
@@ -225,8 +254,8 @@ CONFIG_PARSER_PROTOTYPE(config_parse_split_mode);
 const char* split_mode_to_string(SplitMode s) _const_;
 SplitMode split_mode_from_string(const char *s) _pure_;
 
-int manager_new(Manager **ret);
-int manager_init(Manager *m, const char *namespace);
+int manager_new(Manager **ret, const char *namespace);
+int manager_init(Manager *m);
 Manager* manager_free(Manager *m);
 DEFINE_TRIVIAL_CLEANUP_FUNC(Manager*, manager_free);
 void manager_full_sync(Manager *m, bool wait);

src/journal/journald-native.c

@@ -267,16 +267,16 @@ static int manager_process_entry(
                 if (r <= 0)
                         goto finish;
 
-                if (m->forward_to_syslog)
+                if (m->config.forward_to_syslog)
                         manager_forward_syslog(m, syslog_fixup_facility(priority), identifier, message, ucred, tv);
 
-                if (m->forward_to_kmsg)
+                if (m->config.forward_to_kmsg)
                         manager_forward_kmsg(m, priority, identifier, message, ucred);
 
-                if (m->forward_to_console)
+                if (m->config.forward_to_console)
                         manager_forward_console(m, priority, identifier, message, ucred);
 
-                if (m->forward_to_wall)
+                if (m->config.forward_to_wall)
                         manager_forward_wall(m, priority, identifier, message, ucred);
         }
 

src/journal/journald-socket.c

@@ -23,13 +23,13 @@ static int manager_open_forward_socket(Manager *m) {
         assert(m);
 
         /* Noop if there is nothing to do. */
-        if (m->forward_to_socket.sockaddr.sa.sa_family == AF_UNSPEC || m->namespace)
+        if (m->config.forward_to_socket.sockaddr.sa.sa_family == AF_UNSPEC || m->namespace)
                 return 0;
         /* All ready, nothing to do. */
         if (m->forward_socket_fd >= 0)
                 return 1;
 
-        addr = &m->forward_to_socket;
+        addr = &m->config.forward_to_socket;
 
         family = socket_address_family(addr);
 
@@ -86,7 +86,7 @@ int manager_forward_socket(
         assert(n_iovec > 0);
         assert(ts);
 
-        if (LOG_PRI(priority) > m->max_level_socket)
+        if (LOG_PRI(priority) > m->config.max_level_socket)
                 return 0;
 
         r = manager_open_forward_socket(m);

src/journal/journald-stream.c

@@ -251,16 +251,16 @@ static int stdout_stream_log(
         if (r <= 0)
                 return r;
 
-        if (s->forward_to_syslog || s->manager->forward_to_syslog)
+        if (s->forward_to_syslog || s->manager->config.forward_to_syslog)
                 manager_forward_syslog(s->manager, syslog_fixup_facility(priority), s->identifier, p, &s->ucred, NULL);
 
-        if (s->forward_to_kmsg || s->manager->forward_to_kmsg)
+        if (s->forward_to_kmsg || s->manager->config.forward_to_kmsg)
                 manager_forward_kmsg(s->manager, priority, s->identifier, p, &s->ucred);
 
-        if (s->forward_to_console || s->manager->forward_to_console)
+        if (s->forward_to_console || s->manager->config.forward_to_console)
                 manager_forward_console(s->manager, priority, s->identifier, p, &s->ucred);
 
-        if (s->manager->forward_to_wall)
+        if (s->manager->config.forward_to_wall)
                 manager_forward_wall(s->manager, priority, s->identifier, p, &s->ucred);
 
         m = N_IOVEC_META_FIELDS + 7 + client_context_extra_fields_n_iovec(s->context);

src/journal/journald-syslog.c

@@ -127,7 +127,7 @@ static void forward_syslog_raw(
         assert(m);
         assert(buffer);
 
-        if (LOG_PRI(priority) > m->max_level_syslog)
+        if (LOG_PRI(priority) > m->config.max_level_syslog)
                 return;
 
         iovec = IOVEC_MAKE((char *) buffer, buffer_len);
@@ -154,7 +154,7 @@ void manager_forward_syslog(
         assert(priority <= 999);
         assert(message);
 
-        if (LOG_PRI(priority) > m->max_level_syslog)
+        if (LOG_PRI(priority) > m->config.max_level_syslog)
                 return;
 
         /* First: priority field */
@@ -403,16 +403,16 @@ void manager_process_syslog_message(
 
         syslog_parse_identifier(&msg, &identifier, &pid);
 
-        if (m->forward_to_syslog)
+        if (m->config.forward_to_syslog)
                 forward_syslog_raw(m, priority, buf, raw_len, ucred, tv);
 
-        if (m->forward_to_kmsg)
+        if (m->config.forward_to_kmsg)
                 manager_forward_kmsg(m, priority, identifier, msg, ucred);
 
-        if (m->forward_to_console)
+        if (m->config.forward_to_console)
                 manager_forward_console(m, priority, identifier, msg, ucred);
 
-        if (m->forward_to_wall)
+        if (m->config.forward_to_wall)
                 manager_forward_wall(m, priority, identifier, msg, ucred);
 
         mm = N_IOVEC_META_FIELDS + 8 + client_context_extra_fields_n_iovec(context);

src/journal/journald-wall.c

@@ -23,7 +23,7 @@ void manager_forward_wall(
         assert(m);
         assert(message);
 
-        if (LOG_PRI(priority) > m->max_level_wall)
+        if (LOG_PRI(priority) > m->config.max_level_wall)
                 return;
 
         if (ucred) {

src/journal/journald.c

@@ -52,11 +52,11 @@ static int run(int argc, char *argv[]) {
 
         sigbus_install();
 
-        r = manager_new(&m);
+        r = manager_new(&m, namespace);
         if (r < 0)
                 return log_oom();
 
