Currently with FileDescriptorStorePreserve=yes the FD store is kept
around
regardless of what happens to a unit, which is useful in many cases. But
in
some cases, for example when complex services crash horribly, it's hard
to
reason about what was in the intermediate state, and it's better to
start
fresh.
Add a new 'on-success' option for the FileDescriptorStorePreserve=
setting
that keeps it around only for as long as the unit doesn't go to a
persistently
failed state.
This is especially useful in combination with LUO, where we don't want
to
keep around LUO sessions created by units that then proceeded to crash
and
burn, and might be in a bad state afterwards.
core: add User,Group,SupplementaryGroups,Nice to varlink
Unit.StartTransient
This commit adds more writable fields to the
io.systemd.Unit.StartTransient
varlink method. With this its possible to set:
User,Group,SupplementaryGroups,Nice values.
Plus tests for them.
---
core: report unsupported service fields in varlink calls
Just like for the unsupported/bad exec_fields we should show
a message about what field is bad for service parameters. This
commit adds it using the same pattern. The JSON parser works in
fail-fast mode so we only display the first bad field (and
it depends on the parser what it finds first).
Currently with FileDescriptorStorePreserve=yes the FD store is kept around
regardless of what happens to a unit, which is useful in many cases. But in
some cases, for example when complex services crash horribly, it's hard to
reason about what was in the intermediate state, and it's better to start
fresh.
Add a new 'on-success' option for the FileDescriptorStorePreserve= setting
that keeps it around only for as long as the unit doesn't go to a persistently
failed state.
This is especially useful in combination with LUO, where we don't want to
keep around LUO sessions created by units that then proceeded to crash and
burn, and might be in a bad state afterwards.
This commit moves the io.systemd.Unit.StartTransient tests into
the right place in TEST-74-AUX-UTILS.varlinkctl-unit.sh.
Thanks to Ivan Kruglov for suggesting this.
This commit move the erros in StartTransient from
VARLINK_ERROR_UNIT_BAD_SETTING to SD_VARLINK_ERROR_INVALID_PARAMETER
and it also ensures we have the bad field in the error.
Thanks to Ivan Kruglov for suggesting this.
This commit adds more writable fields to the io.systemd.Unit.StartTransient
varlink method. With this its possible to set:
User,Group,SupplementaryGroups,Nice values.
Plus tests for them.
Add support for filtering IRQ affinity to CPUs on a specific NUMA node
via the new IRQAffinityNUMA= option in .link files. The option accepts:
- "local": use the NUMA node local to the NIC's PCIe slot
- Explicit node number (0, 1, 2, ...): use CPUs on the specified node
When both IRQAffinity= and IRQAffinityNUMA= are specified, their
intersection is used. If the intersection is empty, an error is logged
and IRQ affinity configuration is skipped.
When "local" is specified but the device's NUMA node cannot be
determined (numa_node shows -1), a warning is logged and IRQ affinity
configuration is skipped.
Add IRQAffinity= option to .link files that filters the set of CPUs
eligible for IRQ placement. This works in conjunction with
IRQAffinityPolicy= to constrain which CPUs receive network IRQs.
When specified with spread policy, only the listed CPUs are considered
for IRQ distribution. When specified with single policy, IRQs are
pinned to the first CPU in the allowed set instead of CPU 0.
Implement the spread policy for IRQ affinity distribution using a
topology-aware algorithm. The algorithm:
1. Discovers CPU topology from sysfs (NUMA node, package, die/L3, core)
2. Groups CPUs by L3 cache domain (die) with equidistant ordering
3. Round-robins across dies, spreading IRQs across the system
4. Uses first hyperthread of each core before second hyperthreads
5. Applies IRQ affinity via /proc/irq/<n>/smp_affinity
When there are more IRQs than CPUs, queues wrap around using round-robin.
Add support for configuring IRQ affinity for network interfaces via
systemd .link files. For now, the new IRQAffinityPolicy= option in the [Link]
section only accepts "single", which pins all MSI IRQs to CPU 0.
This allows declarative IRQ affinity configuration for network devices
during udev processing, which is useful for optimizing network
performance on multi-core systems.
Further commits will expand the options supported by IRQAffinityPolicy=.
Just like for the unsupported/bad exec_fields we should show
a message about what field is bad for service parameters. This
commit adds it using the same pattern. The JSON parser works in
fail-fast mode so we only display the first bad field (and
it depends on the parser what it finds first).
