fuzz-unit-file: add __has_feature(memory_sanitizer) when skipping ListenNetlink=
https://clang.llvm.org/docs/MemorySanitizer.html#id5 documents this
check as the way to detect MemorySanitizer at compilation time. We
only need to skip the test if MemorySanitizer is used.
Also, use this condition in cg_slice_to_path(). There, the code that is
conditionalized is not harmful in any way (it's just unnecessary), so remove
the FUZZING_BUILD_MODE_UNSAFE_FOR_PRODUCTION condition.
fuzz-unit-file: adjust check for ListenNetlink yet again
The test for ListenNetlink would abort the loop if a line longer then LINE_MAX
was encountered (read_line() returns -ENOBUFS in that case). Let's use the
the line length limit that the unit file parses uses.
Unfortunately the MIPS toolchains still do not implement PT_GNU_STACK.
This means that while the commit to restrict mmap on MIPS was "correct",
it had the side effect of causing pthread_create to fail because glibc tries
to allocate an executable stack for new threads in the absense of
PT_GNU_STACK. We should wait until PT_GNU_STACK is implemented in all
the relevant parts of the toolchain (at least gcc and glibc) before
enabling this again.
test: bypass selinux integration test if selinux policy devel package is not installed
With this "sudo ./run-integration-tests.sh" should work fully without
exception, even on systems lacking SELinux (in which case that test will
just be skipped)
Michal Sekletar [Fri, 23 Mar 2018 14:28:06 +0000 (15:28 +0100)]
core: delay adding target dependencies until all units are loaded and aliases resolved (#8381)
Currently we add target dependencies while we are loading units. This
can create ordering loops even if configuration doesn't contain any
loop. Take for example following configuration,
If we encounter such unit file early during manager start-up (e.g. load
queue is dispatched while enumerating devices due to SYSTEMD_WANTS in
udev rules) we would add stub unit default.target and we order it Before
test.service. At the same time we add implicit Before to
multi-user.target. Later we merge two units and we create ordering cycle
in the process.
To fix the issue we will now never add any target dependencies until we
loaded all the unit files and resolved all the aliases.
Peter Hutterer [Fri, 23 Mar 2018 14:15:41 +0000 (00:15 +1000)]
udev: don't label high-button mice as joysticks (#8493)
If a device exposes more than 16 mouse buttons, we run into the BTN_JOYSTICK
range, also labelling it as joystick. And since 774ff9b this results in only
ID_INPUT_JOYSTICK but no ID_INPUT_MOUSE.
tree-wide: warn when a directory path already exists but has bad mode/owner/type
When we are attempting to create directory somewhere in the bowels of /var/lib
and get an error that it already exists, it can be quite hard to diagnose what
is wrong (especially for a user who is not aware that the directory must have
the specified owner, and permissions not looser than what was requested). Let's
print a warning in most cases. A warning is appropriate, because such state is
usually a sign of borked installation and needs to be resolved by the adminstrator.
$ build/test-fs-util
Path "/tmp/test-readlink_and_make_absolute" already exists and is not a directory, refusing.
(or)
Directory "/tmp/test-readlink_and_make_absolute" already exists, but has mode 0775 that is too permissive (0755 was requested), refusing.
(or)
Directory "/tmp/test-readlink_and_make_absolute" already exists, but is owned by 1001:1000 (1000:1000 was requested), refusing.
Assertion 'mkdir_safe(tempdir, 0755, getuid(), getgid(), MKDIR_WARN_MODE) >= 0' failed at ../src/test/test-fs-util.c:320, function test_readlink_and_make_absolute(). Aborting.
No functional change except for the new log lines.
This macro will read a pointer of any type, return it, and set the
pointer to NULL. This is useful as an explicit concept of passing
ownership of a memory area between pointers.
It drops ~160 lines of code from our codebase, which makes me like it.
Also, I think it clarifies passing of ownership, and thus helps
readability a bit (at least for the initiated who know the new macro)
test-execute: allow sit0@ to exist in private network namespace
It's always visible:
$ sudo modprobe sit
$ sudo unshare -n ip l
1: lo: <LOOPBACK> mtu 65536 qdisc noop state DOWN mode DEFAULT group default qlen 1000
...