-        r = manager_init(m, namespace);
+        r = manager_init(m);
         if (r < 0)
                 return r;
 

src/kernel-install/50-depmod.install

@@ -33,6 +33,11 @@ case "$COMMAND" in
         exec depmod -a "$KERNEL_VERSION"
         ;;
     remove)
+        [ "$KERNEL_INSTALL_BOOT_ENTRY_TYPE" = "type2" ] || \
+        [ "$KERNEL_INSTALL_BOOT_ENTRY_TYPE" = "type1" ] && \
+        [ -d "/lib/modules/$KERNEL_VERSION/kernel" ] && \
+            echo "Multiple entry types exist, not removing modules.dep or associated files." && \
+            exit 0
         [ "$KERNEL_INSTALL_VERBOSE" -gt 0 ] && \
             echo "Removing /lib/modules/${KERNEL_VERSION}/modules.dep and associated files"
         exec rm -f \

src/kernel-install/90-loaderentry.install.in

@@ -49,6 +49,11 @@ LOADER_ENTRY="$BOOT_ROOT/loader/entries/$ENTRY_TOKEN-$KERNEL_VERSION.conf"
 
 case "$COMMAND" in
     remove)
+        if [ -f "/lib/modules/$KERNEL_VERSION/vmlinuz" ] && [ "$KERNEL_INSTALL_BOOT_ENTRY_TYPE" = "type2" ]; then
+            [ "$KERNEL_INSTALL_VERBOSE" -gt 0 ] && \
+                echo "BOOT_ENTRY_TYPE=type2, not removing loader entries."
+            exit 0
+        fi
         [ "$KERNEL_INSTALL_VERBOSE" -gt 0 ] && \
             echo "Removing ${LOADER_ENTRY%.conf}*.conf"
         exec rm -f \

src/kernel-install/kernel-install.c

@@ -7,6 +7,7 @@
 
 #include "argv-util.h"
 #include "boot-entry.h"
+#include "bootspec.h"
 #include "build.h"
 #include "chase.h"
 #include "conf-files.h"
@@ -88,6 +89,7 @@ typedef struct Context {
         Action action;
         sd_id128_t machine_id;
         bool machine_id_is_random;
+        BootEntryType entry_type;
         KernelImageType kernel_image_type;
         Layout layout;
         char *layout_other;
@@ -427,6 +429,20 @@ static int context_set_initrds(Context *c, char* const* strv) {
         return context_set_path_strv(c, strv, "command line", "initrds", &c->initrds);
 }
 
+static int context_set_entry_type(Context *c, const char *s) {
+        assert(c);
+        BootEntryType e;
+        if (isempty(s) || streq(s, "all")) {
+                c->entry_type = _BOOT_ENTRY_TYPE_INVALID;
+                return 0;
+        }
+        e = boot_entry_type_from_string(s);
+        if (e < 0)
+                return log_error_errno(e, "Invalid entry type: %s", s);
+        c->entry_type = e;
+        return 1;
+}
+
 static int context_load_environment(Context *c) {
         assert(c);
 
@@ -983,6 +999,7 @@ static int context_build_environment(Context *c) {
                                  "KERNEL_INSTALL_LAYOUT",           context_get_layout(c),
                                  "KERNEL_INSTALL_INITRD_GENERATOR", strempty(c->initrd_generator),
                                  "KERNEL_INSTALL_UKI_GENERATOR",    strempty(c->uki_generator),
+                                 "KERNEL_INSTALL_BOOT_ENTRY_TYPE",  boot_entry_type_to_string(c->entry_type),
                                  "KERNEL_INSTALL_STAGING_AREA",     c->staging_area);
         if (r < 0)
                 return log_error_errno(r, "Failed to build environment variables for plugins: %m");
@@ -1487,6 +1504,9 @@ static int help(void) {
                "     --root=PATH               Operate on an alternate filesystem root\n"
                "     --image=PATH              Operate on disk image as filesystem root\n"
                "     --image-policy=POLICY     Specify disk image dissection policy\n"
+               "     --entry-type=type1|type2|all\n"
+               "                               Operate only on the specified bootloader\n"
+               "                               entry type\n"
                "\n"
                "This program may also be invoked as 'installkernel':\n"
                "  installkernel  [OPTIONS...] VERSION VMLINUZ [MAP] [INSTALLATION-DIR]\n"
@@ -1516,6 +1536,7 @@ static int parse_argv(int argc, char *argv[], Context *c) {
                 ARG_ROOT,
                 ARG_IMAGE,
                 ARG_IMAGE_POLICY,
+                ARG_BOOT_ENTRY_TYPE,
         };
         static const struct option options[] = {
                 { "help",                 no_argument,       NULL, 'h'                      },
@@ -1531,6 +1552,7 @@ static int parse_argv(int argc, char *argv[], Context *c) {
                 { "image",                required_argument, NULL, ARG_IMAGE                },
                 { "image-policy",         required_argument, NULL, ARG_IMAGE_POLICY         },
                 { "no-legend",            no_argument,       NULL, ARG_NO_LEGEND            },
+                { "entry-type",           required_argument, NULL, ARG_BOOT_ENTRY_TYPE      },
                 {}
         };
         int t, r;
@@ -1614,6 +1636,12 @@ static int parse_argv(int argc, char *argv[], Context *c) {
                                 return r;
                         break;
 
+                case ARG_BOOT_ENTRY_TYPE:
+                        r = context_set_entry_type(c, optarg);
+                        if (r < 0)
+                                return r;
+                        break;
+
                 case '?':
                         return -EINVAL;
 
