$ mkdir /var/tmp/files
$ touch /var/tmp/files/a
$ mkdir /var/tmp/conf
$ cat >>/var/tmp/conf/esp.conf
[Partition]
Type=esp
Format=vfat
CopyFiles=/var/tmp/files:/
$ truncate /var/tmp/disk -s 300M
$ sudo systemd-repart --dry-run=no --empty=require --definitions=/var/tmp/conf /var/tmp/disk
...
Populating vfat filesystem.
Failed to copy '...' to '/run/systemd/mount-root/': Operation not permitted
(sd-copy) failed with exit status 1.
The issue is that if there's a file owned by non-root and we try to copy
it into a newly-created DOS partition, fchown fails:
fchown(11</run/systemd/mount-root/...>, 1000, 1000) = -1 EPERM (Operation not permitted)
We want to ignore file ownership in such cases, so pass our own UID/GID
to copy_tree_at(), which turns the fchown into a noop and let's the
operation pass through.
Fixes#38863.
Currently almost all the dlopens happen in libbasic or libshared code,
so the ELF dlopen notes all end up in libsystemd-shared. Many
distributions split this library and various binaries in separate
packages, and the library ends up with soft-dependencies, even though
many binaries are either completely useless or do not work at all with
the dlopen dependency. This also makes it impossible to know which
executable uses which dlopen dependency without inspecting the source
code.
If someone only wants to add the soft dependencies from libshared they
can just avoid parsing the executable binaries. By design the code in
libbasic/libshared still does the stamping too, at lower priorities, so
that libsystemd-shared will always list all the optional dependencies,
and if one wants to build a minimal system by default, they can just
parse libsystemd-shared dlopen notes, and ignore the individual
executables. But for many distribution the current setup is insufficient
and requires adding a ton of manual library dependencies, as many
executables become effectively broken or useless without the dlopen
dependencies installed (eg: resolved fails to start without libssl,
repart can do basically nothing without blkid, etc).
Add a new set of DLOPEN_<LIB> macros that wrap the dlopen_lib and also
pull in the ELF note voodoo, so that the callers get their ELF binaries
stamped too. Convert a bunch of callers to use the macro, and use
`required` dependencies for the callers that do not work without the
dlopen library being available.
The one caveat is that, in order to avoid duplicating the exact same
note in a binary due to multiple call sites, some `asm` voodoo is done
instead of the previous bare-C section-creating macro. The drawback of
this approach is that if `--gc-sections` is used to link the binary (as
we do), then binutils >= 2.36 is required for the `SHF_GNU_RETAIN` flag.
This effectively cuts off CentOS 9, so what I did here is adding an
override in meson to detect missing support, and drop `SHF_GNU_RETAIN`.
The build works, but on CentOS 9 there's no dlopen ELF notes anymore.
Given it's just that version, and it goes EOL next year, that seems ok
to me. The alternative is to drop the usage of `--gc-sections` on CentOS
9, or to accept duplicated notes everywhere, and both seem worse.
End result:
```
$ readelf --notes build/systemd-executor
Displaying notes found in: .note.gnu.property
Owner Data size Description
GNU 0x00000010 NT_GNU_PROPERTY_TYPE_0
Properties: x86 ISA needed: x86-64-baseline
Displaying notes found in: .note.gnu.build-id
Owner Data size Description
GNU 0x00000014 NT_GNU_BUILD_ID (unique build ID bitstring)
Build ID: 8a0c3db54adb79ae54e1432255011aa4ab583742
Displaying notes found in: .note.ABI-tag
Owner Data size Description
GNU 0x00000010 NT_GNU_ABI_TAG (ABI version tag)
OS: Linux, ABI: 3.2.0
Displaying notes found in: .note.dlopen
Owner Data size Description
FDO 0x0000006b FDO_DLOPEN_METADATA
Dlopen Metadata: [{"feature":"pam","description":"Support for LinuxPAM","priority":"recommended","soname":["libpam.so.0"]}]
FDO 0x0000007c FDO_DLOPEN_METADATA
Dlopen Metadata: [{"feature":"seccomp","description":"Support for Seccomp Sandboxes","priority":"recommended","soname":["libseccomp.so.2"]}]
FDO 0x00000090 FDO_DLOPEN_METADATA
Dlopen Metadata: [{"feature":"bpf","description":"Support firewalling and sandboxing with BPF","priority":"recommended","soname":["libbpf.so.1","libbpf.so.0"]}]
FDO 0x000000a0 FDO_DLOPEN_METADATA
Dlopen Metadata: [{"feature":"cryptsetup","description":"Support for disk encryption, integrity, and authentication","priority":"recommended","soname":["libcryptsetup.so.12"]}]
FDO 0x00000078 FDO_DLOPEN_METADATA
Dlopen Metadata: [{"feature":"mount","description":"Support for mount enumeration","priority":"recommended","soname":["libmount.so.1"]}]
$ readelf --notes build/systemd
Displaying notes found in: .note.gnu.property
Owner Data size Description
GNU 0x00000010 NT_GNU_PROPERTY_TYPE_0
Properties: x86 ISA needed: x86-64-baseline
Displaying notes found in: .note.gnu.build-id
Owner Data size Description
GNU 0x00000014 NT_GNU_BUILD_ID (unique build ID bitstring)
Build ID: dcd4568842e32da3e71be27db3def51c6b459994
Displaying notes found in: .note.ABI-tag
Owner Data size Description
GNU 0x00000010 NT_GNU_ABI_TAG (ABI version tag)
OS: Linux, ABI: 3.2.0
Displaying notes found in: .note.dlopen
Owner Data size Description
FDO 0x00000075 FDO_DLOPEN_METADATA
Dlopen Metadata: [{"feature":"mount","description":"Support for mount enumeration","priority":"required","soname":["libmount.so.1"]}]
FDO 0x00000072 FDO_DLOPEN_METADATA
Dlopen Metadata: [{"feature":"selinux","description":"Support for SELinux","priority":"recommended","soname":["libselinux.so.1"]}]
```
This macro emits an SD_ELF_NOTE_DLOPEN note so that calling sites,
and not just libsystemd-shared, get the ELF stamped with the notes
documenting the dependency. Currently most of the dlopen notes end
up in libsystemd-shared, and executables have no way to document
the optional dependencies they need. With this new macro, this
issue can be fixed.
