1systemd System and Service Manager 2 3WEB SITE: 4 https://systemd.io 5 6GIT: 7 git@github.com:systemd/systemd.git 8 https://github.com/systemd/systemd 9 10MAILING LIST: 11 https://lists.freedesktop.org/mailman/listinfo/systemd-devel 12 13IRC: 14 #systemd on irc.libera.chat 15 16BUG REPORTS: 17 https://github.com/systemd/systemd/issues 18 19OLDER DOCUMENTATION: 20 21 http://0pointer.de/blog/projects/systemd.html 22 https://www.freedesktop.org/wiki/Software/systemd 23 24AUTHOR: 25 Lennart Poettering 26 Kay Sievers 27 ...and many others 28 29LICENSE: 30 LGPL-2.1-or-later for all code, exceptions noted in LICENSES/README.md 31 32REQUIREMENTS: 33 Linux kernel ≥ 3.15 34 ≥ 4.3 for ambient capabilities 35 ≥ 4.5 for pids controller in cgroup v2 36 ≥ 4.6 for cgroup namespaces 37 ≥ 4.9 for RENAME_NOREPLACE support in vfat 38 ≥ 4.10 for cgroup-bpf egress and ingress hooks 39 ≥ 4.15 for cgroup-bpf device hook and cpu controller in cgroup v2 40 ≥ 4.17 for cgroup-bpf socket address hooks 41 ≥ 4.20 for PSI (used by systemd-oomd) 42 ≥ 5.3 for bounded loops in BPF program 43 ≥ 5.4 for signed Verity images 44 ≥ 5.7 for BPF links and the BPF LSM hook 45 46 Kernel versions below 4.15 have significant gaps in functionality and 47 are not recommended for use with this version of systemd. Taint flag 48 'old-kernel' will be set. Systemd will most likely still function, but 49 upstream support and testing are limited. 50 51 Kernel Config Options: 52 CONFIG_DEVTMPFS 53 CONFIG_CGROUPS (it is OK to disable all controllers) 54 CONFIG_INOTIFY_USER 55 CONFIG_SIGNALFD 56 CONFIG_TIMERFD 57 CONFIG_EPOLL 58 CONFIG_UNIX (it requires CONFIG_NET, but every other flag in it is not necessary) 59 CONFIG_SYSFS 60 CONFIG_PROC_FS 61 CONFIG_FHANDLE (libudev, mount and bind mount handling) 62 63 Kernel crypto/hash API: 64 CONFIG_CRYPTO_USER_API_HASH 65 CONFIG_CRYPTO_HMAC 66 CONFIG_CRYPTO_SHA256 67 68 udev will fail to work with the legacy sysfs layout: 69 CONFIG_SYSFS_DEPRECATED=n 70 71 Legacy hotplug slows down the system and confuses udev: 72 CONFIG_UEVENT_HELPER_PATH="" 73 74 Userspace firmware loading is not supported and should be disabled in 75 the kernel: 76 CONFIG_FW_LOADER_USER_HELPER=n 77 78 Some udev rules and virtualization detection relies on it: 79 CONFIG_DMIID 80 81 Support for some SCSI devices serial number retrieval, to create 82 additional symlinks in /dev/disk/ and /dev/tape: 83 CONFIG_BLK_DEV_BSG 84 85 Required for PrivateNetwork= in service units: 86 CONFIG_NET_NS 87 Note that systemd-localed.service and other systemd units use 88 PrivateNetwork so this is effectively required. 89 90 Required for PrivateUsers= in service units: 91 CONFIG_USER_NS 92 93 Optional but strongly recommended: 94 CONFIG_IPV6 95 CONFIG_AUTOFS_FS 96 CONFIG_TMPFS_XATTR 97 CONFIG_{TMPFS,EXT4_FS,XFS,BTRFS_FS,...}_POSIX_ACL 98 CONFIG_SECCOMP 99 CONFIG_SECCOMP_FILTER (required for seccomp support) 100 CONFIG_KCMP (for the kcmp() syscall, used to be under 101 CONFIG_CHECKPOINT_RESTORE before ~5.12) 102 103 Required for CPUShares= in resource control unit settings: 104 CONFIG_CGROUP_SCHED 105 CONFIG_FAIR_GROUP_SCHED 106 107 Required for CPUQuota= in resource control unit settings: 108 CONFIG_CFS_BANDWIDTH 109 110 Required for IPAddressDeny=, IPAddressAllow=, IPIngressFilterPath=, 111 IPEgressFilterPath= in resource control unit settings unit settings: 112 CONFIG_BPF 113 CONFIG_BPF_SYSCALL 114 CONFIG_BPF_JIT 115 CONFIG_HAVE_EBPF_JIT 116 CONFIG_CGROUP_BPF 117 118 Required for SocketBind{Allow|Deny}=, RestrictNetworkInterfaces= in 119 resource control unit settings: 120 CONFIG_BPF 121 CONFIG_BPF_SYSCALL 122 CONFIG_BPF_JIT 123 CONFIG_HAVE_EBPF_JIT 124 CONFIG_CGROUP_BPF 125 126 For UEFI systems: 127 CONFIG_EFIVAR_FS 128 CONFIG_EFI_PARTITION 129 130 Required for signed Verity images support: 131 CONFIG_DM_VERITY_VERIFY_ROOTHASH_SIG 132 133 Required for RestrictFileSystems= in service units: 134 CONFIG_BPF 135 CONFIG_BPF_SYSCALL 136 CONFIG_BPF_LSM 137 CONFIG_DEBUG_INFO_BTF 138 CONFIG_LSM="...,bpf" or kernel booted with lsm="...,bpf". 