@@ -1641,6 +1669,7 @@ static int run(int argc, char* argv[]) {
                 .action = _ACTION_INVALID,
                 .kernel_image_type = KERNEL_IMAGE_TYPE_UNKNOWN,
                 .layout = _LAYOUT_INVALID,
+                .entry_type = _BOOT_ENTRY_TYPE_INVALID,
                 .entry_token_type = BOOT_ENTRY_TOKEN_AUTO,
         };
         _cleanup_(loop_device_unrefp) LoopDevice *loop_device = NULL;

src/kernel-install/test-kernel-install.sh

@@ -97,6 +97,13 @@ test ! -e "$entry"
 test ! -e "$BOOT_ROOT/the-token/1.1.1/linux"
 test ! -e "$BOOT_ROOT/the-token/1.1.1/initrd"
 
+# Test --entry-type options
+"$kernel_install" -v add 1.1.1 "$D/sources/linux" "$D/sources/initrd"
+"$kernel_install" -v remove 1.1.1 --entry-type=type1
+test ! -e "$entry"
+test ! -e "$BOOT_ROOT/the-token/1.1.1/linux"
+test ! -e "$BOOT_ROOT/the-token/1.1.1/initrd"
+
 # Invoke kernel-install as installkernel
 ln -s --relative -v "$kernel_install" "$D/sources/installkernel"
 "$D/sources/installkernel" -v 1.1.2 "$D/sources/linux" System.map /somedirignored

src/libsystemd/sd-journal/journal-file.c

@@ -4082,6 +4082,54 @@ static void journal_default_metrics(JournalMetrics *m, int fd, bool compact) {
                   m->n_max_files);
 }
 
+static uint64_t get_compress_threshold_bytes(uint64_t compress_threshold_bytes) {
+        return compress_threshold_bytes == UINT64_MAX ?
+                DEFAULT_COMPRESS_THRESHOLD :
+                MAX(MIN_COMPRESS_THRESHOLD, compress_threshold_bytes);
+}
+
+static int set_metrics(JournalFile *f, JournalMetrics *metrics, JournalFile *template) {
+        assert(f);
+        int r;
+
+        if (!journal_file_writable(f))
+                return 0;
+
+        if (metrics) {
+                journal_default_metrics(metrics, f->fd, JOURNAL_HEADER_COMPACT(f->header));
+                f->metrics = *metrics;
+        } else if (template)
+                f->metrics = template->metrics;
+
+        r = journal_file_refresh_header(f);
+        if (r < 0)
+                return log_error_errno(r, "Failed to refresh journal file header. Error to be handled by caller.");
+
+        return 0;
+}
+
+int journal_file_reload(
+                JournalFile *f,
+                JournalFileFlags file_flags,
+                uint64_t compress_threshold_bytes,
+                JournalMetrics *metrics) {
+
+        assert(f);
+        assert((file_flags & ~_JOURNAL_FILE_FLAGS_ALL) == 0);
+        assert(metrics);
+
+        int r;
+
+        f->compress_threshold_bytes = get_compress_threshold_bytes(compress_threshold_bytes);
+
+        r = set_metrics(f, metrics, /* template */ NULL);
+        if (r < 0)
+                /* Journal file header failed to be rotated. The changes may not have taken effect in this case. */
+                return r;
+
+        return 0;
+}
+
 int journal_file_open(
                 int fd,
                 const char *fname,
@@ -4121,9 +4169,7 @@ int journal_file_open(
                 .fd = fd,
                 .mode = mode,
                 .open_flags = open_flags,
-                .compress_threshold_bytes = compress_threshold_bytes == UINT64_MAX ?
-                                            DEFAULT_COMPRESS_THRESHOLD :
-                                            MAX(MIN_COMPRESS_THRESHOLD, compress_threshold_bytes),
+                .compress_threshold_bytes = get_compress_threshold_bytes(compress_threshold_bytes),
                 .strict_order = FLAGS_SET(file_flags, JOURNAL_STRICT_ORDER),
                 .newest_boot_id_prioq_idx = PRIOQ_IDX_NULL,
                 .last_direction = _DIRECTION_INVALID,
@@ -4232,17 +4278,9 @@ int journal_file_open(
         }
 #endif
 
-        if (journal_file_writable(f)) {
-                if (metrics) {
-                        journal_default_metrics(metrics, f->fd, JOURNAL_HEADER_COMPACT(f->header));
-                        f->metrics = *metrics;
-                } else if (template)
-                        f->metrics = template->metrics;
-
-                r = journal_file_refresh_header(f);
-                if (r < 0)
-                        goto fail;
-        }
+        r = set_metrics(f, metrics, template);
+        if (r < 0)
+                goto fail;
 
 #if HAVE_GCRYPT
         r = journal_file_hmac_setup(f);

src/libsystemd/sd-journal/journal-file.h

@@ -149,6 +149,12 @@ int journal_file_open(
                 JournalFile *template,
                 JournalFile **ret);
 
+int journal_file_reload(
+        JournalFile *f,
+        JournalFileFlags file_flags,
+        uint64_t compress_threshold_bytes,
+        JournalMetrics *metrics);
+
 int journal_file_set_offline_thread_join(JournalFile *f);
 JournalFile* journal_file_close(JournalFile *j);
 int journal_file_fstat(JournalFile *f);

src/nspawn/nspawn.c

@@ -1907,14 +1907,14 @@ static int setup_timezone(const char *dest) {
                         log_debug_errno(r, "Timezone %s does not exist (or is not accessible) in container, not creating symlink: %m", z);
                 else {
                         if (unlink(where) < 0 && errno != ENOENT) {
-                                log_full_errno(IN_SET(errno, EROFS, EACCES, EPERM) ? LOG_DEBUG : LOG_WARNING, /* Don't complain on read-only images */
+                                log_full_errno(ERRNO_IS_FS_WRITE_REFUSED(errno) ? LOG_DEBUG : LOG_WARNING, /* Don't complain on read-only images */
                                                errno, "Failed to remove existing timezone info %s in container, ignoring: %m", where);
                                 return 0;
                         }
 
