Some tests will create a subtree of /sys under build/test/sys and
depending on the local system that tree might end up having an infinite
chain of symlinks. For example:
$ ls build/test/sys/devices/pnp0/00:00/subsystem/devices/00:00/subsystem/devices/00:00/subsystem/devices/00:00/subsystem/devices/00:00/subsystem/devices/00:00/subsystem/
devices drivers drivers_autoprobe
Exuberant ctags will by default follow symlinks, so configure it not to
do so through a local .ctags file setting --links=no.
Tested that `ctags -R` doesn't get stuck with the dotfile present.
This new setting allows configuration of CFS period on the CPU cgroup, instead
of using a hardcoded default of 100ms.
Tested:
- Legacy cgroup + Unified cgroup
- systemctl set-property
- systemctl show
- Confirmed that the cgroup settings (such as cpu.cfs_period_ns) were set
appropriately, including updating the CPU quota (cpu.cfs_quota_ns) when
CPUQuotaPeriodSec= is updated.
- Checked that clamping works properly when either period or (quota * period)
are below the resolution of 1ms, or if period is above the max of 1s.
This works like parse_sec() but defaults to USEC_INFINITY when passed an
empty string or only whitespace.
Also introduce config_parse_sec_def_infinity, which can be used to parse
config options using this function.
This is useful for time options that use "infinity" for default and that
can be reset by unsetting them.
Introduce a test case to ensure it works as expected.
The badge is currently serving a broken image, since Coverity Scan is currently
having an outage. See Issue #11185 for more details. We can restore the badge
by reverting this commit once their service is up again.
This uses a {% for %} loop in Jekyll to render the page, from the "title"
information in the Front Matter of the actual page files.
This also makes `make-index-md` build rule unnecessary, since generation is
done by the template engine itself.
Tested this by running Jekyll locally.
It turns out Jekyll (the engine behind GitHub Pages) requires that pages
include a "Front Matter" snippet of YAML at the top for proper rendering.
Omitting it will still render the pages, but including it opens up new
possibilities, such as using a {% for %} loop to generate index.md instead of
requiring a separate script.
I'm hoping this will also fix the issue with some of the pages (notably
CODE_OF_CONDUCT.html) not being available under systemd.io
Tested locally by rendering the website with Jekyll. Before this change, the
*.md files were kept unchanged (so not sure how that even works?!), after this
commit, proper *.html files were generated from it.
In particular, check that the order of the results is consistent.
This test coverage will be useful in order to refactor the compare_func
used while sorting the results.
When introduced, this test also uncovered a memory leak in sd_lldp_stop(),
which was then fixed by a separate commit using a specialized function
as destructor of the LLDP Hashmap.
Tested:
$ ninja -C build/ test
$ valgrind --leak-check=full build/test-lldp
This reverts commit dd102e4d0c.
That test case exposed a memory leak and breaks CI, so let's revert it until
the original issue is fixed, to prevent disruption of automated testing.
The ?: operator is very useful for chaining comparison functions
(strcmp, memcmp, CMP), since its behavior is to return the result
of the comparison function call if non-zero, or continue evaluating
the chain of comparison functions.
This simplifies the code in that using a temporary `r` variable
to store the function results is no longer necessary and the checks
for non-zero to return are no longer needed either, resulting in a
typical three-fold reduction to the number of lines in the code.
Introduce a new memcmp_nn() to compare two memory buffers in
lexicographic order, taking length in consideration.
Tested: $ ninja -C build/ test
All test cases pass. In particular, test_multiple_neighbors_sorted()
in test-lldp would catch regressions introduced by this commit.
In particular, check that the order of the results is consistent.
This test coverage will be useful in order to refactor the compare_func
used while sorting the results.
Tested: ninja -C build/ test
This makes it so that tests no longer need to know the absolute paths to the
source and build dirs, instead using the systemd-runtest.env file to get these
paths when running from the build tree.
Confirmed that test-catalog works on `ninja test`, when called standalone and
also when the environment file is not present, in which case it will use the
installed location under /usr/lib/systemd/catalog.
The location can now also be overridden for this test by setting the
$SYSTEMD_CATALOG_DIR environment variable.
This simplifies get_testdata_dir() to simply checking for an environment
variable, with an additional function to locate a systemd-runtest.env file in
the same directory as the test binary and reading environment variable
assignments from that file if it exists.