Co-developed-by: Claude Opus 4.8 <noreply@anthropic.com>
Many *at()-style helpers accept a directory fd that may be either a
regular, valid fd (>= 0) or one of the special AT_FDCWD/XAT_FDROOT
wildcard values, and open-code that check in their assertions. Add a
wildcard_fd_is_valid() helper for it and use it tree-wide.
This allows multi-call binaries to be easily invoked with a different
name. After installation, the name is set by creating a symlink. But in
build directories, we don't create the symlinks. (There are also other
ways to achieve the same thing, e.g. zsh supports $ARGV0, and exec -a
can be used, but those are either non-portable or are more complicated
to use.) The primary use-case for me is to test --help output for
multicall binaries.
Also reorder the help for env vars to group the more generic ones near
the top.
This was initially proposed in https://github.com/systemd/systemd/pull/24054,
but there were some comments about the implementation. I had a branch
with the patch, but I don't think I ever actually submitted it as a
pull request.
On some architectures like m68k, alignof(void*) is 2, not sizeof(void*)
(which is 4). So the natural alignment of struct Option is 2 and
sizeof(Option) == 26.
However, each variable placed in the SYSTEMD_OPTIONS ELF section via
_OPTION() carries _alignptr_ (= __attribute__((aligned(sizeof(void*))))),
which forces each entry to start at a 4-byte boundary. The linker
therefore inserts 2 bytes of padding between adjacent entries, producing
an actual stride of 28 in the section.
option_parse() iterates over the section with pointer arithmetic
("opt++"), which advances by sizeof(Option) == 26 and lands inside the
padding. The fields read back as zero, and since commit cf88903637
("tree-wide: get rid of most uses of ALIGN_PTR") added the ordering
assertion below, the resulting "0 < 0" trips it when running --help:
Assertion 'opt->id < (opt + 1)->id' failed at src/shared/options.c:116, function option_parse(). Aborting.
#0 0xc0a7b248 in ?? () from /usr/lib/m68k-linux-gnu/libc.so.6
#1 0xc0a7b2ce in pthread_kill () from /usr/lib/m68k-linux-gnu/libc.so.6
#2 0xc0a2edc6 in raise () from /usr/lib/m68k-linux-gnu/libc.so.6
#3 0xc0a1c128 in abort () from /usr/lib/m68k-linux-gnu/libc.so.6
#4 0xc05c6f78 in option_parse (options=0x0, options_end=0x0, state=0xc09ca968) at ../src/shared/options.c:116
Fix this by applying _alignptr_ to the struct definition itself, so that
sizeof(Option) is padded up to a multiple of sizeof(void*) and matches
the actual on-disk stride. Add an assert_cc() so any future regression
is caught at compile time.
The same latent bug applies to Verb and TestFunc, which use the same
section-placement pattern. Their natural sizeof was already a multiple
of sizeof(void*) so no crash was observed, but apply the same fix
defensively.
Follow-up for cf88903637
Co-developed-by: Claude Opus 4.7 <noreply@anthropic.com>
Not the case on the sparc architecture. Replace with two translation
functions. The poll to epoll variant isn't used yet but will be used
when we add io-uring support to sd-event so we already add it here so
that it is in place.
Follow-up for cf4c65afa8
For management purposes it's useful to be able to "tag" a machine with
various labels. Let's add a field for that to /etc/machine-info and make
it settable.
Fixes: #38591
get_user_creds would use getpwnam() and then returns strings pointing
into the static buffer. This seems very iffy. Let's duplicate the
strings properly
The param was input/output, which is unexpected and confusing and
actually made most callers much more complicated than they needed to be.
The function was playing fast and loose with the return value. In some
cases it was returning a static string, which would be completely fine.
But in other case it was returning a pointer into the getgrgid static
buffer, i.e. that return value could be overwritten. AFAICT, this didn't
matter in any of the callers, but we shouldn't do that anyway.
So use a separate output param with an allocated string that the caller
is responsible for.
It turns out that all callers pass NULL (outside of tests) and zero for
flags. So the function _could_ be simplified. But get_user_creds is
called with all the params actually used, and I think having the the
functions so different wouldn't be nice. I first wrote a commit to
drop the unused params, but then I discarded it.
The error message in exec-invoke.c is fixed.
Traditionally, asynchronous programming in systemd has been achieved
using
sd-event along with the asynchronous interfaces of sd-bus and
sd-varlink.
This works well when the system is reacting to events and all code
triggered
by those events can run without blocking. In these scenarios, the global
Manager object is passed as userdata to the callback, and the callback
can
use the stack as usual, declaring local state and ensuring proper
cleanup via
_cleanup_. Control flow structures, such as loops, work as expected, and
everything runs smoothly.
However, challenges arise when the code needs to perform long-running
operations within these callbacks. Since the system cannot block
execution
within the callback, we can't directly invoke a long-running operation
and
wait for its result without introducing complexities. Instead, we need
to
initiate the long-running task, register for completion with sd-event,
sd-bus, or sd-varlink, and provide a callback to be invoked when the
operation completes.
This callback, however, only receives a single userdata pointer, which
forces us to bundle all local variables into a struct and pass it along
as
part of the callback. On top of that, after queuing the asynchronous
operation, the caller continues executing. As the caller's stack unwinds
when the function exits, the resources and state within the local scope
may
be prematurely cleaned up. Therefore, the struct must store copies of
the
local variables or ensure proper reference counting to prevent premature
resource cleanup.