                         what = strjoina("../usr/share/zoneinfo/", z);
                         if (symlink(what, where) < 0) {
-                                log_full_errno(IN_SET(errno, EROFS, EACCES, EPERM) ? LOG_DEBUG : LOG_WARNING,
+                                log_full_errno(ERRNO_IS_FS_WRITE_REFUSED(errno) ? LOG_DEBUG : LOG_WARNING,
                                                errno, "Failed to correct timezone of container, ignoring: %m");
                                 return 0;
                         }
@@ -1949,7 +1949,7 @@ static int setup_timezone(const char *dest) {
                 /* If mounting failed, try to copy */
                 r = copy_file_atomic("/etc/localtime", where, 0644, COPY_REFLINK|COPY_REPLACE);
                 if (r < 0) {
-                        log_full_errno(IN_SET(r, -EROFS, -EACCES, -EPERM) ? LOG_DEBUG : LOG_WARNING, r,
+                        log_full_errno(ERRNO_IS_NEG_FS_WRITE_REFUSED(r) ? LOG_DEBUG : LOG_WARNING, r,
                                        "Failed to copy /etc/localtime to %s, ignoring: %m", where);
                         return 0;
                 }
@@ -2085,7 +2085,7 @@ static int setup_resolv_conf(const char *dest) {
                  * If the disk image is read-only, there's also no point in complaining.
                  */
                 log_full_errno(!IN_SET(RESOLV_CONF_COPY_HOST, RESOLV_CONF_COPY_STATIC, RESOLV_CONF_COPY_UPLINK, RESOLV_CONF_COPY_STUB) &&
-                               IN_SET(r, -ELOOP, -EROFS, -EACCES, -EPERM) ? LOG_DEBUG : LOG_WARNING, r,
+                               (r == -ELOOP || ERRNO_IS_NEG_FS_WRITE_REFUSED(r)) ? LOG_DEBUG : LOG_WARNING, r,
                                "Failed to copy /etc/resolv.conf to %s, ignoring: %m", where);
                 return 0;
         }

src/shared/binfmt-util.c

@@ -11,7 +11,7 @@
 #include "missing_magic.h"
 #include "stat-util.h"
 
-int binfmt_mounted(void) {
+int binfmt_mounted_and_writable(void) {
         _cleanup_close_ int fd = -EBADF;
         int r;
 
@@ -25,7 +25,13 @@ int binfmt_mounted(void) {
         if (r <= 0)
                 return r;
 
-        return access_fd(fd, W_OK) >= 0;
+        r = access_fd(fd, W_OK);
+        if (ERRNO_IS_NEG_FS_WRITE_REFUSED(r))
+                return false;
+        if (r < 0)
+                return r;
+
+        return true;
 }
 
 int disable_binfmt(void) {
@@ -37,7 +43,7 @@ int disable_binfmt(void) {
          * We are a bit careful here, since binfmt_misc might still be an autofs which we don't want to
          * trigger. */
 
-        r = binfmt_mounted();
+        r = binfmt_mounted_and_writable();
         if (r < 0)
                 return log_warning_errno(r, "Failed to determine whether binfmt_misc is mounted: %m");
         if (r == 0) {

src/shared/binfmt-util.h

@@ -1,5 +1,5 @@
 /* SPDX-License-Identifier: LGPL-2.1-or-later */
 #pragma once
 
-int binfmt_mounted(void);
+int binfmt_mounted_and_writable(void);
 int disable_binfmt(void);

src/shared/bootspec.c

@@ -48,7 +48,7 @@ static const char* const boot_entry_type_table[_BOOT_ENTRY_TYPE_MAX] = {
         [BOOT_ENTRY_AUTO]   = "auto",
 };
 
-DEFINE_STRING_TABLE_LOOKUP_TO_STRING(boot_entry_type, BootEntryType);
+DEFINE_STRING_TABLE_LOOKUP(boot_entry_type, BootEntryType);
 
 static const char* const boot_entry_source_description_table[_BOOT_ENTRY_SOURCE_MAX] = {
         [BOOT_ENTRY_ESP]      = "EFI System Partition",

src/shared/bootspec.h

@@ -93,6 +93,7 @@ typedef struct BootConfig {
 
 const char* boot_entry_type_description_to_string(BootEntryType) _const_;
 const char* boot_entry_type_to_string(BootEntryType) _const_;
+BootEntryType boot_entry_type_from_string(const char *s) _pure_;
 
 const char* boot_entry_source_description_to_string(BootEntrySource) _const_;
 const char* boot_entry_source_to_string(BootEntrySource) _const_;

src/shared/loop-util.c

@@ -660,7 +660,7 @@ int loop_device_make_by_path_at(
                 r = fd;
 
                 /* Retry read-only? */
-                if (open_flags >= 0 || !(ERRNO_IS_PRIVILEGE(r) || r == -EROFS))
+                if (open_flags >= 0 || !ERRNO_IS_NEG_FS_WRITE_REFUSED(r))
                         return r;
 
                 fd = xopenat(dir_fd, path, basic_flags|direct_flags|O_RDONLY);

src/shutdown/shutdown.c

@@ -447,7 +447,7 @@ int main(int argc, char *argv[]) {
                 (void) sync_with_progress(-EBADF);
 
         disable_coredumps();
-        disable_binfmt();
+        (void) disable_binfmt();
 
         log_info("Sending SIGTERM to remaining processes...");
         broadcast_signal(SIGTERM, true, true, arg_timeout);

src/ssh-generator/ssh-generator.c

@@ -321,7 +321,7 @@ static int add_export_unix_socket(
                         log_debug("Container manager does not provide /run/host/unix-export/ mount, not binding AF_UNIX socket there.");
                         return 0;
                 }
-                if (errno == EROFS || ERRNO_IS_PRIVILEGE(errno)) {
+                if (ERRNO_IS_FS_WRITE_REFUSED(errno)) {
                         log_debug("Container manager does not provide write access to /run/host/unix-export/, not binding AF_UNIX socket there.");
                         return 0;
                 }

