Yes, there are still a lot of users of bzip2, but it's fallen out of
favour after LZMA/xz, which can compress a lot more and often
decompresses faster than bzip2 too.
We use strtoul() which returns an "unsigned long", but then assign this
to int or unsigned in, i.e. drop 32bit silently on 64bit systems. Let's
clean this up a bit, and retain the right types.
This stuff is likely to fail in many setups (for example when quota is
not supported by the btrfs version), hence only log at debug
level. Previously we'd silently ignore things altogether which makes
things pretty hard to debug.
This changes the output a bit, as the previous multi-line output of each
inhibitor is changed to a single line, but it does unify the output look
with the one of our other tools. Moreover this adds proper sorting.
When we parse an "u" from an sd_bus_message then we need to do that into
a uint32_t, not a pid_t or uid_t, even if this is likely the same.
Also, let's count objects we keep in memory as size_t as usual.
In seccomp code, the code is changed to propagate errors which are about
anything other than unknown/unimplemented syscalls. I *think* such errors
should not happen in normal usage, but so far we would summarilly ignore all
errors, so that part is uncertain. If it turns out that other errors occur and
should be ignored, this should be added later.
In nspawn, we would count the number of added filters, but didn't use this for
anything. Drop that part.
The comments suggested that seccomp_add_syscall_filter_item() returned negative
if the syscall is unknown, but this wasn't true: it returns 0.
The error at this point can only be if the syscall was known but couldn't be
added. If the error comes from our internal whitelist in nspawn, treat this as
error, because it means that our internal table is wrong. If the error comes
from user arguments, warn and ignore. (If some syscall is not known at current
architecture, it is still silently ignored.)
Our logs are full of:
Sep 19 09:22:10 autopkgtest systemd[690]: Failed to add rule for system call oldstat() / -10037, ignoring: Numerical argument out of domain
Sep 19 09:22:10 autopkgtest systemd[690]: Failed to add rule for system call get_thread_area() / -10076, ignoring: Numerical argument out of domain
Sep 19 09:22:10 autopkgtest systemd[690]: Failed to add rule for system call set_thread_area() / -10079, ignoring: Numerical argument out of domain
Sep 19 09:22:10 autopkgtest systemd[690]: Failed to add rule for system call oldfstat() / -10034, ignoring: Numerical argument out of domain
Sep 19 09:22:10 autopkgtest systemd[690]: Failed to add rule for system call oldolduname() / -10036, ignoring: Numerical argument out of domain
Sep 19 09:22:10 autopkgtest systemd[690]: Failed to add rule for system call oldlstat() / -10035, ignoring: Numerical argument out of domain
Sep 19 09:22:10 autopkgtest systemd[690]: Failed to add rule for system call waitpid() / -10073, ignoring: Numerical argument out of domain
...
This is pointless and makes debug logs hard to read. Let's keep the logs
in test code, but disable it in nspawn and pid1. This is done through a function
parameter because those functions operate recursively and it's not possible to
make the caller to log meaningfully.
There should be no functional change, except the skipped debug logs.
This effectively reverts 2bc54be485
and relevant changes in #9920, as it is used to determine the version
of udev, e.g., dracut.
Fixesdracutdevs/dracut#468.
In order to shut down networkd properly, the delegated routes added
need to be removed properly, and as error reporting is wanted, the
network link is needed in the debug output.
Solve this by calling manager_dhcp6_prefix_remove_all(), which will
remove each prefix stored in the Manager structure, and while doing
that reference each link so that it isn't freed before the route
removal callback is called. This in turn causes the network link to
be referenced once more, and an explicit hashmap_remove() must be
called to remove the network link from the m->links hashmap.
Also, since the registered callback is not called when the DHCPv6
client is stopped with sd_dhcp6_client_stop(), an explicit call
to dhcp6_lease_pd_prefix_lost() needs to be made to clean up any
unreachable routes set up for the delegated prefixes.
