When the support for "synthetic errno" was added, we started truncating
the errno value to just the least significant byte. This is generally OK,
because errno values are defined up to ~130.
The docs don't really say what the maximum value is. But at least in principle
higher values could be added in the future. So let's stop truncating
the values needlessly.
The kernel (or libbpf?) have an error where they return 524 as an errno
value (https://bugzilla.redhat.com/show_bug.cgi?id=2036145). We would
confusingly truncate this to 12 (ENOMEM). It seems much nicer to let
strerror() give us "Unknown error 524" rather than to print the bogus
message about ENOMEM.
The COREDUMP_EXE attribute is "optional", i.e. we continue to process the
crash even if we didn't acquire it. The coredump generation code assumed
that it is always available:
#5 endswith at ../src/fundamental/string-util-fundamental.c:41
[ endswith() is called with NULL here, and an assertion fails. ]
#6 submit_coredump at ../src/coredump/coredump.c:823
#7 process_socket at ../src/coredump/coredump.c:1038
#8 run at ../src/coredump/coredump.c:1413
We use the exe path for loop detection, and also (ultimately) pass it to
dwfl_core_file_report(). The latter seems to be fine will NULL, so let's just
change our code to look at COMM, which should be more reliable anyway.
Fixes https://bugzilla.redhat.com/show_bug.cgi?id=2036517.
The code was written unidiomatically, using r as a boolean value, and
confusing errno and r in some places. AFAICS, there wasn't any actual
problem: even in the one place where errno was used instead of r, it would
almost certainly be initialized.
It seems that some libbpf functions set errno, while others return the
error, possibly encoded. Since there are almost no docs, the only way to
know is to read the code of the function. To make matters worse, there is
a global libbpf_mode which can be set to change the convention. With
LIBBPF_STRICT_DIRECT_ERRS in libbpf_mode, some functions set errno while others
return a negative error, and the only way to know is to read the code, except
that the split is now different. We currently don't set
LIBBPF_STRICT_DIRECT_ERRS, but even the possibility makes everything harder
to grok.
This is all very error-prone. Let's at least add some asserts to make sure that
the returned values are as expected.
Not aligning these can create gaps in the section table. Some
firmware does not handle this nicely resulting in secure boot
signature fails.
Using objcopy ensures that any new sections in the future will be
properly aligned.
Fixes: #21956
Changing the efi compiler this way doesn't really work. The gnu-efi
header checks as well as supported compiler flag checks use the
regular cc that meson detects. Changing the compiler this way will
end up with bad compiler flags. For the very same reason, this does
not work with a cross-compiler without going through proper meson
cross-compilation steps either.
The proper way to build systemd-boot with a different compiler is to
use a different build folder and then just use the proper ninja build
target to only build the bootloader/stub.
This fixes a bug introduced by eaba9bb3e6.
The commit mistakenly drops 'x' in ID_NET_NAME_MAC, and adds colons.
The colons were dropped by the commit dfa4876c41,
but the missing 'x' was not added at that time.
Follow-up for dfa4876c41.
Meson would generate the following compile test:
#define crypt_set_metadata_size meson_disable_define_of_crypt_set_metadata_size
#include <limits.h>
#undef crypt_set_metadata_size
#ifdef __cplusplus
extern "C"
#endif
char crypt_set_metadata_size (void);
#if defined __stub_crypt_set_metadata_size || defined __stub___crypt_set_metadata_size
fail fail fail this function is not going to work
#endif
int main(void) {
return crypt_set_metadata_size ();
}
This works fine when the identifier being queried is an actual function. But
crypt_token_max() is an inline function, so getting the address would fail,
leading to a false negative result. Complation would fail because the function
would be defined twice.
With this patch, the check is changed to include the header:
#include <libcryptsetup.h>
#include <limits.h>
#if defined __stub_crypt_set_metadata_size || defined __stub___crypt_set_metadata_size
fail fail fail this function is not going to work
#endif
int main(void) {
void *a = (void*) &crypt_set_metadata_size;
long long b = (long long) a;
return (int) b;
}
which seems to work correctly.
Previously, sethostname_idempotent_full() calls gethostname_full() with
GET_HOSTNAME_ALLOW_NONE and GET_HOSTNAME_ALLOW_LOCALHOST flags. That
intended to get any values set by kernel. But, that does not work, as
the hostname may be empty.
Let's simplify the logic. The function sethostname_idempotent_full()
intends to set the requested hostname only when the current hostname
is different from the requested one. So, no check in getostname_full()
is required. Hence, simply use the result of uname() here.
Fixes#21896.
But handle them gracefully. Otherwise, when the route to the address is
being configured, kernel refuse the route.
Note that kernel's wireguard module handle e.g. 192.168.10.3/24 as
192.168.10.0/24.
Fixes#21929.
Before:
$ systemd-run --service-type=notify --user false
Job for run-rc3fe52ee6ddd4a6eaaf1a20e0a949cdf.service failed because the control process exited with error code.
See "systemctl status run-rc3fe52ee6ddd4a6eaaf1a20e0a949cdf.service" and "journalctl -xeu run-rc3fe52ee6ddd4a6eaaf1a20e0a949cdf.service" for details.
After:
$ systemd-run --service-type=notify --user false
Job for run-r7791e380a7b6400ea01d6a0e5a458b23.service failed because the control process exited with error code.
See "systemctl --user status run-r7791e380a7b6400ea01d6a0e5a458b23.service" and "journalctl --user -xeu run-r7791e380a7b6400ea01d6a0e5a458b23.service" for details.
Fixes https://github.com/systemd/systemd/issues/21933
Luca Boccassi
Zbigniew Jędrzejewski-Szmek
Yu Watanabe
Jan Janssen
Frantisek Sumsal
Evgeny Vereshchagin
Marco Scardovi
dependabot[bot]
Noel Kuntze