Methods:
- `io.systemd.Job.List` — list all queued jobs or look up by `id`/`unit`
name, with streaming support. Uses context/runtime split: `JobContext`
(Unit, JobType) and `JobRuntime` (Id, State, Result,
ActivationDetails). Follows the same SELinux and parameter-conflict
patterns as `io.systemd.Unit.List`.
- `io.systemd.Job.Cancel` — cancel a specific job by ID, with SELinux
and polkit authorization.
- `io.systemd.Job.ClearAll` — cancel all pending jobs, with SELinux and
polkit authorization.
Commit 881e4717c7 ("test: pin stress-ng --vm-method to a portable
scalar method in TEST-55-OOMD") pinned --vm-method=zero-one with the
rationale that it is "a long-standing scalar method". That rationale is
wrong: stress_vm_zero_one() in stress-ng's stress-vm.c is declared
static size_t TARGET_CLONES stress_vm_zero_one(...)
i.e. it carries the exact same TARGET_CLONES attribute as 33 of the 35
other vm methods. On x86_64 with GCC >=5, TARGET_CLONES expands (see
core-target-clones.h in stress-ng) to a target_clones attribute
including "arch=skylake-avx512", "arch=cooperlake", "arch=tigerlake",
"arch=sapphirerapids", and several other AVX-512-bearing arch variants,
plus "default". GCC generates AVX-512 clones of stress_vm_zero_one() and
the IFUNC resolver picks them on any CPU that advertises AVX-512.
The only vm methods in stress-ng's registry whose function definitions
omit TARGET_CLONES entirely (and are therefore guaranteed not to
dispatch to an AVX-512 clone) are lfsr32 (portable, always registered)
and write64ds (x86_64-only, gated on HAVE_ASM_X86_MOVDIRI, i.e. Intel
Tremont / Tiger Lake+ MOVDIRI instruction).
Switch the four stress-ng --vm invocations in TEST-55-OOMD to
--vm-method=lfsr32 so the AVX-512 SIGILL on CPUs without AVX-512 (e.g.
AMD Zen 1-3) can no longer occur regardless of compiler version,
optimization level, or stress-ng package build.
Follow-up for 881e4717c7
Co-developed-by: Claude Opus 4.7 <noreply@anthropic.com>
The test waited for the OnFailure= service's filesystem side effect
(`rmdir` of the directory) and then immediately invoked
`systemctl is-active`. Between `rmdir(2)` returning (which causes the
shell loop to exit) and PID1 reaping the child and transitioning the
oneshot service from `activating` to `active`, there is a small window
where `is-active` can observe `activating` and fail the test.
Wait directly on the unit state instead, matching the pattern used a
few lines above for the `is-failed` case.
[ 1880.326704] TEST-07-PID1.sh[21489]: + timeout --foreground 60 bash -c 'while [[ -d '\''/tmp/TEST-07-PID1-socket-8467/test'\'' ]]; do sleep .5; done'
[ 1880.330482] TEST-07-PID1.sh[21489]: + [[ ! -e /tmp/TEST-07-PID1-socket-8467/test ]]
[ 1880.330482] TEST-07-PID1.sh[21489]: + systemctl is-active TEST-07-PID1-socket-OnFailure.service
[ 1880.347470] TEST-07-PID1.sh[21520]: activating
[ 1880.349508] TEST-07-PID1.sh[21489]: + at_exit
[ 1880.349508] TEST-07-PID1.sh[21489]: + systemctl stop TEST-07-PID1-socket-8467.socket
[ 1880.367331] TEST-07-PID1.sh[107]: Subtest /usr/lib/systemd/tests/testdata/units/TEST-07-PID1.socket-on-failure.sh failed
[ 1880.367331] TEST-07-PID1.sh[107]: + return 1
Co-developed-by: Claude Opus 4.7 <noreply@anthropic.com>
Add COREDUMP_CODE to the fields captured by systemd-coredump. This makes
it possible for system administrators to filter coredumps based on si_code,
which describes the reason why a given signal was sent.