When multiple long-running operations need to be initiated within a
loop,
the complexity grows further. We must introduce additional shared state
to
track the completion of all operations before we can run any code that
depends on their results.
Furthermore, since the daemon may be shut down at any time, we must
track
the lifecycle of each long-running operation in the global Manager
struct,
ensuring proper cleanup even when stack unwinding can no longer manage
the
resources for us.
Fibers, or green threads, provide a more natural way of handling
asynchronous operations. By enabling cooperative multitasking within a
single thread, fibers allow us to write code that looks like it’s
running
synchronously, but with the ability to yield control at predefined
points,
such as when waiting for long-running tasks to complete.
With fibers, we can simplify the control flow by running asynchronous
operations within a fiber, allowing us to "pause" execution while
waiting
for the long-running operation to finish and then "resume" the operation
once
it's complete. This eliminates the need for multiple callback chains,
extensive state tracking, and the potential pitfalls of stack unwinding.
This commit introduces the ability to execute long-running operations in
a
non-blocking manner while maintaining the simplicity and readability of
synchronous code. The fiber-based approach will significantly improve
the
handling of complex workflows, making the code easier to write and
maintain.
The implementation is based on ucontext.h's makecontext() (with a
fallback
to the venerable sigaltstack() approach on musl),
sigsetjmp()/siglongjmp()
and sd-event. ucontext.h provides us with alternate stacks that we can
switch
between. We use sigsetjmp()/siglongjmp() instead of swapcontext()
because the
latter forcibly saves/restores a per context signal mask every time it
is called.
Using sigsetjmp()/siglongjmp(), we can avoid the unnecessary syscall and
maintain
a per thread signal mask, which makes much more sense than having a per
fiber
signal mask.
The default stack size is the same as a regular thread. Because we
use mmap() to allocate the stack, the memory won't actually be used
until it
is paged in by the kernel, so we don't actually use 8MB per fiber.
To integrate fibers with the event loop, each fiber is assigned a
deferred
event source which resumes the fiber when enabled. The deferred event
source
is oneshot by default so the fiber will run immediately until it yields
or
suspends. If it yields, the deferred event source is enabled again
(oneshot)
immediately. If it suspends, before it suspends, one or more event
sources
are registered with sd-event that will enable the deferred event source
(oneshot) to resume the fiber once the operation it is waiting for
completes.
Yielding or suspending the fiber is done by calling sd_fiber_yield() or
sd_fiber_suspend() respectively. Both of these return zero on success or
any
error value from the async operation that caused the fiber to resume.
This is also how fiber cancellation is implemented. When a fiber is
cancelled,
sd_fiber_yield() and sd_fiber_suspend() will return ECANCELED when the
fiber
is resumed, allowing the fiber to unwind its stack (which allows cleanup
to
happen automatically) and finish.
Instead of having applications work directly with fibers, we hide them
behind
a generic futures interface to represent long-running operations,
regardless of
whether those operations are running on a fiber or not. Aside from
fibers, the
futures library (sd-future) will for example allow waiting for sd-event
sources
and doing sd-bus calls in the background as well. Fibers can suspend
until a
future is ready with sd_fiber_await() or by having the future wake up
the fiber
explicitly in its callback. A future always defaults to waking up the
current
fiber.
Each future kind plugs into the library by providing an sd_future_ops
vtable
(alloc, free, cancel, set_priority). The library treats the impl pointer
returned by alloc() as a black box. Future Implementations retrieve it
via
sd_future_get_private().
A future starts in SD_FUTURE_PENDING and transitions exactly once to
SD_FUTURE_RESOLVED, carrying an integer result. Consumers can react to
that
transition either by installing a one-shot callback with
sd_future_set_callback() (callback-style code) or by waiting on it from
a
fiber via sd_fiber_await() (synchronous-looking fiber code).
sd_fiber_await()
is itself built on a "wait future" that resolves when its target
resolves;
sd_future_new_wait() exposes the same primitive directly so non-fiber
callers
can chain futures without involving a fiber.
Cancellation is cooperative: sd_future_cancel() invokes the future
impl's
cancel callback, which is responsible for tearing down its work and
ultimately
resolving the promise with -ECANCELED. For fiber futures this is what
surfaces as the ECANCELED return from
sd_fiber_yield()/sd_fiber_suspend()
mentioned above.
Fire-and-forget fibers — created by passing a NULL ret to sd_fiber_new()
—
take a self-reference on their future so they outlive the caller's
scope.
The self-ref is dropped when the fiber resolves. This floating mechanism
(sd_fiber_set_floating()) is restricted to fiber futures because they
uniquely guarantee resolution; allowing it for arbitrary future kinds
would
risk silent leaks for kinds that may never resolve.
Note that fiber cleanup depends on the runtime operating normally. Each
fiber's _cleanup_-style cleanups live on the fiber's own stack and run
only when the fiber is resumed and allowed to unwind, which requires a
working event loop to drive it to completion. The exit event source
registered for top-level fibers ensures unwind on a normal
sd_event_exit(),
but if the event loop itself terminates abnormally (e.g. an
unrecoverable
allocation failure mid-dispatch) before all fibers have resolved, their
stacks never unwind and any resources they own leak.
The code lives in libsystemd as sd-future (not exported) for the
following reasons:
- We may want to make this a public libsystemd API in the future
- The code can't live in src/basic as it makes heavy use of sd-event
- The code can't live in src/shared as sd-bus and sd-event make use of
it
The log and log-context headers are updated with functions to allow
fibers to have their own log prefix and log context.
Add -fexcess-precision=standard so gcc inserts ISO C99 conformant
rounding at assignments, casts, and returns — without it, double values
on x87 happily stay at 80-bit extended precision across operations and
diverge from the SSE/x86_64 behavior, making strict equality comparisons
architecture-dependent.