139 140 We recommend to turn off Real-Time group scheduling in the kernel when 141 using systemd. RT group scheduling effectively makes RT scheduling 142 unavailable for most userspace, since it requires explicit assignment of 143 RT budgets to each unit whose processes making use of RT. As there's no 144 sensible way to assign these budgets automatically this cannot really be 145 fixed, and it's best to disable group scheduling hence: 146 CONFIG_RT_GROUP_SCHED=n 147 148 It's a good idea to disable the implicit creation of networking bonding 149 devices by the kernel networking bonding module, so that the 150 automatically created "bond0" interface doesn't conflict with any such 151 device created by systemd-networkd (or other tools). Ideally there would 152 be a kernel compile-time option for this, but there currently isn't. The 153 next best thing is to make this change through a modprobe.d drop-in. 154 This is shipped by default, see modprobe.d/systemd.conf. 155 156 Required for systemd-nspawn: 157 CONFIG_DEVPTS_MULTIPLE_INSTANCES or Linux kernel >= 4.7 158 159 Required for systemd-oomd: 160 CONFIG_PSI 161 162 Note that kernel auditing is broken when used with systemd's container 163 code. When using systemd in conjunction with containers, please make 164 sure to either turn off auditing at runtime using the kernel command 165 line option "audit=0", or turn it off at kernel compile time using: 166 CONFIG_AUDIT=n 167 If systemd is compiled with libseccomp support on architectures which do 168 not use socketcall() and where seccomp is supported (this effectively 169 means x86-64 and ARM, but excludes 32-bit x86!), then nspawn will now 170 install a work-around seccomp filter that makes containers boot even 171 with audit being enabled. This works correctly only on kernels 3.14 and 172 newer though. TL;DR: turn audit off, still. 173 174 glibc >= 2.16 175 libcap 176 libmount >= 2.30 (from util-linux) 177 (util-linux *must* be built without --enable-libmount-support-mtab) 178 libseccomp >= 2.3.1 (optional) 179 libblkid >= 2.24 (from util-linux) (optional) 180 libkmod >= 15 (optional) 181 PAM >= 1.1.2 (optional) 182 libcryptsetup (optional), >= 2.3.0 required for signed Verity images support 183 libaudit (optional) 184 libacl (optional) 185 libbpf >= 0.2.0 (optional) 186 libfdisk >= 2.32 (from util-linux) (optional) 187 libselinux (optional) 188 liblzma (optional) 189 liblz4 >= 1.3.0 / 130 (optional) 190 libzstd >= 1.4.0 (optional) 191 libgcrypt (optional) 192 libqrencode (optional) 193 libmicrohttpd (optional) 194 libpython (optional) 195 libidn2 or libidn (optional) 196 gnutls >= 3.1.4 (optional, >= 3.6.0 is required to support DNS-over-TLS with gnutls) 197 openssl >= 1.1.0 (optional, required to support DNS-over-TLS with openssl) 198 elfutils >= 158 (optional) 199 polkit (optional) 200 tzdata >= 2014f (optional) 201 pkg-config 202 gperf 203 docbook-xsl (optional, required for documentation) 204 xsltproc (optional, required for documentation) 205 python-jinja2 206 python-lxml (optional, required to build the indices) 207 python >= 3.5 208 meson >= 0.53.2 209 ninja 210 gcc, awk, sed, grep, and similar tools 211 clang >= 10.0, llvm >= 10.0 (optional, required to build BPF programs 212 from source code in C) 213 gnu-efi >= 3.0.5 (optional, required for systemd-boot) 214 215 During runtime, you need the following additional 216 dependencies: 217 218 util-linux >= v2.27.1 required 219 dbus >= 1.4.0 (strictly speaking optional, but recommended) 220 NOTE: If using dbus < 1.9.18, you should override the default 221 policy directory (--with-dbuspolicydir=/etc/dbus-1/system.d). 222 dracut (optional) 223 polkit (optional) 224 225 To build in directory build/: 226 meson setup build/ && ninja -C build/ 227 228 Any configuration options can be specified as -Darg=value... arguments 229 to meson. After the build directory is initially configured, meson will 230 refuse to run again, and options must be changed with: 231 meson configure -Darg=value build/ 232 meson configure without any arguments will print out available options and 233 their current values. 234 235 Useful commands: 236 ninja -C build -v some/target 237 meson test -C build/ 238 sudo meson install -C build/ --no-rebuild 239 DESTDIR=... meson install -C build/ 240 241 A tarball can be created with: 242 v=250 && git archive --prefix=systemd-$v/ v$v | zstd >systemd-$v.tar.zstd 243 244 When systemd-hostnamed is used, it is strongly recommended to install 245 nss-myhostname to ensure that, in a world of dynamically changing 246 hostnames, the hostname stays resolvable under all circumstances. In 247 fact, systemd-hostnamed will warn if nss-myhostname is not installed. 