For example, to find processes killed due to invalid permissions
(SEGV_ACCERR):
$ journalctl COREDUMP_SIGNAL=11 COREDUMP_CODE=2
I've decided to add the value of si_code to the 'Signal: ' line of
coredumpctl info:
Signal: 11 (SEGV) si_code: SEGV_ACCERR
The stress-ng "vm" stressor's default --vm-method=all cycles through every
VM stress method, including newer ones that use AVX-512 instructions. On
CPUs without AVX-512 support (e.g. AMD Zen 1 to 3) those methods crash with
SIGILL. In testcase_oom_rulesets_lasting_sec all 10 stress-ng workers die
within ~2.34 seconds, so by the time the 6 second sleep elapses the unit
is already in failed/exit-code state and the assert_eq for
ActiveState=active trips.
Pin --vm-method=zero-one, a long-standing scalar method, on all four
stress-ng --vm invocations in this test (the two transient services in
testcase_oom_rulesets and testcase_oom_rulesets_lasting_sec, plus
TEST-55-OOMD-testbloat.service and TEST-55-OOMD-testmunch.service) so the
workers do not crash on AVX-512-less CPUs. testbloat, testmunch and
testcase_oom_rulesets have not been observed failing because they get
OOM-killed by systemd-oomd within ~1 to 2 seconds, before stress-ng cycles
into an AVX-512 method, but they share the same latent flake.
Journal excerpts from the failing run, TEST-55-OOMD-slowrule.service in
testcase_oom_rulesets_lasting_sec (journalctl -o short-monotonic):
[ 58.018676] stress-ng[1015]: invoked with '/usr/bin/stress-ng --timeout 15s --vm 10 --vm-bytes 50M --vm-keep' by user 0 'root'
[ 59.866072] stress-ng[1030]: stress-ng: debug: [1030] caught SIGILL, address 0x000055bd8d609140 (ILL_ILLOPN)
[ 59.921050] stress-ng[1030]: stress-ng: debug: [1030] stress-ng: info: 0x000055bd8d609140:<62>71 fd 48 6f 2d 36 14 1c 00 c5 d1 ef ed 49 29
[ 59.929310] stress-ng[1015]: stress-ng: error: [1015] vm: [1021] terminated with an error, exit status=2 (stressor failed)
[ 60.364111] stress-ng[1015]: stress-ng: info: [1015] failed: 10: vm (10)
[ 60.364493] stress-ng[1015]: stress-ng: info: [1015] unsuccessful run completed in 2.34 secs
[ 60.371290] systemd[1]: TEST-55-OOMD-slowrule.service: Main process exited, code=exited, status=2/INVALIDARGUMENT
[ 60.371396] systemd[1]: TEST-55-OOMD-slowrule.service: Failed with result 'exit-code'.
[ 64.017061] TEST-55-OOMD.sh[1010]: + assert_eq failed active
[ 64.018167] TEST-55-OOMD.sh[1039]: FAIL: expected: 'active' actual: 'failed'
The faulting bytes marked by stress-ng with <62> (the byte at the
instruction pointer) decode unambiguously to an AVX-512 VMOVDQA64 using
the 512-bit zmm13 register, confirmed independently by two disassemblers:
$ printf '\x62\x71\xfd\x48\x6f\x2d\x36\x14\x1c\x00' | ndisasm -b 64 -
00000000 6271FD486F2D3614 vmovdqa64 zmm13,zword [rel 0x1c1440]
-1C00
$ echo '0x62, 0x71, 0xfd, 0x48, 0x6f, 0x2d, 0x36, 0x14, 0x1c, 0x00' | \
llvm-mc -disassemble -triple=x86_64 -mattr=+avx512f
.text
vmovdqa64 1840182(%rip), %zmm13
The leading 0x62 is the EVEX prefix (exclusive to AVX-512 on this target),
zmm13 is a 512-bit register that only exists when AVX-512 is implemented,
and VMOVDQA64 requires the AVX512F (Foundation) CPUID feature (Intel SDM
Vol 2C). Executing this on a CPU without AVX-512 raises #UD, delivered by
the kernel as SIGILL/ILL_ILLOPN, matching the journal entry above. The
same journal shows the kernel reporting "kvm_amd: TSC scaling supported",
i.e. the guest is on AMD KVM, and AMD did not ship AVX-512 before Zen 4.