The flag doesn't fully cover x87: per gcc PR#323
(https://gcc.gnu.org/bugzilla/show_bug.cgi?id=323), a function return
value carried in ST(0) can arrive at the caller still at 80-bit, so a
double that ought to compare equal to a same-magnitude literal picks up
extra mantissa bits and doesn't. Wrap fp_equal in volatile-double
temporaries to force a memory roundtrip — the only operation that
reliably truncates on x87 — so its callers get consistent results
regardless of how the operands were produced.
Add a TEST(fp_equal) case that exercises the previously-broken pattern:
a runtime 1.0/10.0 computed inside a noinline helper, returned across
the function ABI boundary, then compared against the literal 0.1.
Without the volatile truncation this assertion fails on 32-bit gcc.
Some helpers in src/basic — ppoll_usec_full() (used by fd_wait_for_event()),
loop_read(), loop_read_exact(), loop_write_full() and
pidref_wait_for_terminate_full() — block the calling thread. That's the
right behaviour outside a fiber but not inside one, where blocking the
thread also stalls every other fiber running on the same event loop.
Rewriting every caller to pick a fiber or non-fiber variant explicitly
would be a lot of churn and would split otherwise-shared code paths in
two.
Instead, the helpers detect at runtime whether they're running on a fiber
and dispatch to a suspending variant when they are. FiberOps in
fiber-ops.h holds five function pointers (ppoll, read, write, timeout,
cancel_wait_unref); a fiber_ops global constant is populated whenever we
enter a fiber with functions that delegate to suspending variants of common
syscalls. With this approach, the variants themselves stay in libsystemd
which is required because they make use of sd-event.
- loop_read()/loop_read_exact() take the fiber read hook on a fiber
unless the caller asked for a non-blocking attempt (do_poll=false) and
the fd is already non-blocking — in that case we fall through to read()
to preserve the existing return-EAGAIN-immediately semantic. The hook
itself suspends on EAGAIN until data is available, so neither the
do_poll knob nor the explicit fd_wait_for_event() retry loop are
needed on the fiber path.
- loop_write_full() likewise takes the fiber write hook on a fiber,
except when timeout=0 with an already-non-blocking fd (preserving the
fast-return-EAGAIN semantic). The fiber path runs inside a
FIBER_OPS_TIMEOUT() scope so the caller's timeout is honoured via a
deadline future, mirroring SD_FIBER_TIMEOUT() but reachable from
src/basic without pulling in sd-future.h.
- pidref_wait_for_terminate_full() polls the pidfd via fd_wait_for_event()
before each waitid() when either a finite timeout is set or we're on a
fiber, and requires pidref->fd >= 0 in those cases (returning
-ENOMEDIUM otherwise — extending the rule that already applied to
finite timeouts). The poll suspends the fiber via the ppoll hook above;
the subsequent waitid() doesn't block because the pidfd is already
signalled.
Add a family of sd_fiber_*() I/O wrappers that, when called from a
fiber, behave like blocking I/O from the caller's perspective but
yield to the event loop instead of blocking the thread:
sd_fiber_read / sd_fiber_write
sd_fiber_readv / sd_fiber_writev
sd_fiber_recv / sd_fiber_send
sd_fiber_connect
sd_fiber_recvmsg / sd_fiber_sendmsg
sd_fiber_recvfrom / sd_fiber_sendto
sd_fiber_accept
sd_fiber_ppoll
Most of them share a single helper, fiber_io_operation(), which when
invoked outside a fiber falls through to the underlying syscall
directly, preserving the regular blocking behaviour. Inside a fiber
the helper flips the fd to non-blocking (restoring its original mode
on return), tries the syscall once on the fast path, and on EAGAIN/
EWOULDBLOCK creates an sd-event-backed IO future via future_new_io(),
suspends the fiber, and retries the syscall once the event source
fires.
future_new_io() itself is added to sd-event/event-future.{c,h} as a
new IoFuture kind. It wraps sd_event_add_io() into an sd_future:
oneshot enable, EPOLLERR translated via SO_ERROR (suppressed for
non-sockets), and the fd duplicated with F_DUPFD_CLOEXEC to avoid
EEXIST when multiple sources watch the same descriptor.
Together these let fiber-using code write straight-line socket and
pipe I/O without bundling state into callbacks.
Traditionally, asynchronous programming in systemd has been achieved using
sd-event along with the asynchronous interfaces of sd-bus and sd-varlink.
This works well when the system is reacting to events and all code triggered
by those events can run without blocking. In these scenarios, the global
Manager object is passed as userdata to the callback, and the callback can
use the stack as usual, declaring local state and ensuring proper cleanup via
_cleanup_. Control flow structures, such as loops, work as expected, and
everything runs smoothly.
However, challenges arise when the code needs to perform long-running
operations within these callbacks. Since the system cannot block execution
within the callback, we can't directly invoke a long-running operation and
wait for its result without introducing complexities. Instead, we need to
initiate the long-running task, register for completion with sd-event,
sd-bus, or sd-varlink, and provide a callback to be invoked when the
operation completes.
This callback, however, only receives a single userdata pointer, which
forces us to bundle all local variables into a struct and pass it along as
part of the callback. On top of that, after queuing the asynchronous
operation, the caller continues executing. As the caller's stack unwinds
when the function exits, the resources and state within the local scope may
be prematurely cleaned up. Therefore, the struct must store copies of the
local variables or ensure proper reference counting to prevent premature
resource cleanup.
When multiple long-running operations need to be initiated within a loop,
the complexity grows further. We must introduce additional shared state to
track the completion of all operations before we can run any code that
depends on their results.
Furthermore, since the daemon may be shut down at any time, we must track
the lifecycle of each long-running operation in the global Manager struct,
ensuring proper cleanup even when stack unwinding can no longer manage the
resources for us.
Fibers, or green threads, provide a more natural way of handling
asynchronous operations. By enabling cooperative multitasking within a
single thread, fibers allow us to write code that looks like it’s running
synchronously, but with the ability to yield control at predefined points,
such as when waiting for long-running tasks to complete.