This makes it possible to:
- Run `ninja test` from the build dir and have it use ${srcdir}/test for
test unit definitions.
- Run a test directly, such as `build/test-execute` and have it locate
them correctly.
- Run installed tests (from systemd-tests package) and locate the test
units in the installed location (/usr/lib/systemd/tests/testdata), in
which case the absence of the systemd-runtest.env file will have
get_testdata_dir() use the installed location hardcoded into the
binaries.
Explicit setting of $SYSTEMD_TEST_DATA still overrides the contents of
systemd-runtest.env.
Starting with meson 0.46, it is able to detect these argument correctly.
See this commit in meson codebase for more details:
https://github.com/mesonbuild/meson/commit/695b8f3a0377d3e2ce78ba8716adc365b18edea1
We already carry a requirement for meson_version : '>= 0.46', so we can be sure
our build system will include this commit.
Tested by building systemd using a cloned meson synced to the 0.46.0 tag,
confirmed the warnings were detected correctly in that case. The meson messages included this snippet:
> Compiler for C supports arguments -Wno-unused-parameter -Wunused-parameter: YES
> Compiler for C supports arguments -Wno-missing-field-initializers -Wmissing-field-initializers: YES
> Compiler for C supports arguments -Wno-unused-result -Wunused-result: YES
> Compiler for C supports arguments -Wno-format-signedness -Wformat-signedness: YES
> Compiler for C supports arguments -Wno-error=nonnull -Werror=nonnull: YES
> Compiler for C supports arguments -Wno-maybe-uninitialized -Wmaybe-uninitialized: YES
The GitHub guide on contributing file says: "Decide whether to store your
contributing guidelines in your repository's root, docs, or .github directory."
https://help.github.com/articles/setting-guidelines-for-repository-contributors/#adding-a-contributing-file
But there's really no advantage to keeping it in the hidden .github/, since
these are public and really belong together with the other documentation.
We can still keep the issue templates under .github/, since they are not really
documentation on their own.
Updated the links pointing to CONTRIBUTING.md to refer to the one in docs/.
The docs/ directory is special in GitHub, since it can be used to serve GitHub
Pages from, so there's a benefit to switching to it in order to expose it
directly as a website.
Updated references to it from the documentations themselves, from the
CONTRIBUTING.md file and from Meson build files.
Compiler flag -Wmaybe-uninitialized is quite noisy and produces many false
positives, especially when optimization flags are enabled (tested gcc 8.2.1),
so let's just disable it in systemd build.
For example, with CFLAGS=-O2, the build produces 11 such warnings and the
default CFLAGS of Fedora's rpmbuild warns about it in 176 places. A look at a
sample of those shows that most are false positives, where the compiler just
can't figure it out correctly. (While fixing those would be nice, I'm not sure
it's a good use of our time.)
The noisy [-Wmaybe-uninitialized] warnings are not just an annoyance, since
they make it harder to spot warnings that indicate actual problems (such as
variable declared but not used.) Silencing those is beneficial, so that
contributors would see warnings where there are actually actionable problems,
so there's a better chance of having those issues addressed before a PR is
pushed.
Tested:
$ CFLAGS='-O2 -Wp,-D_FORTIFY_SOURCE=2' meson build/
$ ninja -C build/
(NOTE: -Wp,-D_FORTIFY_SOURCE=2 prevents [-Wstringop-truncation] warnings.)
With the commands above, the build will not produce any [-Wmaybe-uninitialized]
warnings (or any other warnings), which is not really the case before this commit.
Also tested with rpmbuild on Fedora, after this commit there are no warnings
produced in the build step.
This makes it possible to wait until boot is finished without having to poll
for this command repeatedly, instead using the syntax:
$ systemctl is-system-running --wait
Waiting is implemented by waiting for the StartupFinished signal to be posted
on the bus.
Register the matcher before checking for the property to avoid race conditions.
Tested by artificially delaying startup with a oneshot service and calling this
command, checked that it emitted `running` and exited with a 0 return code as
soon as the delay service completed startup.
Also tested that booting to degraded state unblocks the command.
Inserted a delay between getting the property and waiting for the signal and
confirmed this seems to work free of race conditions.
Updated the --help text (under --wait) and the man page to document the new
feature.
This function doesn't really implement ordering, but CMP() is still fine to use
there. Keep the comment in place, just update it slightly to indicate that.