Co-developed-by: Claude Opus 4.7 <noreply@anthropic.com>
The current key sealing format has some less-than-ideal weaknesses:
- PBKDF2 with only 65k iterations, where recommendations are ~200k
- AES with null IV, relying on salt for uniqueness
- lack of AES MAC/AEAD
However improbable, it is at least theorically possible that with
a lot of resources an offline bruteforce could be attempted.
Add a v2 sealing format, keeping unsealing compatibility with
the current format:
v2:<iterations>:<salt>:<IV>:<ciphertext>:<aes tag>
and use 600k iterations for the PBKDF2 sha512
Split the monolithic varlinkctl test script into separate files per varlink interface for better organization and easier maintenance:
- varlinkctl.sh: core varlinkctl tool tests (CLI, transports, socket discovery, upgrade/serve) and io.systemd.Manager
- varlinkctl-network.sh: io.systemd.Network
- varlinkctl-unit.sh: io.systemd.Unit (system + user manager)
- varlinkctl-metrics.sh: io.systemd.Metrics
No functional changes — the test content is moved as-is.
Co-developed-by: Claude Opus 4.6 <noreply@anthropic.com>
Introduce COREDUMP_CODE as a new captured field alongside the existing
COREDUMP_SIGNAL. While COREDUMP_SIGNAL identifies the signal number that
terminated the process, COREDUMP_CODE provides the reason the signal was sent.
For example, a process terminated by SIGSEGV due to invalid permissions would
produce COREDUMP_SIGNAL=11 and COREDUMP_CODE=2 (SEGV_ACCERR).
The kernel exposes coredump_code via pidfd starting with v7.1:
https://git.kernel.org/torvalds/c/701f7f4fbabbf4989ba6fbf033b160dd943221d5
System administrators can find both the signal and code in coredumpctl info:
$ coredumpctl info | grep Signal:
Signal: 11 (SEGV) si_code: SEGV_MAPERR
Signed-off-by: Emanuele Rocca <emanuele.rocca@arm.com>
When systemd-pull encountered a file shorter than the compression magic
headers it looks for, then it would complete the download in the
analysis state and fail.
When we are still in the analysis state and the download is done, we
know there is no compression and we should leave the analysis state and
continue writing out to disk as usual.
When systemd-pull encountered a file shorter than the compression magic
headers it looks for, then it would complete the download in the
analysis state and fail.
When we are still in the analysis state and the download is done, we
know there is no compression and we should leave the analysis state and
continue writing out to disk as usual.
For the shell verb we want switches specified after the program name to
be passed to the program to execute, not processed by us. Mirror the
approach in 'userdbctl ssh-authorized-keys': start with
OPTION_PARSER_RETURN_POSITIONAL_ARGS, then lates switch to
STOP_AT_FIRST_NONOPTION for "shell" or NORMAL otherwise.
VERB declarations are placed directly above each function; functions
that dispatch multiple verb names get stacked VERB() declarations.
chainload_importctl() now takes the args strv instead of relying on the
global optind.
--help output is mostly the same.
--no-pager/--no-legend/--no-ask-password/-q/--quiet are now at the end.
bind-volume/unbind-volume are documented.
Fixes c9f461a806.
Co-developed-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
LUO sessions cannot be nested under other sessions. This means we need
to handle them explicitly, and held them open in the shutdown binary
like we do with our own internal session, to allow services to create
their own.
The requirement to support third party sessions comes from VMMs that
wish to preserve VM(s) state(s) across kexec, as some file descriptors
(KVM's vmfd from the KVM_CREATE_VM ioctl) cannot be transfered between
processes via SCM_RIGHTS, so they cannot be stashed in the FD Store
directly. Also some file descriptors have to be handled all together or
not at all, again to do with KVM and devices that are all part of the
same vm.
When there is a NOTIFY_SOCKET, and FDs are received from the
payload following the FD Store protocol, forward them up the
chain to the service manager that is managing nspawn.
This allows FD Store persistence across container restarts,
and can chain up for user managers as well to survive restarting
those, or reexecs, and in the future reboots too via LUO.
Add a new test case to exercise the PID1 -> user session -> nspawn -> payload
chain.
This PR adds some more properties to the io.systemd.Unit.StartTransient
varlink interface: WorkingDirectory, Environment and
SetCredential{,Encrypted}. Its also hopefully a useful starting point to
establish a pattern to add even more.
Implement support for the --json= flag in the info subcommand
(issue #38844). Previously, coredumpctl info always produced
human-readable text output regardless of --json=.