With fibers, we can simplify the control flow by running asynchronous
operations within a fiber, allowing us to "pause" execution while waiting
for the long-running operation to finish and then "resume" the operation once
it's complete. This eliminates the need for multiple callback chains,
extensive state tracking, and the potential pitfalls of stack unwinding.
This commit introduces the ability to execute long-running operations in a
non-blocking manner while maintaining the simplicity and readability of
synchronous code. The fiber-based approach will significantly improve the
handling of complex workflows, making the code easier to write and maintain.
The implementation is based on ucontext.h's makecontext() (with a fallback
to the venerable sigaltstack() approach on musl), sigsetjmp()/siglongjmp()
and sd-event. ucontext.h provides us with alternate stacks that we can switch
between. We use sigsetjmp()/siglongjmp() instead of swapcontext() because the
latter forcibly saves/restores a per context signal mask every time it is called.
Using sigsetjmp()/siglongjmp(), we can avoid the unnecessary syscall and maintain
a per thread signal mask, which makes much more sense than having a per fiber
signal mask.
The default stack size is the same as a regular thread. Because we
use mmap() to allocate the stack, the memory won't actually be used until it
is paged in by the kernel, so we don't actually use 8MB per fiber.
To integrate fibers with the event loop, each fiber is assigned a deferred
event source which resumes the fiber when enabled. The deferred event source
is oneshot by default so the fiber will run immediately until it yields or
suspends. If it yields, the deferred event source is enabled again (oneshot)
immediately. If it suspends, before it suspends, one or more event sources
are registered with sd-event that will enable the deferred event source
(oneshot) to resume the fiber once the operation it is waiting for completes.
Yielding or suspending the fiber is done by calling sd_fiber_yield() or
sd_fiber_suspend() respectively. Both of these return zero on success or any
error value from the async operation that caused the fiber to resume.
This is also how fiber cancellation is implemented. When a fiber is cancelled,
sd_fiber_yield() and sd_fiber_suspend() will return ECANCELED when the fiber
is resumed, allowing the fiber to unwind its stack (which allows cleanup to
happen automatically) and finish.
Instead of having applications work directly with fibers, we hide them behind
a generic futures interface to represent long-running operations, regardless of
whether those operations are running on a fiber or not. Aside from fibers, the
futures library (sd-future) will for example allow waiting for sd-event sources
and doing sd-bus calls in the background as well. Fibers can suspend until a
future is ready with sd_fiber_await() or by having the future wake up the fiber
explicitly in its callback. A future always defaults to waking up the current
fiber.
Each future kind plugs into the library by providing an sd_future_ops vtable
(alloc, free, cancel, set_priority). The library treats the impl pointer
returned by alloc() as a black box. Future Implementations retrieve it via
sd_future_get_private().
A future starts in SD_FUTURE_PENDING and transitions exactly once to
SD_FUTURE_RESOLVED, carrying an integer result. Consumers can react to that
transition either by installing a one-shot callback with
sd_future_set_callback() (callback-style code) or by waiting on it from a
fiber via sd_fiber_await() (synchronous-looking fiber code). sd_fiber_await()
is itself built on a "wait future" that resolves when its target resolves;
sd_future_new_wait() exposes the same primitive directly so non-fiber callers
can chain futures without involving a fiber.
Cancellation is cooperative: sd_future_cancel() invokes the future impl's
cancel callback, which is responsible for tearing down its work and ultimately
resolving the promise with -ECANCELED. For fiber futures this is what
surfaces as the ECANCELED return from sd_fiber_yield()/sd_fiber_suspend()
mentioned above.
Fire-and-forget fibers — created by passing a NULL ret to sd_fiber_new() —
take a self-reference on their future so they outlive the caller's scope.
The self-ref is dropped when the fiber resolves. This floating mechanism
(sd_fiber_set_floating()) is restricted to fiber futures because they
uniquely guarantee resolution; allowing it for arbitrary future kinds would
risk silent leaks for kinds that may never resolve.
Note that fiber cleanup depends on the runtime operating normally. Each
fiber's _cleanup_-style cleanups live on the fiber's own stack and run
only when the fiber is resumed and allowed to unwind, which requires a
working event loop to drive it to completion. The exit event source
registered for top-level fibers ensures unwind on a normal sd_event_exit(),
but if the event loop itself terminates abnormally (e.g. an unrecoverable
allocation failure mid-dispatch) before all fibers have resolved, their
stacks never unwind and any resources they own leak.
The code lives in libsystemd as sd-future (not exported) for the following reasons:
- We may want to make this a public libsystemd API in the future
- The code can't live in src/basic as it makes heavy use of sd-event
- The code can't live in src/shared as sd-bus and sd-event make use of it
The log and log-context headers are updated with functions to allow
fibers to have their own log prefix and log context.
Drop the -fundamental suffix from src/fundamental/ headers in favor of names
that match their src/basic/ or src/shared/ counterparts (e.g.
macro-fundamental.h -> macro.h, assert-fundamental.h -> assert-util.h,
cleanup-fundamental.h -> cleanup-util.h). Rename src/basic/{btrfs,label}.{c,h}
to use the -util suffix to match the existing shared/btrfs-util and
shared/label-util siblings. Rename src/shared/mkdir-label.{c,h} to mkdir.{c,h}
and src/shared/tmpfile-util-label.{c,h} to tmpfile-util.{c,h} to match the
corresponding src/basic names.
This saves us from having to come up with separate names for files that do
the same thing across tiers, and it makes it easier to move stuff between
src/fundamental, src/basic and src/shared: consumers just #include "foo.h"
and pick up whichever tier their -I path resolves to first, so call sites
don't need to be updated when an API moves between layers.
