As it turns out if you pass a va_list to a function its state becomes
undefined after that function returns, and this actually does break on
x86-32.
Hence, let's reimplement message_read_ap() without the use of recursion.
Instead we now build our own stack of types in an array so that we can
decode the entire parameter list in a single stackframe.
Make "systemd-analyze dot" output only lines with units matching
given glob(7) patterns. Add --from-pattern and --to-pattern options.
Without any patterns all relationships are printed as before.
A relationship must match the follwing expression:
(isempty(from) || from[0] || from[1] || .. || from[n]) &&
(isempty(to) || to[0] || to[1] || .. || to[n]) &&
(isempty(P) || P[0] || P[1] || ... || P[n])
where from[] and to[] are lists of patterns provided with subsequent
--from-pattern and --to-pattern respectively. P[] is a list of additional
patterns provided after the "dot" subcommand.
Avoid "sender uid=65534, message ignored" case, where no credentials can
be read on the sender side.
Seems, the server socket does not enable credential receiving fast
enough, and the message from the client (without credential) sometimes
is queued before the credential passing was active.
Avoid the dynamic allocation for the _UID, _GID, and _PID strings.
The maximum size of the string can be determined at compile time.
The code has only been compile tested.
When systemd was compiled without audit support, do not collect the
audit session and loginuid in the journal. This is saving a couple of
syscalls and memory allocations per log message.
Avoid memory allocations to construct the path for files in the
procfs. The procfs paths are way shorter than the PATH_MAX so we
can use snprintf on a string located on the stack. This shows up
as a win on x86 using the benchmark program below.
$ make libsystemd-shared.la; gcc -O2 -Isrc/systemd/ -Isrc/ \
-o simple-perf-test simple-perf-test.c \
.libs/libsystemd-shared.a -lrt
#include "shared/util.h"
void test_once(void) {
pid_t pid = getpid();
char *tmp = NULL;
get_process_comm(pid, &tmp);
free(tmp);
tmp = NULL;
get_process_cmdline(pid, 0, 1, &tmp);
free(tmp);
is_kernel_thread(pid);
tmp = NULL;
get_process_exe(pid, &tmp);
free(tmp);
}
int main(int argc, char **argv)
{
int i;
for (i = 0; i < 50000; ++i)
test_once();
}
When switching root, i.e. LANG can be set to the locale of the initramfs
or "C", if it was unset. When systemd deserializes LANG in the real root
this would overwrite the setting previously gathered by locale_set().
To reproduce, boot with an initramfs without locale.conf or change
/etc/locale.conf to a different language than the initramfs and check a
daemon started by systemd:
$ tr "$\000" '\n' </proc/$(pidof sshd)/environ | grep LANG
LANG=C
To prevent that, serialization of environment variables is skipped, when
serializing for switching root.
https://bugzilla.redhat.com/show_bug.cgi?id=949525
In recent glibc, many commonly-used librt functions have moved
from librt to libc. This results in dbus' configure.ac
concluding that we don't need to link in librt. However,
sd-daemon.c needs it for mq_getattr(), causing the build
to fail.
dbus doesn't use POSIX message queues, so I'd prefer to be
able to avoid "if on Linux, link librt for sd-daemon.c".
Fix for:
b92bea5d2a
Causing:
systemd-logind[265]: Assertion 'd = event.data.ptr' failed at src/shared/dbus-loop.c:233, function bus_loop_dispatch(). Aborting.
Before, we would initialize many fields twice: first
by filling the structure with zeros, and then a second
time with the real values. We can let the compiler do
the job for us, avoiding one copy.
A downside of this patch is that text gets slightly
bigger. This is because all zero() calls are effectively
inlined:
$ size build/.libs/systemd
text data bss dec hex filename
before 897737 107300 2560 1007597 f5fed build/.libs/systemd
after 897873 107300 2560 1007733 f6075 build/.libs/systemd
… actually less than 1‰.
A few asserts that the parameter is not null had to be removed. I
don't think this changes much, because first, it is quite unlikely
for the assert to fail, and second, an immediate SEGV is almost as
good as an assert.
For some reason this shows up on i686 only:
src/udev/udev-builtin-usb_id.c:192:20: warning: comparison between signed and unsigned integer expressions [-Wsign-compare]
So far we didn't place spaces between the box drawing chars and the
values next to them. Let's be consistent here.
(Or to turn this around: if we really want to place a space there we
probably should do that in all our tree outputs, not just here...)
It's probably a good idea to minimize the number of field names to show
in the "systemctl status" output, in order to make them useful as a
guide for the reader how things are "grouped". This patch moves
information about the used socket technology to the end of the output
lines in brackets, rather than into the field names. This turns the used
socket technology into what it is -- peripheral meta information --
instead of something that was at the core.
New output:
systemd-journald.socket - Journal Socket
Loaded: loaded (/usr/lib/systemd/system/systemd-journald.socket; static)
Active: active (running) since Fr 2013-03-29 02:16:30 CET; 1 weeks 0 days ago
Docs: man:systemd-journald.service(8)
man:journald.conf(5)
man:systemd-journald.service(8)
man:journald.conf(5)
Listen: /run/systemd/journal/stdout (Stream)
/run/systemd/journal/socket (Datagram)
/dev/log (Datagram)
It is possible to build systemd without logind or run logind without systemd
init. Commit 66e41181 fixed sd_booted() to only succeed for systemd init; with
that, testing for systemd init is wrong in the parts that talk to logind.
In particular, this affects the PAM module and the "uaccess" udev builtin.
Change sd_booted() to a new logind_running() which tests for
/run/systemd/seats/.
For details, see:
<https://mail.gnome.org/archives/desktop-devel-list/2013-March/msg00092.html>
https://bugs.freedesktop.org/show_bug.cgi?id=62754