Add a CoredumpFields struct that holds all journal fields extracted
for a coredump entry, along with coredump_fields_done() to release
member resources and coredump_fields_load() to populate the struct
from a journal entry. Both print_info() and the new print_info_json()
use this shared loader, eliminating the duplicate RETRIEVE loop.
print_info_json() builds a JSON object with the same fields shown by
print_info(). Missing fields are omitted via SD_JSON_BUILD_PAIR_CONDITION,
matching the tolerant behavior of print_info() rather than skipping the
entry entirely. Signal/Reason handling mirrors print_info(): normal
coredumps (MESSAGE_ID == SD_MESSAGE_COREDUMP_STR) emit a numeric Signal
field; non-normal entries (kernel oops, etc.) emit a Reason field with
the raw text from COREDUMP_SIGNAL.
Co-developed-by: Claude Opus 4.6 <noreply@anthropic.com>
When the kernel hands the command line to PID 1, any KEY=VALUE assignment
whose KEY does not contain a '.' is exported as an environment variable
(with '-' replaced by '_') rather than passed as an argument. Mimic the
same split in --boot mode so kernel-cmdline-style arguments passed after
the container path behave as they would on a real boot.
This commit adds support to set `Environment` in the
`io.systemd.Unit.StartTransient` varlink call.
The behavior is similar to D-Bus, i.e. a `null` or `[]` clears the
environment. This is not needed for StartTransient() as there the env
always starts empty but it seems a good property to have if this is
reused.
This commit adds setting the WorkingDirectory to the
`io.systemd.Unit.StartTransient` varlink call. This is a
first step towards more complete StartTransient in varlink.
The goal is to be as close as possible to the D-Bus parameters.
The exception is WorkingDirectory which is an object here so that
we avoid the `-` prefixes and use a more type-safe approach by
making it an explicit `missingOK` parameter.
The key names stay the same as the D-Bus properties (PascalCase).
If there are no equivalent D-Bus properties the native varlink
convention of camelCase is used.
qemu by default only has 30 PCI slots and with vmspawn now reserving
some of those for its hotplug features, we go over the limit for the
nvme test.
Let's drop the number of nvme devices to 12 to fix the conflict.
The fix for the corrupted state when units become aliased on reload
leaks the now-aliased unit's resources, which become untracked and
essentially lost.
While fixing the state corruption is of course necessary, leaking
processes/etc. is not ideal for a system and service manager, so
instead attempt to keep track of them by creating stub units
on-the-fly.
This way resources are not leaked, there are clear indications of
where they moved, and all state can be tracked as expected.
This series adds a new `RestrictFileSystemAccess=` setting in the
`[Manager]` section of `system.conf` that enforces a deny-default
execution policy: only binaries residing on signed dm-verity block
devices (and the initramfs during early boot) are permitted to execute.
Everything else — tmpfs, procfs, sysfs, anonymous executable mappings,
unsigned dm-verity devices — is denied.
The directive takes the values `no` (default), `exec` (lock down
execution), and accepts `yes` as an alias for `exec`. The name is
deliberately broader than what the initial values cover so the same
setting can grow to restrict other filesystem access categories in the
future (e.g. `any` to deny all access from untrusted filesystems, not
just execution).
### How it works
The BPF program is entirely self-contained; PID1 loads it and the kernel
does the rest. When dm-verity brings up a device, the kernel calls
`security_bdev_setintegrity()` twice during `verity_preresume()`: once
with the root hash and once with the signature validity status. Our
`lsm/bdev_setintegrity` hook captures the second call and records the
device number in a BPF hash map if the signature is valid. When a device
is torn down, `lsm/bdev_free_security` cleans up the map entry. No
userspace map population is needed at any point.
The enforcement side hooks `bprm_check_security` (execve), `mmap_file`
(PROT_EXEC mappings including shared libraries), and `file_mprotect`
(W→X transitions like JIT and libffi). Each hook resolves the file's
backing device via `file->f_inode->i_sb->s_dev` and looks it up in the
verity device map. For block-backed filesystems, `s_dev` equals
`s_bdev->bd_dev`, which avoids an extra pointer chase and NULL check on
`s_bdev` — non-block filesystems simply miss in the map and get denied
by the default policy.