2: sit0@NONE: <NOARP> mtu 1480 qdisc noop state DOWN mode DEFAULT group default qlen 1000
...
test-execute: simplify checks if grep output is empty
grep already indicates if it matched anything by return value.
Additional advantage is then that if the test fails, the unexpected
matching lines are visible in the log output.
safe_atou16_full() is like safe_atou16() but also takes a base
parameter. safe_atou16() is then implemented as inline function on top
of it, passing 0 as base. Similar safe_atoux16() is reworked as inline
function too, with 16 as base.
Let's use first_word() instead of startswith(), it's more explanatory
and a bit more correct. Also, let's use the return value instead of
adding +9 when looking for the second part of the directive.
conf-parser: let's explicitly deprecate .include in unit files
.include lines are already deprecated somewhat, and for example
explicitly not mentioned in the documentation for this reason. Let's get
one step further and generatea warning when we encounter them (but still
process them).
Why are they deprecated? Because they are semantically awful — they
complicate stat() based mtime checks for configuration files and they
allow arbitrary loops we currently have zero protection against and
really shouldn't have to have.
When mangling names, optionally emit a warning (#8400)
The warning is not emitted for absolute paths like /dev/sda or /home, which are
converted to .device and .mount unit names without any fuss.
Most of the time it's unlikely that users use invalid unit names on purpose,
so let's warn them. Warnings are silenced when --quiet is used.
$ build/systemctl show -p Id hello@foo-bar/baz
Invalid unit name "hello@foo-bar/baz" was escaped as "hello@foo-bar-baz" (maybe you should use systemd-escape?)
Id=hello@foo-bar-baz.service
$ build/systemd-run --user --slice foo-bar/baz --unit foo-bar/foo true
Invalid unit name "foo-bar/foo" was escaped as "foo-bar-foo" (maybe you should use systemd-escape?)
Invalid unit name "foo-bar/baz" was escaped as "foo-bar-baz" (maybe you should use systemd-escape?)
Running as unit: foo-bar-foo.service
move MANAGER_IS_RELOADING() check into manager_recheck_{dbus|journal}() (#8510)
Let's better check this inside of the call than before it, so that we
never issue this while reloading, even should these calls be called due
to other reasons than just the unit notify.
This makes sure the reload state is unset a bit earlier in
manager_reload() so that we can safely call this function from there and
they do the right thing.
Long Li [Wed, 21 Mar 2018 10:51:28 +0000 (03:51 -0700)]
v3: Properly parsing SCSI Hyperv devices (#8509)
Since 2016, Hyperv devices moved to using standard way to expose UUID to sysfs. Fix the parsing function to work with the newer format.
Change log:
v2: changed code to work with both old and new path format
v3: changed guid_str_len type to size_t, fixed length in char guid[] in handle_scsi_hyperv()
Yu Watanabe [Wed, 21 Mar 2018 04:39:03 +0000 (13:39 +0900)]
nss-systemd: make dynamic users enumerable by `getent`
This adds `setpwent()`, `getpwent_r()`, `endpwent()`, `setgrent()`,
`getgrent_r()`, and `endgrent()` interfaces to nss-systemd library.
Thus, dynamic users can be enumerated by e.g. `getent passwd` command.
core/load-fragment: reject overly long paths early
No need to go through the specifier_printf() if the path is already too long in
the unexpanded form (since specifiers increase the length of the string in all
practical cases).
In the oss-fuzz test case, valgrind reports:
total heap usage: 179,044 allocs, 179,044 frees, 72,687,755,703 bytes allocated
and the original config file is ~500kb. This isn't really a security issue,
since the config file has to be trusted any way, but just a matter of
preventing accidental resource exhaustion.
journal: don't insist that the journal file header's boot ID matches the last entry
We update the boot ID whenever the file is opened for writing (i.e. set
to ONLINE stat), even if we never write a single entry to it. Hence,
don't insist that the last entry's boot ID matches the file header.