src/sysctl/sysctl.c

@@ -95,7 +95,7 @@ static int sysctl_write_or_warn(const char *key, const char *value, bool ignore_
                  * permission problem here, since that's how container managers usually protected their
                  * sysctls.)
                  * In all other cases log an error and make the tool fail. */
-                if (ignore_failure || (!arg_strict && (r == -EROFS || ERRNO_IS_PRIVILEGE(r))))
+                if (ignore_failure || (!arg_strict && ERRNO_IS_NEG_FS_WRITE_REFUSED(r)))
                         log_debug_errno(r, "Couldn't write '%s' to '%s', ignoring: %m", value, key);
                 else if (ignore_enoent && r == -ENOENT)
                         log_warning_errno(r, "Couldn't write '%s' to '%s', ignoring: %m", value, key);

src/test/test-binfmt-util.c

@@ -3,8 +3,8 @@
 #include "binfmt-util.h"
 #include "tests.h"
 
-TEST(binfmt_mounted) {
-        ASSERT_OK(binfmt_mounted());
+TEST(binfmt_mounted_and_writable) {
+        ASSERT_OK(binfmt_mounted_and_writable());
 }
 
 DEFINE_TEST_MAIN(LOG_DEBUG);

src/test/test-cgroup.c

@@ -64,10 +64,8 @@ TEST(cg_create) {
         (void) cg_trim(test_b, /* delete_root= */ true);
 
         r = cg_create(test_a);
-        if (IN_SET(r, -EPERM, -EACCES, -EROFS)) {
-                log_info_errno(r, "Skipping %s: %m", __func__);
-                return;
-        }
+        if (ERRNO_IS_NEG_FS_WRITE_REFUSED(r))
+                return (void) log_tests_skipped_errno(r, "%s: Failed to create cgroup %s", __func__, test_a);
 
         ASSERT_OK_EQ(r, 1);
         ASSERT_OK_ZERO(cg_create(test_a));

src/test/test-procfs-util.c

@@ -51,7 +51,7 @@ int main(int argc, char *argv[]) {
                 log_info("Reducing limit by one to %"PRIu64"…", limit-1);
 
                 r = procfs_tasks_set_limit(limit-1);
-                if (IN_SET(r, -ENOENT, -EROFS) || ERRNO_IS_PRIVILEGE(r))
+                if (r == -ENOENT || ERRNO_IS_NEG_FS_WRITE_REFUSED(r))
                         return log_tests_skipped_errno(r, "can't set tasks limit");
                 assert_se(r >= 0);
 

src/test/test-sysctl-util.c

@@ -54,14 +54,14 @@ TEST(sysctl_read) {
         assert_se(STR_IN_SET(s, "0", "1"));
 
         r = sysctl_write_ip_property(AF_INET, "lo", "forwarding", s, NULL);
-        assert_se(r >= 0 || ERRNO_IS_PRIVILEGE(r) || r == -EROFS);
+        assert_se(r >= 0 || ERRNO_IS_NEG_FS_WRITE_REFUSED(r));
         s = mfree(s);
 
         assert_se(sysctl_read_ip_property(AF_INET, NULL, "ip_forward", &s));
         assert_se(STR_IN_SET(s, "0", "1"));
 
         r = sysctl_write_ip_property(AF_INET, NULL, "ip_forward", s, NULL);
-        assert_se(r >= 0 || ERRNO_IS_PRIVILEGE(r) || r == -EROFS);
+        assert_se(r >= 0 || ERRNO_IS_NEG_FS_WRITE_REFUSED(r));
         s = mfree(s);
 
         assert_se(sysctl_read("kernel/hostname", &s) >= 0);
@@ -69,7 +69,7 @@ TEST(sysctl_read) {
         ASSERT_STREQ(s, u.nodename);
 
         r = sysctl_write("kernel/hostname", s);
-        assert_se(r >= 0 || ERRNO_IS_PRIVILEGE(r) || r == -EROFS);
+        assert_se(r >= 0 || ERRNO_IS_NEG_FS_WRITE_REFUSED(r));
 }
 
 DEFINE_TEST_MAIN(LOG_INFO);

src/userdb/userdbctl.c

@@ -1181,18 +1181,32 @@ static int ssh_authorized_keys(int argc, char *argv[], void *userdata) {
         return r;
 }
 
-static int load_credential_one(int credential_dir_fd, const char *name, int userdb_dir_fd) {
+static int load_credential_one(
+                int credential_dir_fd,
+                const char *name,
+                int userdb_dir_persist_fd,
+                int userdb_dir_transient_fd) {
+
         int r;
 
         assert(credential_dir_fd >= 0);
         assert(name);
-        assert(userdb_dir_fd >= 0);
+        assert(userdb_dir_persist_fd >= 0);
+        assert(userdb_dir_transient_fd >= 0);
 
-        const char *user = startswith(name, "userdb.user.");
-        const char *group = startswith(name, "userdb.group.");
+        const char *suffix = startswith(name, "userdb.");
+        if (!suffix)
+                return 0;
+
+        const char *transient = startswith(suffix, "transient."),
+                *user = startswith(transient ?: suffix, "user."),
+                *group = startswith(transient ?: suffix, "group.");
         if (!user && !group)
                 return 0;
 