During early boot the initramfs needs to be trusted as well, since it
runs before any dm-verity volume is mounted. PID1 writes the initramfs
superblock's device number into a BPF global before attaching the
programs, and clears it after `switch_root` to close the trust window.
As a prerequisite, PID1 also verifies that
`dm_verity.require_signatures=1` is active — without it, unsigned
dm-verity devices could be created, which would weaken the security
model even though the BPF program would correctly deny execution from
them.
### Surviving daemon-reexec
The BPF programs and their verity device map must survive PID1
re-execution (daemon-reexec, switch_root, soft-reboot). Without
preservation, `manager_free()` would destroy the skeleton, the link FDs
would close, programs would detach, and the map would be freed. After
exec, a fresh skeleton would have an empty map — but existing dm-verity
devices have already signaled their integrity and won't do so again. A
deny-default policy plus an empty map means all execution denied and the
system is bricked.
We solve this by serializing the raw BPF link FDs and the `.bss` map FD
across exec using systemd's existing `serialize_fd` / `fdset_cloexec` /
`deserialize_fd` infrastructure. The kernel reference chain (link FD →
`struct bpf_link` → `struct bpf_prog` → `struct bpf_map`) keeps programs
attached and map data intact as long as the dup'd FDs survive. After
exec, PID1 detects the deserialized FDs and skips skeleton re-creation
entirely. If switching root, it uses the deserialized `.bss` map FD to
clear `initramfs_s_dev` via a targeted `mmap()` write, preserving the
other guard globals in `.bss`.
We intentionally avoid bpffs pinning. Pinned objects are discoverable
and manipulable by any process with sufficient privileges
(`BPF_OBJ_GET`, unlink). FD serialization keeps everything private to
PID1 with no external attack surface.
### Self-protection
BPF LSM programs attached via the tracing trampoline (`BPF_LSM_MAC`) are
inherently tamper-resistant — `bpf_tracing_link_lops` has no
`.update_prog` and no `.detach` callbacks, so the kernel rejects
`BPF_LINK_UPDATE` with `-EINVAL` and `BPF_LINK_DETACH` with
`-EOPNOTSUPP`. Once attached, our programs cannot be modified or
detached through the `bpf()` syscall.
The remaining attack vector is map injection: `BPF_MAP_GET_FD_BY_ID` to
obtain an FD to `verity_devices`, then `BPF_MAP_UPDATE_ELEM` to insert a
fake trusted device. The self-protection guard blocks this with three
hooks. `lsm/bpf_map` fires inside `bpf_map_new_fd()`, the chokepoint for
all code paths that produce a map FD, and denies access to our map IDs
from any process other than PID1 (identified via `tgid == 1`, which is
unspoofable — `bpf_get_current_pid_tgid()` reads `current->tgid` from
`pid->numbers[0].nr`, the init-namespace PID). `lsm/bpf_prog` provides
analogous protection for program FDs as defense-in-depth. `lsm/bpf`
handles `BPF_LINK_GET_FD_BY_ID` at the command level since there is no
`security_bpf_link()` hook in the kernel.
The guard starts inactive — all protected IDs default to 0 in `.bss`,
and no real BPF object has ID 0 — so there is no window where it
interferes with PID1's own setup. After attaching all programs, PID1
queries the kernel-assigned IDs via `bpf_obj_get_info_by_fd()` and
writes them into the guard's globals. From that point on, the guard is
active. The guard has zero collateral damage: it only denies access to
our specific object IDs, leaving bpftrace, bpftool,
`RestrictFileSystems=`, and all other BPF usage completely unaffected.
Additionally, a ptrace guard (`lsm/ptrace_access_check`) blocks
`PTRACE_MODE_ATTACH` to PID1 from other processes, preventing extraction
of sensitive state from PID1's address space via ptrace, `/proc/1/mem`,
`process_vm_readv()`, or `pidfd_getfd()`. `PTRACE_MODE_READ` is allowed
so that monitoring tools and `systemctl` continue to work normally.
### Limitations
- The enforcement hooks resolve trust by looking at
`file->f_inode->i_sb->s_dev` — the device number of the superblock that
owns the file's inode. This works correctly for files directly on a
dm-verity block device, but it does not see through overlayfs. When a
file is accessed on an overlay mount, `f_inode` points to the overlay
inode, and `i_sb->s_dev` is the overlay superblock's anonymous device
number — not the underlying dm-verity device. The overlay superblock has
no backing block device, so the lookup misses in the verity map and
execution is denied by the default policy.