248 249 nss-systemd must be enabled on systemd systems, as that's required for 250 DynamicUser= to work. Note that we ship services out-of-the-box that 251 make use of DynamicUser= now, hence enabling nss-systemd is not 252 optional. 253 254 Note that the build prefix for systemd must be /usr. (Moreover, packages 255 systemd relies on — such as D-Bus — really should use the same prefix, 256 otherwise you are on your own.) -Dsplit-usr=false (which is the default 257 and does not need to be specified) is the recommended setting. 258 -Dsplit-usr=true can be used to give a semblance of support for systems 259 with programs installed split between / and /usr. Moving everything 260 under /usr is strongly encouraged. 261 262 Additional packages are necessary to run some tests: 263 - busybox (used by test/TEST-13-NSPAWN-SMOKE) 264 - nc (used by test/TEST-12-ISSUE-3171) 265 - python3-pyparsing 266 - python3-evdev (used by hwdb parsing tests) 267 - strace (used by test/test-functions) 268 - capsh (optional, used by test-execute) 269 270POLICY FOR SUPPORT OF DISTRIBUTIONS AND ARCHITECTURES: 271 272 systemd main branch and latest major or stable releases are generally 273 expected to compile on current versions of popular distributions (at 274 least all non-EOL versions of Fedora, Debian unstable/testing/stable, 275 latest Ubuntu LTS and non-LTS releases, openSUSE Tumbleweed/Leap, 276 CentOS Stream 8 and 9, up-to-date Arch, etc.) We will generally 277 attempt to support also other non-EOL versions of various distros. 278 Features which would break compilation on slightly-older distributions 279 will only be introduced if there are significant reasons for this 280 (i.e. supporting them interferes with development or requires too many 281 resources to support). In some cases backports of specific libraries or 282 tools might be required. 283 284 The policy is similar wrt. architecture support. systemd is regularly 285 tested on popular architectures (currently amd64, i386, arm64, ppc64el, 286 and s390x), but should compile and work also on other architectures, for 287 which support has been added. systemd will emit warnings when 288 architecture-specific constants are not defined. 289 290USERS AND GROUPS: 291 Default udev rules use the following standard system group names, which 292 need to be resolvable by getgrnam() at any time, even in the very early 293 boot stages, where no other databases and network are available: 294 295 audio, cdrom, dialout, disk, input, kmem, kvm, lp, render, tape, tty, video 296 297 During runtime, the journal daemon requires the "systemd-journal" system 298 group to exist. New journal files will be readable by this group (but 299 not writable), which may be used to grant specific users read access. In 300 addition, system groups "wheel" and "adm" will be given read-only access 301 to journal files using systemd-tmpfiles.service. 302 303 The journal remote daemon requires the "systemd-journal-remote" system 304 user and group to exist. During execution this network facing service 305 will drop privileges and assume this uid/gid for security reasons. 306 307 Similarly, the network management daemon requires the "systemd-network" 308 system user and group to exist. 309 310 Similarly, the name resolution daemon requires the "systemd-resolve" 311 system user and group to exist. 312 313 Similarly, the coredump support requires the "systemd-coredump" system 314 user and group to exist. 315 316NSS: 317 systemd ships with four glibc NSS modules: 318 319 nss-myhostname resolves the local hostname to locally configured IP 320 addresses, as well as "localhost" to 127.0.0.1/::1. 321 322 nss-resolve enables DNS resolution via the systemd-resolved DNS/LLMNR 323 caching stub resolver "systemd-resolved". 324 325 nss-mymachines enables resolution of all local containers registered 326 with machined to their respective IP addresses. 327 328 nss-systemd enables resolution of users/group registered via the 329 User/Group Record Lookup API (https://systemd.io/USER_GROUP_API), 330 including all dynamically allocated service users. (See the 331 DynamicUser= setting in unit files.) 