+        const char *userdb_dir = transient ? "/run/userdb" : "/etc/userdb";
+        int userdb_dir_fd = transient ? userdb_dir_transient_fd : userdb_dir_persist_fd;
+
         _cleanup_(sd_json_variant_unrefp) sd_json_variant *v = NULL;
         unsigned line = 0, column = 0;
         r = sd_json_parse_file_at(NULL, credential_dir_fd, name, SD_JSON_PARSE_SENSITIVE, &v, &line, &column);
@@ -1350,12 +1364,12 @@ static int load_credential_one(int credential_dir_fd, const char *name, int user
 
         r = write_string_file_at(userdb_dir_fd, fn, formatted, WRITE_STRING_FILE_CREATE|WRITE_STRING_FILE_ATOMIC);
         if (r < 0)
-                return log_error_errno(r, "Failed to write JSON record to /etc/userdb/%s: %m", fn);
+                return log_error_errno(r, "Failed to write JSON record to %s/%s: %m", userdb_dir, fn);
 
         if (symlinkat(fn, userdb_dir_fd, link) < 0)
                 return log_error_errno(errno, "Failed to create symlink from %s to %s: %m", link, fn);
 
-        log_info("Installed /etc/userdb/%s from credential.", fn);
+        log_info("Installed %s/%s from credential.", userdb_dir, fn);
 
         if ((ur && !sd_json_variant_is_blank_object(ur_privileged->json)) ||
             (gr && !sd_json_variant_is_blank_object(gr_privileged->json))) {
@@ -1371,7 +1385,7 @@ static int load_credential_one(int credential_dir_fd, const char *name, int user
 
                 r = write_string_file_at(userdb_dir_fd, fn, formatted, WRITE_STRING_FILE_CREATE|WRITE_STRING_FILE_ATOMIC|WRITE_STRING_FILE_MODE_0600);
                 if (r < 0)
-                        return log_error_errno(r, "Failed to write JSON record to /etc/userdb/%s: %m", fn);
+                        return log_error_errno(r, "Failed to write JSON record to %s/%s: %m", userdb_dir, fn);
 
                 link = mfree(link);
 
@@ -1386,7 +1400,7 @@ static int load_credential_one(int credential_dir_fd, const char *name, int user
                 if (symlinkat(fn, userdb_dir_fd, link) < 0)
                         return log_error_errno(errno, "Failed to create symlink from %s to %s: %m", link, fn);
 
-                log_info("Installed /etc/userdb/%s from credential.", fn);
+                log_info("Installed %s/%s from credential.", userdb_dir, fn);
         }
 
         if (ur)
@@ -1399,7 +1413,7 @@ static int load_credential_one(int credential_dir_fd, const char *name, int user
                         if (fd < 0)
                                 return log_error_errno(errno, "Failed to create %s: %m", membership);
 
-                        log_info("Installed /etc/userdb/%s from credential.", membership);
+                        log_info("Installed %s/%s from credential.", userdb_dir, membership);
                 }
         else
                 STRV_FOREACH(u, gr->members) {
@@ -1411,7 +1425,7 @@ static int load_credential_one(int credential_dir_fd, const char *name, int user
                         if (fd < 0)
                                 return log_error_errno(errno, "Failed to create %s: %m", membership);
 
-                        log_info("Installed /etc/userdb/%s from credential.", membership);
+                        log_info("Installed %s/%s from credential.", userdb_dir, membership);
                 }
 
         if (ur && user_record_disposition(ur) == USER_REGULAR) {
@@ -1461,13 +1475,21 @@ static int load_credentials(int argc, char *argv[], void *userdata) {
         if (r < 0)
                 return log_error_errno(r, "Failed to enumerate credentials: %m");
 
-        _cleanup_close_ int userdb_dir_fd = xopenat_full(
-                AT_FDCWD, "/etc/userdb",
-                /* open_flags= */ O_DIRECTORY|O_CREAT|O_CLOEXEC,
-                /* xopen_flags= */ XO_LABEL,
-                /* mode= */ 0755);
-        if (userdb_dir_fd < 0)
-                return log_error_errno(userdb_dir_fd, "Failed to open %s: %m", "/etc/userdb/");
+        _cleanup_close_ int userdb_persist_dir_fd = xopenat_full(
+                        AT_FDCWD, "/etc/userdb",
+                        /* open_flags= */ O_DIRECTORY|O_CREAT|O_CLOEXEC,
+                        /* xopen_flags= */ XO_LABEL,
+                        /* mode= */ 0755);
+        if (userdb_persist_dir_fd < 0)
+                return log_error_errno(userdb_persist_dir_fd, "Failed to open /etc/userdb/: %m");
+
+        _cleanup_close_ int userdb_transient_dir_fd = xopenat_full(
+                        AT_FDCWD, "/run/userdb",
+                        /* open_flags= */ O_DIRECTORY|O_CREAT|O_CLOEXEC,
+                        /* xopen_flags= */ XO_LABEL,
+                        /* mode= */ 0755);
+        if (userdb_transient_dir_fd < 0)
+                return log_error_errno(userdb_transient_dir_fd, "Failed to open /run/userdb/: %m");
 
         FOREACH_ARRAY(i, des->entries, des->n_entries) {
                 struct dirent *de = *i;
@@ -1475,7 +1497,11 @@ static int load_credentials(int argc, char *argv[], void *userdata) {
                 if (de->d_type != DT_REG)
                         continue;
 
-                RET_GATHER(r, load_credential_one(credential_dir_fd, de->d_name, userdb_dir_fd));
+                RET_GATHER(r, load_credential_one(
+                                credential_dir_fd,
+                                de->d_name,
+                                userdb_persist_dir_fd,
+                                userdb_transient_dir_fd));
         }
 
         return r;