This means that overlayfs mounts whose lower layers are on
dm-verity-protected volumes will currently have execution blocked, even
though the actual data is integrity-protected. The correct fix requires
a kernel extension that allows the BPF program to call something like
`d_real_inode()` to resolve through the overlay to the real inode on the
underlying filesystem, and then check that inode's superblock device
number against the verity map. I plan to add a BPF kfunc exposing this
functionality in a follow-up kernel series.
- Multi-device filesystems such as btrfs use entirely synthetic device
numbers and there is no way to reach the actual device backing the inode
from the inode itself. So `RestrictFileSystemAccess=` only works
reliably with a subset of filesystems. In practice this isn't a problem
because the feature is tailored to erofs; using it on arbitrary
filesystems requires careful vetting of the actual filesystem behaviour.
- The initial implementation also blocks JIT-style execution that relies
on memory mapped executable. This is part of `exec` semantics today and
can be loosened later by introducing finer-grained values (a common
pattern in systemd — following the precedent of `ProtectSystem=`, which
started as a boolean and later grew `auto`/`yes`/`full`/`strict`
semantics).
- The configuration is a system-wide setting with no per-unit opt-out.
This is intentional for the initial implementation: a global invariant
is easier to reason about and harder to accidentally weaken. Per-unit
relaxation can be added later if a concrete need arises.
### Testing
The series includes unit tests and integration tests covering both the
core enforcement logic and the self-protection guard. The unit test
loads the skeleton, attaches programs, populates guard globals, and
verifies that protected IDs are set correctly. The integration tests
exercise the guard by attempting `BPF_MAP_GET_FD_BY_ID` and
`BPF_PROG_GET_FD_BY_ID` from a non-PID1 process and verifying that
access is denied.
What we cannot currently test end-to-end is actual execution enforcement
against a dm-verity-signed root filesystem. The systemd test suite does
not yet have infrastructure for booting a VM with a signed dm-verity
rootfs image — the existing mkosi-based test framework lacks the ability
to produce and boot such images. This will hopefully change soon when
Daan integrates barrage into the test suite.
Signed-off-by: Christian Brauner (Amutable) <brauner@kernel.org>
Convert test-btrfs to use the test framework and
assertions, merge the physical offset test into it
and beef it up to include what TEST-83-BTRFS does and
finally get rid of TEST-83-BTRFS as it is unneeded now.
Add TEST-90-RESTRICT-FSACCESS with two subtests:
config subtest — Tests PID1's RestrictFileSystemAccess= configuration parsing and
failure modes via system.conf drop-ins and daemon-reexec:
- Default RestrictFileSystemAccess=no produces no log messages
- RestrictFileSystemAccess=yes without BPF LSM logs appropriate warning
- RestrictFileSystemAccess=yes without require_signatures is correctly rejected
by the test helper binary's precondition check
enforce subtest — Tests actual BPF LSM enforcement using a test helper
binary (test-bpf-restrict-fsaccess) that loads the BPF skeleton with
initramfs_s_dev set to the rootfs s_dev, pins BPF links, and exits:
- Execution from rootfs continues to work (trusted via initramfs_s_dev)
- Execution from tmpfs is blocked with EPERM
- Execution from a signed dm-verity device is allowed, driven via
systemd-run -p RootImage= against the pre-built signed minimal_0
images that mkosi ships and signs at image build time (no on-the-fly
squashfs / verity hash tree / signature build required)
- After BPF detach, enforcement is lifted
All tests skip gracefully when prerequisites are not met (BPF LSM, BPF
framework, dm-verity tools, signing keys).
Signed-off-by: Christian Brauner <brauner@kernel.org>
A follow-up to the AddStorage / RemoveStorage series. ReplaceStorage
swaps the *backing file* of an already-attached storage device on a
running vmspawn-managed VM, leaving the guest-visible device frontend
(virtio-blk, virtio-scsi, nvme, scsi-cd) and every other property of
the device untouched. The intended use is to point an existing disk
at a new image without the guest seeing a hot-unplug/hot-plug cycle.
The signature mirrors AddStorage minus the 'config' field: the
device frontend doesn't change, only the backing behind it. Read-
only / read-write is derived from the new fd's O_ACCMODE; scsi-cd is
forced read-only to match the boot-time policy. S_ISBLK on the new
fd selects host_device vs file driver, matching AddStorage.