Looked for definitions of functions using the *_compare_func() suffix.
Tested:
- Unit tests passed (ninja -C build/ test)
- Installed this build and booted with it.
Macro returns -1, 0, 1 depending on whether a < b, a == b or a > b.
It's safe to use on unsigned types.
Add tests to confirm corner cases are properly covered.
Drop __extension__, since we don't use gcc -Wpedantic or -ansi.
Reformat code for spacing. Add spaces after commas almost everywhere.
Reindent code blocks in macro definitions, for consistency.
Test it when sending an FD without any contents, or an FD and some contents,
or only contents and no FD (using a bare send().)
Also fix the previous test which forked but was missing an _exit() at the
end of the child execution code.
These take a struct iovec to send data together with the passed FD.
The receive function returns the FD through an output argument. In case data is
received, but no FD is passed, the receive function will set the output
argument to -1 explicitly.
Update code in dynamic-user to use the new helpers.
These custom macros make the expression go through a function, in order
to prevent ASSERT_SIDE_EFFECT false positives on our macros such as
assert_se() and assert_return() that cannot be disabled and will always
evaluate their expressions.
This technique has been described and recommended in:
https://community.synopsys.com/s/question/0D534000046Yuzb/suppressing-assertsideeffect-for-functions-that-allow-for-sideeffects
Tested by doing a local cov-build and uploading the resulting tarball to
scan.coverity.com, confirmed that the ASSERT_SIDE_EFFECT false positives
were gone.
The switch to memory_startswith() changed the logic to only look for a space or
NUL byte after the matched word, but matching the full size should also be
acceptable.
This changed the behavior of parsing of "AUTH\r\n", where m will be set to 4,
since even though the word will match, the check for it being followed by ' '
or NUL will make line_begins() return false.
Tested:
- Using netcat to connect to the private socket directly:
$ echo -ne '\0AUTH\r\n' | sudo nc -U /run/systemd/private
REJECTED EXTERNAL ANONYMOUS
- Running the Ignition blackbox test:
$ sudo sh -c 'PATH=$PWD/bin/amd64:$PATH ./tests.test'
PASS
Fixes: d27b725abf
The current CLI does not support a way to clear these lists, since without any
additional arguments, the command will list the current values.
Introduce a new way to clear the lists by passing a single '' argument to these
subcommands.
Update the man page to document this.
Tested:
$ build/resolvectl domain eth1
Link 3 (eth1): ~.
$ build/resolvectl domain eth1 ''
$ build/resolvectl domain eth1
Link 3 (eth1):
$ build/resolvectl domain eth1 '~.' '~example.com'
$ build/resolvectl domain eth1
Link 3 (eth1): ~. ~example.com
$ build/resolvectl domain eth1 ''
$ build/resolvectl domain eth1
Link 3 (eth1):
$ build/resolvectl domain eth1 '~.'
$ build/resolvectl domain eth1
Link 3 (eth1): ~.
And similar for "dns" and "nta".
Starting with glibc 2.27.9000-36.fc29, include file sys/stat.h will have a
definition for struct statx, in which case include file linux/stat.h should be
avoided, in order to prevent a duplicate definition.
In file included from ../src/basic/missing.h:18,
from ../src/basic/util.h:28,
from ../src/basic/hashmap.h:10,
from ../src/shared/bus-util.h:12,
from ../src/libsystemd/sd-bus/bus-creds.c:11:
/usr/include/linux/stat.h:99:8: error: redefinition of ‘struct statx’
struct statx {
^~~~~
In file included from /usr/include/sys/stat.h:446,
from ../src/basic/util.h:19,
from ../src/basic/hashmap.h:10,
from ../src/shared/bus-util.h:12,
from ../src/libsystemd/sd-bus/bus-creds.c:11:
/usr/include/bits/statx.h:36:8: note: originally defined here
struct statx
^~~~~
Extend our meson.build to look for struct statx when only sys/stat.h is
included and, in that case, do not include linux/stat.h anymore.
Tested that systemd builds correctly when using a glibc version that includes a
definition for struct statx.
glibc Fedora RPM update:
https://src.fedoraproject.org/rpms/glibc/c/28cb5d31fc1e5887912283c889689c47076278ae
glibc upstream commit:
https://sourceware.org/git/?p=glibc.git;a=commitdiff;h=fd70af45528d59a00eb3190ef6706cb299488fcd
Use PRIu64 constant to get the format right on LP-64 architectures,
cast to (uint64_t) to solve incompatibility of __u64.