Alex Gartrell [Tue, 27 Feb 2018 05:56:35 +0000 (21:56 -0800)]
journal: provide compress_threshold_bytes parameter
Previously the compression threshold was hardcoded to 512, which meant that
smaller values wouldn't be compressed. This left some storage savings on the
table, so instead, we make that number tunable.
basic/fs-util: skip fsync_directory_of_file() if /proc/self/fd/ is not available (#8386)
When systemd is running under lorax (in Fedora compose process), it'd think that
it failed to write /etc/machine-id, even though the write succeeded, because
fsync_directory_of_file() would fail, because /proc/self/fd/ is not available.
fsync_directory_of_file() is mostly an additional safety net, so I think it's best
to just silently ignore the error.
Strace of pid1:
35791 stat("/etc", {st_mode=S_IFDIR|0755, st_size=4096, ...}) = 0
35791 openat(AT_FDCWD, "/etc/machine-id", O_RDWR|O_CREAT|O_NOCTTY|O_CLOEXEC, 0444) = 3
35791 umask(022) = 000
35791 read(3, "", 38) = 0
35791 openat(AT_FDCWD, "/var/lib/dbus/machine-id", O_RDONLY|O_NOCTTY|O_NOFOLLOW|O_CLOEXEC) = -1 ENOENT (No such file o
r directory)
35791 openat(AT_FDCWD, "/sys/class/dmi/id/product_name", O_RDONLY|O_CLOEXEC) = -1 ENOENT (No such file or directory)
35791 openat(AT_FDCWD, "/sys/class/dmi/id/sys_vendor", O_RDONLY|O_CLOEXEC) = -1 ENOENT (No such file or directory)
35791 openat(AT_FDCWD, "/sys/class/dmi/id/board_vendor", O_RDONLY|O_CLOEXEC) = -1 ENOENT (No such file or directory)
35791 openat(AT_FDCWD, "/sys/class/dmi/id/bios_vendor", O_RDONLY|O_CLOEXEC) = -1 ENOENT (No such file or directory)
35791 access("/proc/xen", F_OK) = -1 ENOENT (No such file or directory)
35791 openat(AT_FDCWD, "/sys/hypervisor/type", O_RDONLY|O_CLOEXEC) = -1 ENOENT (No such file or directory)
35791 openat(AT_FDCWD, "/proc/cpuinfo", O_RDONLY|O_CLOEXEC) = -1 ENOENT (No such file or directory)
35791 getrandom("\xb8\x82\xed\xd4\x35\x11\xd0\xeb\xa6\x79\xd7\x31\x6e\x7b\x99\xce", 16, GRND_NONBLOCK) = 16
35791 writev(2, [{iov_base="Initializing machine ID from random generator.", iov_len=46}, {iov_base="\n", iov_len=1}],
2) = 47
35791 lseek(3, 0, SEEK_SET) = 0
35791 ftruncate(3, 0) = 0
35791 write(3, "b882edd4351140eba679d7316e7b99ce\n", 33) = 33
35791 fsync(3) = 0
35791 fstat(3, {st_mode=S_IFREG|0444, st_size=33, ...}) = 0
35791 readlinkat(AT_FDCWD, "/proc/self/fd/3", 0x564df8c694c0, 99) = -1 ENOENT (No such file or directory)
35791 close(3) = 0
35791 umask(022) = 022
35791 openat(AT_FDCWD, "/run/machine-id", O_WRONLY|O_CREAT|O_NOCTTY|O_TRUNC|O_CLOEXEC, 0444) = 3
35791 write(3, "b882edd4351140eba679d7316e7b99ce\n", 33) = 33
35791 close(3) = 0
35791 umask(022) = 022
35791 mount("/run/machine-id", "/etc/machine-id", NULL, MS_BIND, NULL) = 0
35791 writev(2, [{iov_base="Installed transient /etc/machine-id file.", iov_len=41}, {iov_base="\n", iov_len=1}], 2) = 42
35791 mount(NULL, "/etc/machine-id", NULL, MS_RDONLY|MS_REMOUNT|MS_BIND, NULL) = 0
nspawn: move network namespace creation to a separate step (#8430)
Fixes #8427.
Unsharing the namespace in a separate step changes the ownership of
/proc/net/ip_tables_names (and related files) from nobody:nobody to
root:root. See [1] and [2] for all the details.
mkosi: set file permissions in copy of source tree (#8370)
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.