Where a higher-tier wrapper exists (e.g. src/basic/macro.h wrapping
src/fundamental/macro.h), the wrapper uses an explicit "../fundamental/foo.h"
or "../basic/foo.h" relative include for the lower-tier header. We can't use
GCC's #include_next directive for this — when the wrapper is reachable both
via same-dir-as-source lookup and via -I (e.g. -Isrc/shared) for the
directory it lives in, #include_next advances by exactly one slot in libcpp's
internal directory chain and lands on the same physical directory it was
already in, never reaching the lower-tier sibling (see make_cpp_dir() in
gcc/libcpp/files.cc:1986).
To make sure the right headers are always picked up, the include directories
are reordered so that e.g. src/shared always takes priority over src/basic and
similar for the other directories.
Co-developed-by: Claude Opus 4.7 <noreply@anthropic.com>
# Context
#40195 defines the initial proposal and the motivation behind this PR.
This PR introduces 3 new options for `.link` files `[Link]` section:
- `IRQAffinityPolicy=`
- `IRQAffinity=`
- `IRQAffinityNUMA=`
The purpose is to allow `systemd-udevd` to configure a NIC's IRQs
affinity to specific CPU(s).
`IRQAffinityPolicy=` supports two policies:
- `single`: assign all the NIC IRQs to CPU 0, or the first CPU in the
CPU set resulting from the union of `IRQAffinity=` and
`IRQAffinityNUMA=`.
- `spread`: assign all the NIC IRQs to all the CPUs (or the union of
`IRQAffinity=` and `IRQAffinityNUMA=` if defined) in a round-robin
fashion while optimizing for cache locally while spreading apart queues
on CPUs as much as possible.
Both `IRQAffinity=` and `IRQAffinityNUMA=` behaves as filters to reduce
the CPU set to assign IRQs to, and are only valid if
`IRQAffinityPolicy=` is defined.
# Spreading IRQs
This section describes the algorithm responsible for spreading IRQs over
different CPUs to maximize performance.
## 1. Discover CPU topology
Read from `/sys/devices/system/cpu/cpu*/topology` to identify:
- L3 cache domains (dies)
- Physical cores VS hyperthreads
- NUMA nodes
- Core ordering within each die
```
Example: Dual-socket server with 2 dies per socket, 4 cores per die
┌─────────────────────────────────────────────────────────────────┐
│ NUMA Node 0 │
│ ┌─────────────────────────┐ ┌─────────────────────────┐ │
│ │ Die 0 (L3 Cache) │ │ Die 1 (L3 Cache) │ │
│ │ ┌────┐┌────┐┌────┐┌────│ │ ┌────┐┌────┐┌────┐┌────│ │
│ │ │ C0 ││ C1 ││ C2 ││ C3 │ │ │ C4 ││ C5 ││ C6 ││ C7 │ │
│ │ │0,16││1,17││2,18││3,19│ │ │4,20││5,21││6,22││7,23│ │
│ │ └────┘└────┘└────┘└────│ │ └────┘└────┘└────┘└────│ │
│ └─────────────────────────┘ └─────────────────────────┘ │
└─────────────────────────────────────────────────────────────────┘
┌─────────────────────────────────────────────────────────────────┐
│ NUMA Node 1 │
│ ┌─────────────────────────┐ ┌─────────────────────────┐ │
│ │ Die 2 (L3 Cache) │ │ Die 3 (L3 Cache) │ │
│ │ ┌────┐┌────┐┌────┐┌────│ │ ┌────┐┌────┐┌────┐┌────│ │
│ │ │ C8 ││ C9 ││C10 ││C11 │ │ │C12 ││C13 ││C14 ││C15 │ │
│ │ │8,24││9,25││10,2││11,2│ │ │12,2││13,2││14,3││15,3│ │
│ │ └────┘└────┘└────┘└────│ │ └────┘└────┘└────┘└────│ │
│ └─────────────────────────┘ └─────────────────────────┘ │
└─────────────────────────────────────────────────────────────────┘
(Numbers show CPU IDs: first HT, second HT)
```
## 2. Filter the first hyperthread only
Use only the first hyperthread of each physical core to avoid SMT
contention. Two IRQs on sibling HTs contend for ALU/cache without cache
benefit.
```
Before: CPUs 0-31 (16 cores × 2 HTs)
After: CPUs 0-15 (first HT of each core)
```
## 3. Equidistant permutations
Reorder dies and CPUs so consecutive selections are maximally spread
apart.
```
Original order: [0, 1, 2, 3] (adjacent dies/CPUs)
|
v
Equidistant: [0, 2, 1, 3] (spread apart)
```
This ensures that even if only 2 IRQs are assigned, they land on dies 0
and 2 (not 0 and 1), maximizing physical distance. The permutation is
also applied within each die:
```
Die permutation: Die0 -> Die2 -> Die1 -> Die3
Within each die:
┌───────────────────────────────────────────────┐
│ Die 0: [C0,C1,C2,C3] -> [C0,C2,C1,C3] │
│ Die 1: [C4,C5,C6,C7] -> [C4,C6,C5,C7] │
│ Die 2: [C8,C9,C10,C11] -> [C8,C10,C9,C11] │
│ Die 3: [C12,C13,C14,C15] -> [C12,C14,C13,C15] │
└───────────────────────────────────────────────┘
```
## 4. Round-robin selection across dies
Pick one CPU from each die in rotation, following permuted order.
```
Round 1: Die0->C0 Die2->C8 Die1->C4 Die3->C12
| | | |
v v v v
IRQs: [IRQ0] [IRQ1] [IRQ2] [IRQ3]
Round 2: Die0->C2 Die2->C10 Die1->C6 Die3->C14
| | | |
v v v v
IRQs: [IRQ4] [IRQ5] [IRQ6] [IRQ7]
```
If there are more IRQs than physical cores, this logic wraps around and
reuse CPUs. Only the first hyperthread of each core is used to avoid
cache line contention between queues.
Closes#40195.
- test-random-util is reworked to not use sqrt()
- pretty-print.c inlines ceil() so libm doesn't have
to be linked into libshared
- We add fno-math-errno to allow inlining of more math
functions by not requiring standard math functions to
set errno on invalid input.