This was missed in ad4bc33522, so fix it
with this follow up.
Also use compat_main() when called as `resolvconf`, since the interface
is closer to that of `systemd-resolve`.
Use a heap allocated string to set arg_ifname, since a stack allocated
one would be lost after the function returns. (This last one broke the
case where an interface name was suffixed with a dot, such as in
`resolvconf -a tap0.dhcp`.)
Tested:
$ build/resolvconf -a nonexistent.abc </etc/resolv.conf
Unknown interface 'nonexistent': No such device
Fixes#9423.
Use PRIu64 and PRIu32 constants to also get the format right on LP-64
architectures.
For the 64-bit fields, we need a cast to (uint64_t), since __u64 is
defined as a `long long unsigned` and PRIu64 expects a `long unsigned`.
In practice, both are the same, so the cast should be OK.
Using an assertion is fine, since calls to job_merge_into_installed()
are protected by a check for job_type_is_conflicting().
Uncovered by Coverity, fixes CID 996307.
This introduces a has_data boolean field in struct unit_files which can
be used to detect the end of the array.
Use a _cleanup_ for struct unit_files in acquire_time_data and its
callers. Code for acquire_time_data is also simplified by replacing
goto's with straight returns.
Tested: By running the commands below, also checking them under valgrind.
- build/systemd-analyze blame
- build/systemd-analyze critical-chain
- build/systemd-analyze plot
Fixes: Coverity finding CID 996464.
This simplifies the code a bit and hopefully fixes Coverity finding
CID 1382966. There was not actually a resource leak here (Coverity
seemed to be confused by thinking log_oom() could actually return 0),
but the fix doesn't hurt and should make this code more resilient to
future refactorings.
Tested: builds fine, manually called scsi_id, seems to work ok.
This fixes an insecure use of tainted data as argument to functions that
allocate memory and read from files, which could be tricked into getting
networkctl to allocate a large amount of memory and fill it with file
data.
This was uncovered by Coverity. Fixes CID 1393254.
This should fix a leak of the allocated Prefix if sd_radv_prefix_new
fails for some reason.
The code was already initializing prefix to NULL and using TAKE_PTR to
return it, so only the _cleanup_ was missing.
Fixes Coverity finding CID 1382976.
This shouldn't be necessary, since read() should never return a size
larger than the size of the buffer passed in, but Coverity doesn't seem
to understand that.
We could possibly fix this with a model file for Coverity, but given
changing the code is not that much of a biggie, let's just do that
instead.
Fixes CID 996458: Overflowed or truncated value (or a value computed
from an overflowed or truncated value) `pos` used as array index.
Tested: `ninja -C build/ test`, builds without warnings, test cases pass.
The refactor in #9200 inadvertently dropped the variable assignment to
traverse the device and its hierarchy in add_matches_for_device().
This was uncovered by Coverity (CID #1393310).
Fix that by restoring the assignment.
Tested: `journalctl /dev/sda` now filters journalctl output again.
This fixes the copy routines on overlay filesystem, which typically
returns the underlying st_dev for files, symlinks, etc.
The value of st_dev is guaranteed to be the same for directories, so
checking it on directories only fixes this code on overlay filesystem
and still keeps it from traversing mount points (which was the original
intent.)
There's a small side effect here, by which regular (non-directory) files
with bind mounts will be copied by the new logic (while they were
skipped by the previous logic.)
Tested: ./build/test-copy with an overlay on /tmp.
Fixes: #9134
This makes it behave the same whether there is a blank line or not at
the end of the file. This is also consistent with the behavior of the
shell on a shell script that ends on a trailing backslash at the last
line.
Added tests to test_config_parse(), which only pass if the corresponding
change to config_parse() is included.
