asprintf(3) says that the pointer is "undefined" after a failed call.
In the current glibc implementation it is just NULL. In principle the
call could return a valid pointer with bad contents or something.
We have two styles of error handling: in a majority of cases we would
check the return value, but sometimes we used (void) and relied on the
pointer not being set. In practice both styles should be equivalent,
but gcc doesn't like the second one with -Wunused-result. (Though only
sometimes. E.g. on my F34 box I don't get the same warnings as in CI,
even though the compiler version is very similar and the compilation
options are the same…). It's also nice to be consistent in our code base.
So let's always use the first style of error checking.
We disabled it in f73fb7b742f294b6d2126afa16001bd2ff6ab461 in response to an
apparent gcc bug. It seems that depending on the combination of optimization
options, gcc still ignores (void). But this seems to work fine with clang, so
let's re-enable the warning conditionally.
basic/macro: make CONST_MAX(DECIMAL_STR_MAX(…), STRLEN(…)) possible
When those two macros were used together in CONST_MAX(), gcc would complain
about a type mismatch. So either DECIMAL_STR_MAX() should be made size_t like
STRLEN(), or STRLEN() be made unsigned.
Since those macros are only usable on arguments of (small) fixed size, any type
should be fine (even char would work…). For buffer size specifications, both
size_t and unsigned are OK. But unsigned was used for DECIMAL_STR_MAX macros
and FORMAT_foo_MAX macros, making STRLEN the only exception, so let's adjust
STRLEN() to be unsigned too.
Also: I don't think this is currently used anywhere, but if any of those macros
were used as an argument to sprintf, size_t would require a cast. ("%*s"
requires an int/unsigned argument.)
test-ether-addr-util: add a simple test that HW_ADDR_TO_STR works with nesting
It seems to, but I was a bit incredulous… The comment is adjusted to match
the standard.
The trick with a temporary buffer is neat. I wasn't sure if it is valid, but
the standard says so. The test really tests that we are reading the rules right
and that the compiler doesn't do anythign strange or emit an unexpected
warning.
repart: when we can't fit the partitions in, report needed disk size current disk size
This improves error output in repart if we can't fit the defined
partitions into the disk image. With this change we'll now show not only
the disk size we need (as before), but also the current one, as well as
the largest free area on disk.
This should make it a bit easier to debug disk space issues that repart
runs into.
If the auto-discovered swap partition is LUKS encrypted, decrypt it
automatically.
This aligns with the Discoverable Partitions Specification, though I've
also updated it to explicitly mention that LUKS is now supported here.
Since systemd retries any key already in the kernel keyring, if the swap
partition has the same passphrase as the root partition, the user won't
be prompted a second time for a second passphrase.
See https://github.com/systemd/systemd/issues/20019
This is preparation for adding encryption support to the credentials
logic, and we thus would like to add more deps. Let's hence move things
from src/basic/ to src/shared, so that we can rely on the OpenSSL
utilities already in src/shared.
chattr-util: generalize chattr manipulation for files with secrets from journalctl
This moves the code for setting chattr file attributes appropriate for
"secrets" files from journalctl into generic chattr-util.c code so that
we can use it elsewhere.
Also, let's reuse the "bitwise" logic already implemented in the chattr
code, instead of doing it again.
sd-id128: make sure sd_id128_get_machine_app_specific() logic also works without "khash"
So, as it turns out AF_ALG is turned off in a lot of kernels/container
environments, including our CI. Hence, if we link against OpenSSL
anyway, let's just use that client side. It's also faster.
One of those days we should drop the khash code, and ust use OpenSSL,
once the licensing issues are resolved.
process-util: explicitly handle processes lacking parents in get_process_ppid()
Let's make sure we signal out-of-band via an error message if a process
doesn't have a parent process whose PID we could return. Otherwise we'll
too likely hide errors, as we return an invalid PID 0, which in other
contexts has special meaning (i.e. usually "myself").
Replaces: #20153
This is based on work by @dtardon, but goes a different route, by
ensuring we propagate a proper error in this case.
This modernizes the function in question a bit in other ways, i.e.
renames stuff and makes the return parameter optional.
Apparently it's an important feature for some folks:
https://utcc.utoronto.ca/\~cks/space/blog/linux/NetworkdMACMatchesWidely.
I think we considered this more of a bugfix, but it's somewhere on the border.
Let's add this it's easier to discover.
oomd: don't collect candidate stats on every interval
cb13961ada52c1b27f6d6c2c6e37a2901f01ed30 updated the oomd logic to
collect candidate data when a kill was about to happen. However there
was still a call left over in the main loop to collect candidate data on
every interval. Remove this since it's unneeded.