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.
exp10() has a symbol version > 2.34 on latest glibc. To allow
dropping our baseline required glibc runtime version to <= 2.34,
let's add our own version to prevent pulling in the newer symbol
from glibc.
Each dlopen_*() wrapper kept its dl handle as a file-scope
'static void *xxx_dl = NULL;' even though only the wrapper itself
ever referenced it. Move each one inside the corresponding function
so its scope matches its actual use, leaving the rest of each
translation unit free of the unused file-scope name.
In pcre2-util.c the assert(pcre2_dl) in pattern_matches_and_log()
becomes assert(sym_pcre2_match), which carries the same invariant
(pattern_compile_and_log() ran dlopen_pcre2()).
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
Filesystems like xfs, btrfs, gfs2 and ocfs2 handle quotas internally
and do not need external quotacheck/quotaon services. When usrquota or
grpquota mount options are used in fstab for these filesystems,
generator_hook_up_quotacheck() falls through to the !fstype_needs_quota()
branch and emits a misleading warning that quotas are "not supported"
when they actually work fine — the kernel handles them internally.
Add fstype_has_internal_quota() to return early with a debug message,
and adopt a tri-state return convention so the caller skips quotaon
when quotacheck was not needed.
The buggy code path was introduced in #24824 and #24880.
Co-developed-by: Claude Opus 4.6 <noreply@anthropic.com>
Add signal_code_to_string() in signal-util.c and cover the si_code values
defined in libc's siginfo-consts.h. Fall back to the numeric value when no
symbolic name is known.
Co-developed-by: Codex (GPT-5) <noreply@openai.com>
Signed-off-by: Emanuele Rocca <emanuele.rocca@arm.com>
- **bpf-util: rename from bpf-dlopen, unify version-specific symbol
handling**
- **cryptsetup: dlopen libcryptsetup in tokens**
- **tree-wide: dlopen libpam in pam plugins**
- **test-bus-marshal: dlopen() glib and libdbus instead of linking
directly**
- **lock-util: Simplify timeout for lock_generic_with_timeout()**
- **color-util: simplify hsv_to_rgb, fix rgb_to_hsv negative-hue wrap**
- **tree-wide: Use our own macros instead of fabs()/fmax()/fmin()**
- **locale-util: dlopen() libintl instead of linking against it**
- **home: Use log2u64() over log2()**
- **meson: drop libdl, threads, and librt dependencies**
- **libc: Use dlsym() from a constructor instead of weak symbols**
- **libc: Make sure C23 versions of strtol(), sscanf() are not used**
Our baseline glibc is 2.34, which merged libdl, libpthread (the
dependency('threads') target), and librt into libc. Empty .so/.a stubs
remain for backward compatibility with old binaries, but new builds
resolve dl_*, pthread_*, mq_*, timer_*, etc. directly from libc.
On musl the same libraries are likewise empty stubs.
Drop the libdl, threads, and librt entries from every meson.build, and
remove the now-stale 'Libs.private: -lrt -pthread' from libudev.pc.in
since both flags resolve to empty link-time stubs on glibc 2.34+ and
musl.
Verified with readelf -d that libsystemd.so, libudev.so, and systemd no
longer carry DT_NEEDED entries for libdl/libpthread/librt.
dgettext() lives in libc on glibc and in libintl.so.8 on musl with
gettext. Resolve it via dlsym() so neither configuration produces a
hard link-time dependency on libintl: try libintl.so.8 first and fall
back to RTLD_DEFAULT (which finds dgettext in libc on glibc).
The _() macro now expands to a runtime check that returns the
untranslated string if dlopen_libintl() has not run successfully, so
callers don't have to gate every translatable message on a runtime
check. pam_systemd_home — currently the only consumer of _() — calls
dlopen_libintl() best-effort from each PAM entry point.
The meson find_library('intl') dance is replaced with a has_header()
check; the only thing we need at build time is the prototype.
The getxattr(2) man page only enumerates xattr-specific errors (ENODATA,
ENOTSUP, ERANGE, E2BIG, ...) in its own ERRORS section, but at the
bottom of that section notes that "the errors documented in stat(2) can
also be returned." stat(2) returns -ENOENT when a component of the path
does not exist, so any xattr lookup against a path can fail with -ENOENT
exactly the same way as -ENODATA — both mean "there is nothing here for
me to read." The previous definition of ERRNO_IS_NEG_XATTR_ABSENT()
reflected only the directly-enumerated errors and missed -ENOENT, so
callers that should semantically swallow "the xattr is absent" instead
bubbled -ENOENT up as a hard error.
The most visible consequence on real fleets has been systemd-journald
spamming dmesg with one line per dispatched log message whenever a
unit's cgroup directory cannot be found at the time
client_context_read_log_filter_patterns() is called — typically inside
containers whose journald observes clients whose unit cgroup is no
longer present in its view (cgroup-namespace boundary, unit teardown
race, transient sub-scope already collapsed back to its unit cgroup,
etc.). The same bug pattern lurks at every other cgroup-xattr callsite:
systemd-oomd reading user.oomd_avoid / user.oomd_omit / user.oomd_ooms
on cgroups it is concurrently killing; killall reading
user.survive_final_kill_signal during shutdown; cg_is_delegated() /
cg_has_coredump_receive() / cgroup_get_managed_oom_kill_last(); etc. For
these, "path is gone" is by construction the same answer as "xattr is
not set" — there is no way for the user to have attached an xattr to a
path that does not exist.
A quick survey of non-cgroup callers (src/portable/portable.c,
src/home/{homework-luks,user-record-util}.c,
src/random-seed/random-seed-tool.c, src/basic/os-util.c) confirms they
all operate on fds or on paths whose absence is already the desired
silent-skip outcome, so widening the macro to also fold in -ENOENT does
not change observable behavior at any other site.