The leak can be reproduced by running systemd-path --suffix .tmp under valgrind or asan:
$ ./build/systemd-path --suffix .tmp search-binaries
/usr/local/bin/.tmp:/usr/bin/.tmp:/usr/local/sbin/.tmp:/usr/sbin/.tmp:/home/vagrant/.local/bin/.tmp:/home/vagrant/bin/.tmp
=================================================================
==19177==ERROR: LeakSanitizer: detected memory leaks
Direct leak of 56 byte(s) in 1 object(s) allocated from:
*0 0x7fd6adf72850 in malloc (/lib64/libasan.so.4+0xde850)
*1 0x7fd6ad2b93d2 in malloc_multiply ../src/basic/alloc-util.h:69
*2 0x7fd6ad2bafd2 in strv_split ../src/basic/strv.c:269
*3 0x7fd6ad42ba67 in search_from_environment ../src/libsystemd/sd-path/sd-path.c:409
*4 0x7fd6ad42bffe in get_search ../src/libsystemd/sd-path/sd-path.c:482
*5 0x7fd6ad42c55b in sd_path_search ../src/libsystemd/sd-path/sd-path.c:607
*6 0x7fd6ad42b3a2 in sd_path_home ../src/libsystemd/sd-path/sd-path.c:348
*7 0x55f59c65ebea in print_home ../src/path/path.c:97
*8 0x55f59c65f157 in main ../src/path/path.c:177
*9 0x7fd6abaea009 in __libc_start_main (/lib64/libc.so.6+0x21009)
Indirect leak of 68 byte(s) in 5 object(s) allocated from:
*0 0x7fd6adf72850 in malloc (/lib64/libasan.so.4+0xde850)
*1 0x7fd6abb5f689 in strndup (/lib64/libc.so.6+0x96689)
Indirect leak of 25 byte(s) in 1 object(s) allocated from:
*0 0x7fd6adf72850 in malloc (/lib64/libasan.so.4+0xde850)
*1 0x7fd6abb5f689 in strndup (/lib64/libc.so.6+0x96689)
*2 0x6c2e2f746e617266 (<unknown module>)
SUMMARY: AddressSanitizer: 149 byte(s) leaked in 7 allocation(s).
The NULSTR_FOREACH iterator needs to see an empty string at the end, so
we need to insert an extra NUL explicitly.
Also update PATH0_BIN_SBIN(x) to include an extra NUL terminator, rename
it to PATH_BIN_SBIN_NULSTR(x), which is more consistent with the similar
CONF_PATHS_NULSTR(x) macro.
Fixes: 5008da1ec1
Running `test-path` under an umask such as 027 fails with:
Assertion '(s.st_mode & S_IRWXO) == 0004' failed at ../src/test/test-path.c:247, function test_path_makedirectory_directorymode(). Aborting.
Looking at directory /tmp/test-path_makedirectory, it was indeed created
with mode 0740, applying the umask to the requested 0744.
Set an explicit umask for this test, to ensure reproducible results.
Use `systemctl --user --force exit` to implement the systemd-exit
user service.
This removes our dependence on an external `kill` binary and the
concerns about whether they recognize SIGRTMIN+n by name or what their
interpretation of SIGRTMIN is.
Tested: `systemctl --user start systemd-exit.service` kills the
`systemd --user` instance for my user.
The ninja binary is deployed as `ninja-build` in older distros such as
RHEL 7/CentOS 7. Detect that and use `ninja-build` instead of `ninja`
when it's available.
Meson keeps permissions around during the build, so details of how umask
was set when cloning the original git tree will leak all the way to the
installed files in the mkosi image.
So reset the permissions of the files in the copy of the tree before
starting the build.
Also set the umask explicitly.
Tested by creating a mkosi image and booting it on a tree that was
cloned with a umask of 027, confirmed that the *.target files were not
created as world-unreadable anymore.
Also fix one case where the presence of a newline was used to generate
an invalid environment assignment.
Tested: with mkosi, which builds the local tree and run ninja tests.
This was a bug inadvertently added by commit 73fc96c8ac.
The intent of the check is to "match slot address with device by
stripping the function" (as the comment above states it), for example
match network device PCI address 0000:05:00.0 (including a .0 for
function) to PCI slot address 0000:05:00, but changing that to a streq()
call prevented the match.
Change that to startswith(), which should both fix the bug and make the
intent of the check more clear and prevent unintentional bugs from being
introduced by future refactorings.
Otherwise having a .socket unit start a .service running a binary under
a chroot fails as the unit is unable to determine the SELinux label of
the binary.
Using C.UTF-8 (as was done before #7244) breaks Arch Linux, but using
en_US.UTF-8 (after #7244) breaks Debian in our .mkosi/mkosi.debian.
