This is mostly intended as test case for the early enum comment bugfix,
as this Varlink IDL description now contains such comments, and
test-varlink-idl will process it forth and back aleady.
This normalizes how we report an empty list of boot entries in
ListBootEntries(). Our usual pattern is to return one item per method
call, but when there is none we usually return a NoSuchXYZ error. Do so
here too.
Before this we'd return a null item instead here, and only here.
This is a minor compat break, but given that this IPC interface is very
new and probably not used so far (we don't use it in our code at least,
and google doesn#t find any other use) I think this normalization is OK
at this point.
And drop spurious assertion.
Fortunately, the previous logic worked, as /run/systemd/unit-private-tmp
is ordered earlier than /tmp or /var/tmp. But, let's ensure the tmpfs
mounted earlier to make the logic clearer.
Follow-up for 0e551b04ef.
In 99% of uses of sd_json_build() we want to build an object as
outermost construct. Let's shorten this most common case a bit, by
adding sd_json_buildo() that implies this. This allows us to shorten
much of our code, all across the tree.
Even though we don't export json_log() in the public API, let's
officially make the SD_JSON_WARNING/SD_JSON_DEBUG that control its
effect in the public API.
After all, for our own dispatcher functions they have a nice effect, and
they are trivially reimplemented in user code independently.
(We might eventually consider exporting json_log() as public API, but
this is quite involved, given its use of macros/inline functions and
iternal logging API).
This mostly just swaps around the bit flags and cleans up comments.
Currently, install_info_apply() only updates r if it's 0,
meaning that if one of the earlier install_info_symlink_alias/wants()
calls returns > 0, errors generated by later calls will be discarded.
Fix that.
This also
- merges basic/linux and shared/linux,
- moves BPF_JUMP_A() to basic/missing_bpf.h,
- copies from usrspace kernel headers directory generated by 'make headers',
rather than copying from kernel tree,
- copies const.h into our tree to reduce change in ethtool.h,
- copies auto_fs.h into our tree to reduce change in auto_dev-ioctl.h.
This reverts commit f5bdecba08.
Some kmsg sent before sleep may be received by systemd-journald after
sleep. In that case, map_clock_usec() does not provide correct
timestamp.
So, _SOURCE_MONOTONIC_TIMESTAMP field is anyway unreliable.
Let's not use the field.
This partially reverts commit a9357c2ce2.
Some kmsg sent before sleep may be received by systemd-journald after
sleep. In that case, map_clock_usec() does not provide correct
timestamp.
So, we cannot provide reliable _SOURCE_MONOTONIC_TIMESTAMP.
If a symlink is leftover, still allow cleaning it up via 'disable'. This
happens when a unit is stopped and removed, but not disabled, and a reload
has already happened. At that point, cleaning up the old symlinks becomes
impossible through the APIs, and needs to be done manually. Always allow
cleaning up symlinks, if they exist, by only erroring out if there is an
OOM.
Follow-up for f31f10a620