Extend test-xattr-util's getxattr_at_malloc test with an explicit
non-existent-path lookup that asserts ERRNO_IS_NEG_XATTR_ABSENT() now
matches, alongside the pre-existing non-existent-xattr (-ENODATA) check.
This bothered me for a while, but I didn't think too much about it and just
copied the existing usage pattern. But it really doesn't make sense. We expect
the compiler to align the section properly. But if it didn't align it, applying
alignment after the fact would just cause our pointer to point to the middle
of the structure. That'd be even worse than a misaligned pointer.
Similarly, when doing pointer arithmetic, p++ should really result in a value
with the appropriate alignment. This is the basic principle of C pointer
addition. So we really shouldn't try to adjust the pointer ourselves. At most,
we can assert that it is indeed aligned in tests.
The BPF kprobe that fires on user namespace destruction is the only
thing
that triggers registry cleanup, so any time it doesn't run — ring buffer
overflow, kprobe missing, fdstore entry dropped outside our cleanup path
— a registry entry is left behind forever.
Stamp each registry entry with the kernel's unique namespace identifier
(NS_GET_ID, kernel ≥ 6.13) at allocation time. At manager startup, after
the existing fdstore→registry sweep, walk the registry and ask the
kernel
to look each namespace up by id via open_by_handle_at() on nsfs; if the
lookup returns -ESTALE the namespace is gone and we release the entry.
Old entries written before this change carry no identifier and are left
alone.
Add a namespace_open_by_id() helper for the lookup. The kernel restricts
open_by_handle_at() on nsfs to processes in the initial user namespace,
collapsing both permission denials and dead namespaces onto -ESTALE; the
helper refuses early with -EPERM outside the initial user namespace
so callers can tell the two apart.
The BPF kprobe that fires on user namespace destruction is the only thing
that triggers registry cleanup, so any time it doesn't run — ring buffer
overflow, kprobe missing, fdstore entry dropped outside our cleanup path
— a registry entry is left behind forever.
Stamp each registry entry with the kernel's unique namespace identifier
(NS_GET_ID, kernel ≥ 6.13) at allocation time. At manager startup, after
the existing fdstore→registry sweep, walk the registry and ask the kernel
to look each namespace up by id via open_by_handle_at() on nsfs; if the
lookup returns -ESTALE the namespace is gone and we release the entry.
Old entries written before this change carry no identifier and are left
alone.
Add a namespace_open_by_id() helper for the lookup. The kernel restricts
open_by_handle_at() on nsfs to processes in the initial user namespace,
collapsing both permission denials and dead namespaces onto -ESTALE; the
helper refuses early with -EHOSTDOWN outside the initial user namespace
so callers can tell the two apart.
This function is used in a quite a few places and it calls getenv and
does a bit of parsing, so let's try to speed this up.
Also, mark it _const_. With the caching, it now always returns the same
result, so we can tell the compiler to eliminate subsequent invocations.
It'd be nice if we could not repeat the information about the option
list a second time. But I don't see a nice way to do this, since
(by design) with the macro approach, the macros must be intertwined
with the parse_argv() code. But that code in turn refers to a bunch
of variables, so lifting out the function is not immediately possible.
So I think it's best to keep the existing approach where we provide
a list of options, without additional context, and skip them using
a custom routine.
For libc syscall wrappers (pidfd_open, fsopen, openat2, etc.) we previously
gated the calls behind build-time HAVE_* checks. Replace these with shim
functions in src/libc/ that fall back to the raw syscall at runtime when the
loaded glibc lacks the symbol. The infrastructure lives in src/libc/libc-shim.h:
DEFINE_SYSCALL_SHIM falls back to a direct syscall, DEFINE_LIBC_SHIM returns
ENOSYS (for posix_spawn-family helpers that have no corresponding syscall), and
DEFINE_LIBC_ERRNO_SHIM sets errno=ENOSYS and returns -1 (for read/write-style
helpers). The weak reference to the libc symbol is bound via __asm__(\"name\")
rename so the bare libc identifier never appears as a C token — this avoids
both #undef boilerplate against override-header redirects and the resulting
-Wredundant-decls warning. Drop the corresponding cc.has_function() loop from
meson.build.
For optional libraries (libcryptsetup, libdw, libarchive), drop the per-symbol
HAVE_* checks. Always declare the prototypes, suppressing the redundant-decl
warnings via DISABLE_WARNING_REDUNDANT_DECLS and NOLINT, and resolve the symbols
after the main dlopen via a new DLSYM_OPTIONAL() helper that only assigns on
success. libcryptsetup's crypt_set_keyring_to_link / crypt_token_set_external_path
and libarchive's *_is_set wrappers use fallback functions as their pointer
initializers (returning -ENOSYS and 0 respectively), so call sites can invoke
the symbol unconditionally and just check for -ENOSYS where the \"not supported\"
distinction matters.
The same shim treatment applies to pidfd_spawn / posix_spawnattr_setcgroup_np
(src/libc/spawn.c) and epoll_pwait2 (src/libc/epoll.c), with corresponding
override headers in src/include/override/spawn.h and
src/include/override/sys/epoll.h. posix_spawn_wrapper() in process-util.c and
epoll_wait_usec() in sd-event.c now detect ENOSYS in the return value instead
of checking the function pointer, falling back to plain posix_spawn() and
epoll_wait() respectively. coredump-config and coredump-submit get a
dlopen_dw_has_dwfl_set_sysroot() helper. The kexec arch gate now uses
defined(__NR_kexec_file_load) directly; pidfd.h uses __has_include_next() to
decide whether to pull in glibc's header.
This lets binaries built against newer glibc / libcryptsetup / libdw /
libarchive headers still load and run on older targets where these symbols are
absent.
Luca Boccassi
Zbigniew Jędrzejewski-Szmek
Lennart Poettering
Yu Watanabe
Daan De Meyer
Quentin Deslandes
Vincent Mihalkovic
Emanuele Rocca
Ivan Kruglov