test/units/TEST-04-JOURNAL.journal-reload.sh

@@ -0,0 +1,316 @@
+#!/usr/bin/env bash
+# SPDX-License-Identifier: LGPL-2.1-or-later
+# shellcheck disable=SC2016
+set -eux
+set -o pipefail
+
+MACHINE_ID="$(</etc/machine-id)"
+TEST_MSG_PREFIX="JOURNAL-RELOAD TEST"
+SYSLOG_ID="$(systemd-id128 new)"
+
+write_to_journal() {
+    local rand_val
+
+    rand_val=$(head /dev/urandom | tr -dc A-Za-z0-9 | head -c 14)
+    echo "$TEST_MSG_PREFIX $rand_val" | systemd-cat -t "$SYSLOG_ID"
+    journalctl --sync
+
+    echo "$rand_val"
+}
+
+verify_archived_journals() {
+    local msg="$1"
+    local journal_path_prefix="$2"
+    local journal_dir="/$journal_path_prefix/log/journal/$MACHINE_ID"
+
+    for journal_file in "$journal_dir"/*; do
+        filename=$(basename -- "$journal_file")
+        if [[ "$filename" != "system.journal" ]]; then
+            if journalctl --file="$journal_file" | grep -q "$msg"; then
+                echo "Message present in archived journal: $filename"
+                exit 0
+            fi
+        fi
+    done
+
+    exit 1
+}
+
+# shellcheck disable=SC2317
+verify_journal() {
+    local msg="$1"
+    local entry_expected="$2"
+    local test_name="$3"
+    local journal_path_prefix="$4"
+
+    local path="/$journal_path_prefix/log/journal/$MACHINE_ID/system.journal"
+
+    if [ ! -e "$path" ] || ! grep -Fxq "MESSAGE=$TEST_MSG_PREFIX $msg" "$path"; then
+        if [ "$entry_expected" == true ]; then
+            echo "$test_name ERROR: Message not present in $journal_path_prefix journal. Checking archived journals..."
+            if ! verify_archived_journals "$msg" "$journal_path_prefix"; then
+                echo "$test_name ERROR: Message also not present in archived journals"
+                cleanup
+                exit 1
+            fi
+
+        fi
+    else
+        if [ "$entry_expected" == false ]; then
+            echo "$test_name ERROR: Message present in $journal_path_prefix journal"
+            cleanup
+            exit 1
+        fi
+    fi
+}
+
+verify_journals() {
+    local msg="$1"
+    local runtime_expected="$2"
+    local system_expected="$3"
+    local test_name="$4"
+
+    local failed=false
+
+    if ! verify_journal "$msg" "$runtime_expected" "$test_name" "run"; then
+        failed=true
+    fi
+
+    if ! verify_journal "$msg" "$system_expected" "$test_name" "var"; then
+        failed=true
+    fi
+
+    if [ "$failed" == true ]; then
+        cleanup
+        exit 1
+    fi
+}
+
+get_num_archived_journals() {
+    local journal_path_prefix="$1"
+
+    local journal_dir="/$journal_path_prefix/log/journal/$MACHINE_ID/"
+    num_journal_files=$(find "$journal_dir" -type f -name "*.journal" ! -name "system.journal" | wc -l)
+
+    echo "$num_journal_files"
+}
+
+cleanup() {
+    rm /run/systemd/journald.conf.d/reload.conf
+    journalctl --vacuum-size=1M
+    systemctl daemon-reload
+    systemctl reload systemd-journald.service
+}
+
+# Start clean slate.
+mkdir -p /run/systemd/journald.conf.d
+cat <<EOF >/run/systemd/journald.conf.d/reload.conf
+[Journal]
+Storage=persistent
+EOF
+
+systemctl daemon-reload
+systemctl restart systemd-journald.service
+
+# Add entries in system.
+journalctl --flush
+rand_val1=$(write_to_journal)
+verify_journals "$rand_val1" false true "Confirming test setup after flush."
+
+# Reload journald (persistent->persistent)
+systemctl reload systemd-journald.service
+
+# Reload should persist persistent journal.
+verify_journals "$rand_val1" false true "Persistent->Persistent System Reload: "
+
+rand_val1=$(write_to_journal)
+verify_journals "$rand_val1" false true "Persistent->Persistent System Post-Reload: "
+
+# Add entries in runtime
+journalctl --relinquish
+rand_val2=$(write_to_journal)
+verify_journals "$rand_val2" true false "Confirming test setup after relinquish."
+
+# Reload journald (persistent->persistent)
+systemctl reload systemd-journald.service
+
+# System journal entries should stay in system journal, runtime in runtime.
+verify_journals "$rand_val1" false true "Persistent->Persistent Runtime Reload 1: "
+verify_journals "$rand_val2" true false "Persistent->Persistent Runtime Reload 2: "
+
+# Write new message and confirm it's written to runtime.
+rand_val=$(write_to_journal)
+verify_journals "$rand_val" true false "Persistent->Persistent New Message After Reload: "
+
+# Flush and confirm that messages are written to system.
+journalctl --flush
+rand_val=$(write_to_journal)
+verify_journals "$rand_val" false true "Persistent->Volatile New Message Before Reload: "
+
+# Test persistent->volatile
+cat <<EOF >/run/systemd/journald.conf.d/reload.conf
+[Journal]
+Storage=volatile
+EOF
+
+# Confirm old message exists where it was written to (storage->storage).
+systemctl reload systemd-journald.service
+verify_journals "$rand_val" false true "Persistent->Volatile Reload: "
+
+# Confirm that messages are written to only runtime journal.
+rand_val=$(write_to_journal)
+verify_journals "$rand_val" true false "Persistent->Volatile New Message After Reload: "
+
+# Test volatile works and logs are NOT getting written to system journal despite flush.
+journalctl --flush
+rand_val=$(write_to_journal)
