| Age | Commit message (Collapse) | Author |
|---|
|
Let's be strict here, since its better to be safe than sorry.
|
|
The barrier implementation tracks remote states internally. There is no
need to check the return value of any barrier_*() function if the caller
is not interested in the result. The barrier helpers only return the state
of the remote side, which is usually not interesting as later calls to
barrier_sync() will catch this, anyway.
Shut up coverity by explicitly ignoring return values of barrier_place()
if we're not interested in it.
|
|
Imagine a constructor like this:
int object_new(void **out) {
void *my_object;
int r;
...
r = ioctl(...);
if (r < 0)
return -errno;
...
*out = my_object;
return 0;
}
We have a lot of those in systemd. If you now call those, gcc might inline
the call and optimize it. However, gcc cannot know that "errno" is
negative if "r" is. Therefore, a caller like this will produce warnings:
r = object_new(&obj);
if (r < 0)
return r;
obj->xyz = "foobar";
In case the ioctl in the constructor fails, gcc might assume "errno" is 0
and thus the error-handling is not triggered. Therefore, "obj" is
uninitialized, but accessed. Gcc will warn about that.
The new negative_errno() helper can be used to mitigate those warnings.
The helper is guaranteed to return a negative integer. Furthermore, it
spills out runtime warnings if "errno" is non-negative.
Instead of returning "-errno", you can use:
return negative_errno();
gcc will no longer assume that this can return >=0, thus, it will not warn
about it.
Use this new helper in libsystemd-terminal to fix some grdev-drm warnings.
|
|
This macro exists for MIPS since v3.17:
https://git.kernel.org/cgit/linux/kernel/git/stable/linux-stable.git/commit/?id=42944521af97a3b25516f15f3149aec3779656dc
|
|
A small readability improvement...
|
|
And conditionalize journald audit support with it
|
|
|
|
|
|
Also change the default prefixlen function to only access the first octet of the in_addr.
|
|
|
|
event loop
We really should not run manual event loops anymore, but standardize on
sd_event, so that we can run sd_bus connections from it eventually.
|
|
__attribute__((used)) is not enough to force static variables to
be carried over to a compiled program from a library. Mappings defined
in libsystemd-shared.a were not visible in the compiled binaries.
To ensure that the mappings are present in the final binary, the
tables are made non-static and are given a real unique name by which
they can be referenced.
To use a mapping defined not in the local compilation unit (e.g. in
a library) a reference to the mapping table is added. This is done
by including a declaration in the header file.
Expected values in test-engine are fixed to reflect the new mappings.
|
|
f7101b7368df copied some logic to prevent enabling masked units, but
also added a check which causes attempts to enable templated units to
fail. Since we know the logic beyond this check will properly handle
units which truly do not exist, we can rely on the unit file state
comparison to suffice for expressing the intent of f7101b7368df.
ref: https://bugs.archlinux.org/task/42616
|
|
|
|
|
|
I tried to preserve most errno values, but in some cases they were
inconsistent (different errno values for the same error name) or just
mismatched.
|
|
This is a rewrite of the hashmap implementation. Its advantage is lower
memory usage.
It uses open addressing (entries are stored in an array, as opposed to
linked lists). Hash collisions are resolved with linear probing and
Robin Hood displacement policy. See the references in hashmap.c.
Some fun empirical findings about hashmap usage in systemd on my laptop:
- 98 % of allocated hashmaps are Sets.
- Sets contain 78 % of all entries, plain Hashmaps 17 %, and
OrderedHashmaps 5 %.
- 60 % of allocated hashmaps contain only 1 entry.
- 90 % of allocated hashmaps contain 5 or fewer entries.
- 75 % of all entries are in hashmaps that use trivial_hash_ops.
Clearly it makes sense to:
- store entries in distinct entry types. Especially for Sets - their
entries are the most numerous and they require the least information
to store an entry.
- have a way to store small numbers of entries directly in the hashmap
structs, and only allocate the usual entry arrays when the direct
storage is full.
The implementation has an optional debugging feature (enabled by
defining the ENABLE_HASHMAP_DEBUG macro), where it:
- tracks all allocated hashmaps in a linked list so that one can
easily find them in gdb,
- tracks which function/line allocated a given hashmap, and
- checks for invalid mixing of hashmap iteration and modification.
Since entries are not allocated one-by-one anymore, mempools are not
used for entries. Originally I meant to drop mempools entirely, but it's
still worth it to use them for the hashmap structs. My testing indicates
that it makes loading of units about 5 % faster (a test with 10000 units
where more than 200000 hashmaps are allocated - pure malloc: 449±4 ms,
mempools: 427±7 ms).
Here are some memory usage numbers, taken on my laptop with a more or
less normal Fedora setup after booting with SELinux disabled (SELinux
increases systemd's memory usage significantly):
systemd (PID 1) Original New Change
dirty memory (from pmap -x 1) [KiB] 2152 1264 -41 %
total heap allocations (from gdb-heap) [KiB] 1623 756 -53 %
|
|
Add mempool_alloc0_tile(). It's like mempool_alloc_tile(), but it
initializes the allocated tile's memory to zero.
|
|
Two's logarithms for unsigned.
|
|
provided headers
|
|
|
|
We really shouldn't create fds ever that have the flag unset.
|
|
|
|
|
|
278 is vmsplice on x86_64. 318 is what we want:
http://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/tree/arch/x86/syscalls/syscall_64.tbl
|
|
|
|
|
|
needing entropy
Doesn't require an fd, and could be a bit faster, so let's make use of
it, if it is available.
|
|
something fails
|
|
|
|
These use the (deprecated) IPv4 address classes to deduce the corresponding subnet masks. This is useful when addresses
without subnet masks and prefix lengths are given.
Make use of these new functions from sd-dhcp-lease.
|
|
|
|
Reported-by: sztanpet on irc
|
|
Function queries system hostname and applies changes only when necessary. Also,
migrate all client of sethostname to sethostname_idempotent while at it.
|
|
On an ro fs, systemctl disable ... would fail silently.
|
|
https://bugs.freedesktop.org/show_bug.cgi?id=85447
|
|
|
|
|
|
|
|
Let's make parse_usec() and parse_nsec() work similar
|
|
In order to make object destruction easier (in particular in combination
with _cleanup_) we usually make destructors deal with NULL objects as
NOPs. Change the calendar spec destructor to follow the same scheme.
|
|
|
|
|
|
|
|
|
|
|
|
Normally we shouldn#t log from "library" functions, but SELinux is
weird, hence upgrade security messages uniformly to LOG_ERR when in
enforcing mode.
|
|
APIs that query and return something cannot silently fail, they must
either return something useful, or an error. Fix that.
Also, properly rollback socket unit fd creation when something goes
wrong with the security framework.
|
|
|
|
|