So try to detect which one is available and works, first checking
whether we're already running under a valid UTF-8 locale, then trying
C.UTF-8 and finally en_US.UTF-8.
If we fail to find a valid UTF-8 locale, then fail early, instead of
letting the whole build complete only for Mesos to fail midway through
the `ninja test` step.
Tested on all of mkosi.fedora, mkosi.debian and mkosi.arch.
Fixes: #7238
A runaway string should still be returned by the code that splits on
commas, so add a '?' to the regex so that the last '"?' in a string
still produces a valid block for the split code.
Tested:
ACTION=="remove\"GOTO=""
Which then produced:
$ test/rule-syntax-check.py src/login/70-uaccess.rules
# looking at src/login/70-uaccess.rules
Invalid line src/login/70-uaccess.rules:10: ACTION=="remove\"GOTO=""
clause: ACTION=="remove\"GOTO=""
Using a regex to match the groups is smarter than the split(',') that
would break in those cases.
Tested:
SUBSYSTEM=="usb", ENV{ID_USB_INTERFACES}=="*:060101:*,*:070202:*", TAG+="uaccess"
Rule checker doesn't break there after this commit.
Add support to backslash-escaped double quote inside a string.
Tested by modifying src/login/70-uaccess.rules to include:
ACTION=="remove" it", GOTO="uaccess_end"
And had the rule checker complain about it:
$ test/rule-syntax-check.py src/login/70-uaccess.rules
# looking at src/login/70-uaccess.rules
Invalid line src/login/70-uaccess.rules:10: ACTION=="remove" it", GOTO="uaccess_end"
clause: ACTION=="remove" it"
If ./configure --disable-resolved has been used, do not try to build
test-dns-packet and test-resolve-tables which depend on it.
Previously, the SOURCES, LIBS and LDADDs for these tests were made conditional
while the main rules for them weren't, causing build failures trying to build a
binary with no sources.
This was uncovered when trying to build udeb for systemd in CI, which uses
--disable-resolved for a minimal build, which uncovered the issue.
Fixes#2651.
@evverx brought up that test-execute runs under MANAGER_USER which
forwards all its environment variables to the services. It turns out it
only forwards those that were in the environment at the time of manager
creation, so this test was still working.
It was still possible to attack it by running something like:
$ sudo VAR1=a VAR2=b VAR3=c ./test-execute
Prevent that attack by unsetting the three variables explicitly before
creating the manager for the test case.
Also add comments explaining the interactions with MANAGER_USER and,
while it has some caveats, this tests are still valid in that context.
Tested by checking that the test running with the variables set from the
external environment will still pass.
Check the base case, plus erasing the list, listing the same variable
name more than once and when variables are absent from the manager
execution environment.
Confirmed that `sudo ./test-execute` passes and that modifying the test
cases (or the values of the set variables in test-execute.c) is enough
to make the test cases fail.
This directive allows passing environment variables from the system
manager to spawned services. Variables in the system manager can be set
inside a container by passing `--set-env=...` options to systemd-spawn.
Tested with an on-disk test.service unit. Tested using multiple variable
names on a single line, with an empty setting to clear the current list
of variables, with non-existing variables.
Tested using `systemd-run -p PassEnvironment=VARNAME` to confirm it
works with transient units.
Confirmed that `systemctl show` will display the PassEnvironment
settings.
Checked that man pages are generated correctly.
No regressions in `make check`.
In most cases, systemd requires escaping $ (for systemd variable
substitution) and % (for specifiers) by doubling them. This was somewhat
of an issue in tests like exec-environment*.service where systemd was
doing the substitutions and we were not really checking that those were
available in the actual environment of the command. Fix that.
Expressions such as `exit $(test ...)` are incorrect. They only work
because $(test ...) will produce no output, so the command will become a
bare "exit" which will exit with the status of the latest executed
command which turns out to be the test... The direct approach is simply
calling "test" as the last command, for which the shell will propagate
the exit status.
One situation where this was breaking tests was on `exit $(test ...) &&
$(test ...) && $(test ...)` where the second and third tests were not
really executing, since the first command is actually `exit` so && was
doing nothing there. Fixed it by just using `test ... && test ... &&
test ...` as it was initially intended.
Pass -x to all shell executions for them to produce useful debugging
output to stderr. Consequently, removed most of the explicit `echo`s
that are no longer needed.
Mark all units as Type=oneshot explicitly.