It seems that fd_set_perms() is always called after checking that
fd >= 0 (also when called as action() in glob_item_recursively()),
so it seems that the assertion really came from fd==0.
man: correct return value of sd_bus_open_with_description
Since https://github.com/systemd/systemd/commit/f4b2933ee7890e5d414ab266d8586f19027a2bd9
if a description is not set, sd_bus_open_with_description returns -ENXIO, but the
documnetation stated that it returned successfully with a NULL string.
login: XGI Z7/Z9 (XG20 core) graphic chip requires master-of-seat to be set
This graphic chip doesn't have a DRM driver and fall back to vesa-framebuffer
driver.
Without this patch, users of such chip suddenly see their GUI broken without
any indication or reason of what happened (no error message). Hence this
regression is near to impossible to troubleshoot for end users. Such case was
reported https://bugzilla.opensuse.org/show_bug.cgi?id=1187154.
Rather than adding another exception in the udev rules to deal with such
HWs, they instead get their own hwdb file '60-seat.hwdb'.
hwdb: add resolution override for Pinebook Pro touchpad
The Pinebook Pro touchpad returns a resolution data that is 2 times of
the real value, which makes libinput think the touchpad is only 1/4 the
real size.
Add a resolution override value for it, to allow libinput to calculate
the distance moved on it correctly.
Let's unify handling of the boolean values throughout the test-functions
code, since we use 0/1, true/false, and yes/no almost randomly in many
places, so picking the right values during CI configuration can be a real
pain.
Roughly reorder entries, without rewording anything, by component, so
that there's some structure to the text.
Only 3 lines are deleted: an empty line, 'External:' at the bottom since
it was merged with 'External:' at the top, and the weird last line:
'String is not UTF-8 clean, ignoring assignment' which was likely an error
from some editor
Dan Streetman [Fri, 2 Jul 2021 14:38:14 +0000 (10:38 -0400)]
test: optionally, only save test journal for failing tests
Saving the journal for passing tests creates a huge amount of unneeded
data stored for each full test run. Add a env var to allow saving the
journal only for failed tests.
hashmap: make sure hashmap_get_strv()+set_get_strv() work with a NULL object
Before we invoke n_entries() we need to check for non-NULL here, like in
all other calls to the helper function. Otherwise we'll crash when
invoked with a NULL object, which we usually consider equivalent to an
empty one though.
This was updated incorrectly in https://github.com/systemd/systemd/pull/20058/commits/4fd3fc66396026f81fd5b27746f2faf8a9a7b9ee. As https://github.com/systemd/systemd/blob/main/man/systemd.special.xml decribes, this unit is about shutdown rather than boot.
udev: when booting without root= specification, and searching a root partition actually do the version comparison magic
Since 08fe0a53869f27a9bfbc5bd31f27058145d46745 when dissecting a disk
image we'll automatically pick the "newest" root fs if multiple exist,
by comparing GPT partition labels. This works in systemd-nspawn,
systemd-dissect, systemd-tmpfiles --image, … and so on. It also works
already in systemd-gpt-auto-generator. However, there was one missing
place: in the logic that automatically finds a root fs in case no root=
was specified on the kernel logic at all. This logic doesn't use the
dissection logic, but a much simpler one.
coredumpctl: show --help text if "coredumpctl help" is called
Most of our programs that take "verbs" make the "help" verb either
equivalent to passing the --help switch (or at least print a message
redirecting the user to that switch). Do so in coredumpctl too, in order
to minimize surprises.
ubifs volumes have a UUID and the built-in blkid is able to determine
it. The disk/by-uuid symlink isn't created because ubifs volumes are
not on block devices but on SUBSYSTEM="ubi" devices. See #20071.
Allow ubi subsystem devices to be processed by the persistent storage
rules too. The kernel device name matching already allows ubi* to pass.
The existing rules are sufficient to create the link.
The links look like other by-uuid symlinks, for example:
/dev/disk/by-uuid/9a136158-585b-4ba4-9b70-cbaf2cf78a1c -> ../../ubi0_1
network: drop old ndisc configurations after new ones are configured
Previously, `ndisc_remove_old_one()` checked `ndisc_{addresses,routes}_configured`
flags, but they are not unset when all addresses or routes are already
assigned.
After the request queue is implemented, the address or route requests
are not processed within the same event of ndisc handler is called, but
will processed later when they are ready. So, calling `ndisc_remove_old()`
in the event of ndisc handler will remove all addresses and routes
previously assigned even they are requested to be updated.
This makes `ndisc_remove_old()` do nothing when there exist some
requests to configure addresses and routes, thus previously assigned
addresses and routes are kept until all requests are processed.
The time-out when waiting to reach the online state is a pretty
regularly seen error, let's print an explicit log message for it. The
previous "Event loop failed: timed out" message is a bit too low-level I
think for regular users (as event loops are a developer's concept, not a
user's, really).
Note that outputting low-level error messages is generally actually OK I
think — for unexpected errors, but this timeout is a pretty expected one,
directly configurable by the user, hence output something friendly.
projects / thirdparty / systemd.git / log