332 333 To make use of these NSS modules, please add them to the "hosts:", 334 "passwd:" and "group:" lines in /etc/nsswitch.conf. The "resolve" module 335 should replace the glibc "dns" module in this file (and don't worry, it 336 chain-loads the "dns" module if it can't talk to resolved). 337 338 The four modules should be used in the following order: 339 340 passwd: compat systemd 341 group: compat systemd 342 hosts: files mymachines resolve [!UNAVAIL=return] dns myhostname 343 344SYSV INIT.D SCRIPTS: 345 When calling "systemctl enable/disable/is-enabled" on a unit which is a 346 SysV init.d script, it calls /usr/lib/systemd/systemd-sysv-install; 347 this needs to translate the action into the distribution specific 348 mechanism such as chkconfig or update-rc.d. Packagers need to provide 349 this script if you need this functionality (you don't if you disabled 350 SysV init support). 351 352 Please see src/systemctl/systemd-sysv-install.SKELETON for how this 353 needs to look like, and provide an implementation at the marked places. 354 355WARNINGS and TAINT FLAGS: 356 systemd will warn during early boot if /usr is not already mounted at 357 this point (that means: either located on the same file system as / or 358 already mounted in the initrd). While in systemd itself very little 359 will break if /usr is on a separate late-mounted partition, many of its 360 dependencies very likely will break sooner or later in one form or 361 another. For example, udev rules tend to refer to binaries in /usr, 362 binaries that link to libraries in /usr or binaries that refer to data 363 files in /usr. Since these breakages are not always directly visible, 364 systemd will warn about this. Such setups are not really supported by 365 the basic set of Linux OS components. Taint flag 'split-usr' will be 366 set when this condition is detected. 367 368 For more information on this issue consult 369 https://www.freedesktop.org/wiki/Software/systemd/separate-usr-is-broken 370 371 systemd will warn if the filesystem is not usr-merged (i.e.: /bin, /sbin 372 and /lib* are not symlinks to their counterparts under /usr). Taint flag 373 'unmerged-usr' will be set when this condition is detected. 374 375 For more information on this issue consult 376 https://www.freedesktop.org/wiki/Software/systemd/TheCaseForTheUsrMerge 377 378 systemd requires that the /run mount point exists. systemd also 379 requires that /var/run is a symlink to /run. Taint flag 'var-run-bad' 380 will be set when this condition is detected. 381 382 Systemd will also warn when the cgroup support is unavailable in the 383 kernel (taint flag 'cgroups-missing'), the system is using the old 384 cgroup hierarchy (taint flag 'cgroupsv1'), the hardware clock is 385 running in non-UTC mode (taint flag 'local-hwclock'), the kernel 386 overflow UID or GID are not 65534 (taint flags 'overflowuid-not-65534' 387 and 'overflowgid-not-65534'), the UID or GID range assigned to the 388 running systemd instance covers less than 0…65534 (taint flags 389 'short-uid-range' and 'short-gid-range'). 390 391 Taint conditions are logged during boot, but may also be checked at any 392 time with: 393 394 busctl get-property org.freedesktop.systemd1 /org/freedesktop/systemd1 org.freedesktop.systemd1.Manager Tainted 395 396VALGRIND: 397 To run systemd under valgrind, compile with meson option 398 -Dvalgrind=true and have valgrind development headers installed 399 (i.e. valgrind-devel or equivalent). Otherwise, false positives will be 400 triggered by code which violates some rules but is actually safe. Note 401 that valgrind generates nice output only on exit(), hence on shutdown 402 we don't execve() systemd-shutdown. 403 404STABLE BRANCHES AND BACKPORTS: 405 Stable branches with backported patches are available in the 406 systemd-stable repo at https://github.com/systemd/systemd-stable. 407 408 Stable branches are started for certain releases of systemd and named 409 after them, e.g. v238-stable. Stable branches are managed by 410 distribution maintainers on an as needed basis. See 411 https://www.freedesktop.org/wiki/Software/systemd/Backports/ for some 412 more information and examples. 413