Also made sure all shell variables are properly quoted.
v2: Added an explicit LC_ALL=C to ionice invocations since some locales
(such as French) will add a space before the colon in the output.
Tested by running `sudo ./test-execute` and confirming all tests enabled
on my system (essentially all of them except for the s390 one) passed.
Tweaked the variables or options or expected values and confirmed the
tests do indeed fail when the values are not exactly the expected ones.
v2: Also tested with `LANG=fr_FR.UTF-8 sudo ./test-execute` to confirm
it still works in a different locale.
Just skip them in place, instead of setting separator=true. We only do
that in a single place (while finding a separator outside of quote or
backslash states) so we don't really need a separate state for it.
Tested that no regressions were introduced in test-extract-word. Ran a
full `make check` and also installed the binaries on a test system and
did not see any issues related to parsing unit files or starting units
after a reboot.
Use inner loops to keep processing the same state, except when there is
a state change, then break back to the outer loop so that the correct
branch can be selected again.
Tested that no regressions were introduced in test-extract-word.
This will make it easier to use inner loops to keep looping in the same
state, by just updating p and c in the same way in the inner loops.
Tested that no regressions were created in test-extract-word.
Using `goto` might be appropriate for the "finish" cases but it was
really not necessary at this point of the code... Just use if/else
blocks to accomplish the same.
Confirmed that the test cases in test-extract-word keep working as
expected.
This block runs once before all the other handling, so move it outside
the main loop and put it in its own loop until it's finished doing its
job.
Tested by confirming `make check` (and particularly test-extract-word)
still passes and by booting a system with binaries including this
commit.
Tests for the functions defined in src/basic/parse-util.c. Reorder them
to match the order in which the functions are defined in the source
file. Adjusted the list of include files to remove the ones no longer
needed in test-util.c.
Tested that `make check` still passes as expected. Also checked the
number of lines removed from test-util.c matches the expected, as an
additional verification that no tests were dropped or duplicated in the
move.
Document support for commas as a separator and possibility of specifying
ranges of CPU indices.
Tested by regenerating the manpages locally and reading them on man.
Tested CPUAffinity ranges on both a service unit and in system.conf and
confirmed they work as expected (by inspecting /proc/PID/status, for the
main pid of the service and for pid 1). Also mixed ranges with both
spaces, commas, trailing commas and spaces.
Added new tests to increase coverage of ranges and prevent regressions.
Tested CPUAffinity settings on both a service unit and in system.conf
and confirmed they work as expected.
Added a new test to confirm that trailing commas and spaces work and to
prevent any regressions in that area.
This simplifies the logic and uniformizes the way single and double
quotes are handled. In the end, the code is about 40 lines shorter.
Tested by running the excellent test cases from test-util. Also
installed the systemd binaries including this patch and booted a
system with it, everything looked normal.
Tested with a dummy service running 'sleep', modifying its CPUAffinity,
restarting the service and checking the ^Cpus_allowed entries in the
/proc/PID/status file.
Use the new code in config_parse_cpu_affinity2.
Tested by modifying CPUAffinity=... setting in /etc/systemd/system.conf
and reloading the daemon, then checking ^Cpus_allowed in /proc/1/status
to confirm the correct CPU mask is in place.
gcc versions 4.6 and earlier used to complain when a local variable
shadows a global function, 4.7 and above only complain if a local
variable shadows a global variable.
Fix this by checking whether gcc 4.7+ behavior is in place before
deciding to use -Werror=shadow in $(CFLAGS), by using a custom test
program source that shadows a global function with a local variable and
confirming that -Werror=shadow does not make the compile to break.
Tested:
- On gcc 4.7 and 4.8, confirmed nothing changed (other than the order of
the -Werror=shadow argument, going to the end of CFLAGS.)
- On gcc 4.6, confirmed by looking at the config.log output that the
check for -Werror=shadow failed and it was not included in CFLAGS.
- Ran `make V=1` to confirm -Werror=shadow was still in use, introduced
a bogus shadowing issue and confirmed it was caught when building with
a recent gcc.
The variables should be quoted inside [...] to avoid double macro
expansion. This is currently not an issue, since the values (-W...) are
not really macros, but we might as well just fix that issue now.
Tested by re-running autogen.sh and comparing the value of OUR_CFLAGS in
the generated Makefile. Ran a full build from a clean tree to confirm
no other issues were introduced.