+verify_journals "$rand_val" true false "Persistent->Volatile New Message After Flush: "
+
+# Test that the new limits (e.g., RuntimeMaxUse) take effect on reload.
+# Write 1M of data to runtime journal
+max_size=1048656 # (1 * 1024 * 1024) = 1048576, but centos has a different minimum value for some reason.
+set +x
+dd if=/dev/urandom bs=1M count=5 | base64 | systemd-cat -t "$SYSLOG_ID"
+set -x
+journalctl --vacuum-size=2M
+
+total_size=$(du -sb "/run/log/journal/$MACHINE_ID" | cut -f1)
+if [ "$total_size" -lt "$max_size" ]; then
+    echo "ERROR: Journal size does not exceed RuntimeMaxUse limit"
+    cleanup
+    exit 1
+fi
+
+# Reload with RuntimeMaxUse=1M.
+cat <<EOF >/run/systemd/journald.conf.d/reload.conf
+[Journal]
+Storage=volatile
+RuntimeMaxUse=1M
+EOF
+
+# systemctl daemon-reload
+systemctl reload systemd-journald.service
+
+sleep 15 # Wait for RuntimeMaxUse change to take effect.
+
+# Confirm that runtime journal size shrunk to <=1M.
+total_size=$(du -sb "/run/log/journal/$MACHINE_ID" | cut -f1)
+if [ "$total_size" -gt "$max_size" ]; then
+    echo "ERROR: Journal size exceeds RuntimeMaxUse limit"
+    cleanup
+    exit 1
+fi
+
+# Prepare for volatile->persistent by getting rid of runtime limit. Otherwise, it will not write.
+cat <<EOF >/run/systemd/journald.conf.d/reload.conf
+[Journal]
+Storage=volatile
+EOF
+systemctl daemon-reload
+systemctl reload systemd-journald.service
+sleep 15 # Wait for RuntimeMaxUse change to take effect.
+
+journalctl --vacuum-size=1M
+sleep 5
+
+rand_val=$(write_to_journal)
+verify_journals "$rand_val" true false "Volatile->Persistent New Message Before Reload: "
+
+# Reload volatile->persistent
+cat <<EOF >/run/systemd/journald.conf.d/reload.conf
+[Journal]
+Storage=persistent
+EOF
+
+systemctl reload systemd-journald.service
+
+# Confirm that previous message is still in runtime journal.
+verify_journals "$rand_val" true false "Volatile->Persistent Reload: "
+
+# Confirm that new messages are written to runtime journal.
+rand_val=$(write_to_journal)
+verify_journals "$rand_val" true false "Volatile->Persistent New Message After Reload: "
+
+# Confirm that flushing writes to system journal.
+journalctl --flush
+verify_journals "$rand_val" false true "Volatile->Persistent New Message After Flush: "
+
+set +x
+dd if=/dev/urandom bs=1M count=5 | base64 | systemd-cat -t "$SYSLOG_ID"
+set -x
+
+max_size=$((2 * 1024 * 1024))
+total_size=$(du -sb "/var/log/journal/$MACHINE_ID" | cut -f1)
+if [ "$total_size" -lt "$max_size" ]; then
+    echo "ERROR: Journal size does not exceed SystemMaxUse limit"
+    cleanup
+    exit 1
+fi
+
+# Ensure reloading without limit does not interfere with SystemMaxUse test.
+systemctl reload systemd-journald.service
+total_size=$(du -sb "/var/log/journal/$MACHINE_ID" | cut -f1)
+if [ "$total_size" -lt "$max_size" ]; then
+    echo "ERROR: Journal size does not exceed SystemMaxUse limit"
+    cleanup
+    exit 1
+fi
+
+# Write to storage to prepare for SystemMaxFiles test.
+journalctl --flush
+
+num_var_journals=$(get_num_archived_journals "var")
+limit_var_journals=3
+if [ "$num_var_journals" -lt "$limit_var_journals" ]; then
+    echo "Creating archive files."
+    for (( i=0; i<=num_var_journals; i++ ))
+    do
+        echo "$TEST_MSG_PREFIX" | systemd-cat -t "$SYSLOG_ID"
+        journalctl --rotate
+    done
+
+    num_var_journals=$(get_num_archived_journals "var")
+    if [ "$num_var_journals" -lt "$limit_var_journals" ]; then
+        echo "ERROR: Number of journal files in /var/log/journal/$MACHINE_ID/ is less than $limit_var_journals"
+        cleanup
+        exit 1
+    fi
+fi
+
+# Reload with less SystemMaxUse and SystemMaxFiles.
+cat <<EOF >/run/systemd/journald.conf.d/reload.conf
+[Journal]
+Storage=persistent
+RuntimeMaxUse=2M
+SystemMaxUse=2M
+SystemMaxFiles=3
+EOF
+
+systemctl daemon-reload
+systemctl reload systemd-journald.service
+
+# New system journal needs to be created with the new configuration for change to take effect.
+journalctl --flush
+
+# Check SystemMaxFiles
+num_var_journals=$(get_num_archived_journals "var")
+if [ "$num_var_journals" -gt "$limit_var_journals" ]; then
+    echo "ERROR: Number of journal files in /var/log/journal/$MACHINE_ID/ is greater than $limit_var_journals"
+    cleanup
+    exit 1
+fi
+
+sleep 15
+
+# Check SystemMaxUse
+total_size=$(du -sb "/var/log/journal/$MACHINE_ID" | cut -f1)
+if [ "$total_size" -gt "$max_size" ]; then
+    echo "ERROR: Journal size exceeds SystemMaxUse limit"
+    cleanup
+    exit 1
+fi
+
+rm /run/systemd/journald.conf.d/reload.conf
+journalctl --vacuum-size=1M
+systemctl daemon-reload
+systemctl reload systemd-journald.service

units/systemd-journald.service.in

@@ -57,7 +57,7 @@ StandardOutput=null
 SystemCallArchitectures=native
 SystemCallErrorNumber=EPERM
 SystemCallFilter=@system-service
-Type=notify
+Type=notify-reload
 PassEnvironment=TERM
 {{SERVICE_WATCHDOG}}
 

units/systemd-userdb-load-credentials.service

@@ -26,6 +26,8 @@ RemainAfterExit=yes
 ExecStart=userdbctl load-credentials
 ImportCredential=userdb.user.*
 ImportCredential=userdb.group.*
+ImportCredential=userdb.transient.user.*
+ImportCredential=userdb.transient.group.*
 
 [Install]
 WantedBy=sysinit.target