The QMP primitive is blockdev-reopen. It cannot change a file /
host_device node's 'filename' so we can't just point the existing
file node at a new fd, but it can swap a format node's 'file' child
to a different existing monitor-owned node by node-name reference
(case 3 in qemu/qapi/block-core.json:5034-5040). The chain is:
add-fd (host fd → new fdset)
blockdev-add (new file node, filename=/dev/fdset/N — fd-only)
remove-fd (release monitor's ref; new file holds the dup)
blockdev-reopen (format node, file = new file node-name)
blockdev-del (old file node; its dup release frees old fdset)
The reopen options must restate every option the original blockdev-
add emitted on the format node — blockdev-reopen resets any
unspecified option to its driver default. The 'file' field is a
node-name string reference, never a path.
No new errors and no new IDL types beyond the method itself;
everything is built on the existing NoSuchStorage / StorageImmutable
/ NotConnected / EBUSY vocabulary.
The series is:
vmspawn: split blockdev-add into separate file and format calls
Preparatory refactor. qemu/blockdev.c:3440 only marks the
top-level BDS returned by blockdev-add as monitor-owned;
inline children are NOT, so blockdev-del later rejects them
with "Node X is not owned by the monitor". Split into two
blockdev-add calls so the file node is independently
deletable. DriveInfo gains qmp_file_node_name and a
file_generation counter; the teardown helper deletes format
then file (file-first is rejected as "node used as 'file'
of Y"). The ephemeral path was already structured this way;
only the regular add path changes. Drops the now-unused
qmp_build_blockdev_add_inline().
shared/varlink-io.systemd.MachineInstance: add ReplaceStorage method
IDL only: ReplaceStorage(fileDescriptorIndex, name). No new
errors.
vmspawn: implement io.systemd.MachineInstance.ReplaceStorage
vmspawn_qmp_replace_block_device() entry point, ReplaceCtx
(refcounted, ReplaceCtxStateFlags for partial-state tracking)
and four async callbacks plus an idempotent replace_fail.
file_generation is bumped before issuing blockdev-add so
retries don't collide on node-name.
BLOCK_DEVICE_STATE_REPLACE_PENDING gates concurrent
Replace / Remove on the same drive. On reopen success the
trailing blockdev-del of the old file node fires from the
reopen callback; its failure logs a warning and still replies
success (the swap already committed; the orphan resolves at VM
exit). QMP disconnect mid-replace routes via
qmp_client_fail_pending → replace_fail → NotConnected.
test: integration test for io.systemd.MachineInstance.ReplaceStorage
TEST-87-AUX-UTILS-VM.replace-storage covers happy-path replace,
successive replaces (file_generation rotation), StorageImmutable
rejection on the boot-time drive, NoSuchStorage on unknown
names, InvalidParameter on malformed names, and clean
RemoveStorage after a replace (proves the new file node is
monitor-owned and the teardown order works). Backing files are
passed via 'varlinkctl --push-fd'; no machinectl front-end is
added in this round.
Signed-off-by: Christian Brauner (Amutable) <brauner@kernel.org>
JSON only supports string keys in objects, but D-Bus specification is a
bit more lenient and allows dict entries to have any basic type as key.
Let's stringify allowed non-string keys so that we can represent them as
JSON objects.
Relevant snippet from the D-Bus specification:
A DICT_ENTRY works exactly like a struct, but rather than parentheses
it uses curly braces, and it has more restrictions. The restrictions
are: it occurs only as an array element type; it has exactly two
single complete types inside the curly braces; the first single
complete type (the "key") must be a basic type rather than a container
type. Implementations must not accept dict entries outside of arrays,
must not accept dict entries with zero, one, or more than two fields,
and must not accept dict entries with non-basic-typed keys. A dict
entry is always a key-value pair.
Resolves: #32904
Luca Boccassi
Daan De Meyer
Michael Vogt
Quentin Deslandes
Yu Watanabe
Ivan Kruglov
Emanuele Rocca
Matteo Croce
Kai Lüke
Zbigniew Jędrzejewski-Szmek
Philip Withnall
noxiouz
Chris Down
Christian Brauner
Christian Brauner
Frantisek Sumsal