Instead of setting many small unreachable routes for each of the /64
subnets that were not distributed between the links requesting delegated
prefixes, set one unreachable route for the size of the delegated
prefix. Each subnet asssigned to a downstream link will add a routable
subnet for that link, and as the subnet assigned to the downstream link
has a longer prefix than the whole delegated prefix, the downstream
link subnet routes are preferred over the unroutable delegated one.
The unreachable route is not added when the delegated prefix is exactly
a /64 as the prefix size cannot be used to sort out the order of routing
into a bigger blocking subnet with the smaller /64 punching routable
"holes" into it.
When stopping the DHCPv6 client, the unroutable delegated prefix is
removed before the downstream link prefixes are unassigned.
Since we now have the possibility to request prefixes to be delegated
without corresponding IPv6 addresses, it does not make sense to store
lease T1 and T2 timeouts in the otherwise unused IA_NA structure.
Therefore lease timeouts T1 and T2 are moved to the DHCPv6 client
structure, as there will be only one set of stateful timeouts required
by RFC 7550, Section 4.3.
Select T1 and T2 timeouts based on whether addresses or prefixes were
requested and what the server offered. The address and prefix timeouts
values have been computed earlier when the relevant DHCPv6 options were
parsed.
Add function to fetch the IAID for the delegated IA_PD prefix. In
order to keep things simple in the implemntation, the same IAID
is used with IA_NA addresses and IA_PD prefixes. But the DHCPv6
server can choose to return only IA_PD prefixes, and the client
can nowadays omit requesting of IA_NA addresses. Since the function
fetching said IAID from the lease looks only for IA_NA ones, it
will return an empty IAID, which of course does not match the one
set for prefixes.
Fix this by adding a function returning the IAID for the prefix.
RFC 7084, WPD-4, requires Customer Edge end routers to behave
according to the following:
"WPD-4: By default, the IPv6 CE router MUST initiate DHCPv6 prefix
delegation when either the M or O flags are set to 1 in a
received Router Advertisement (RA) message. Behavior of the
CE router to use DHCPv6 prefix delegation when the CE router
has not received any RA or received an RA with the M and the
O bits set to zero is out of scope for this document."
Since it cannot be automatically detected whether DHCPv6 is to be
operated as an CE end router or whether to initiate an Informational
exchange to obtain other useful network information via DHCPv6 when the
Router Advertisement 'O' bit is set, a 'ForceDHCPv6PDOtherInformation'
boolean network configuration option in the '[DHCP]' section of a is
introduced. Setting this option causes DHCPv6 to be started in stateful
mode, although only the 'O' bit is seen in the Router Advertisement.
When 'ForceDHCPv6PDOtherInformation' is set and the Router Advertisement
has only the Other information 'O' bit set, disable requests for IA_NA
addresses.
Fixes#9745.
Add function to enable/disable IA_NA address requests. Internally
handle the request as a bit mask and add IA_PD prefix delegation
to the same bit mask instead of having a separate boolean. Thus
the calling code can set requests for prefix and address delegation
separately. This is handy when supporting RFC 7084.
Add a check in the code that at least something is requested from
the server in Managed mode. By default request IA_NA addresses from
the DHCPv6 server. Although a value has been defined for IA_TA,
temporay IA_TA addresses are not yet requested.
Add helper function for fetching enabled/disabled state of Prefix
Delegation for a DHCPv6 client. Update function setting prefix
delegation to use an int instead of a boolean.
Both 'systemd-hwdb update' and 'udevadm hwdb --update' creates hwdb
database. The database created by systemd-hwdb containes additional
information such that priority, line number, and source filename.
The unified function 'hwdb_update()' can take a flag 'compat' which
controls the format version of created database.
Several functions are shared by qsort and hash_ops or Prioq.
This makes these functions explicitly specify argument type,
and cast to __compar_fn_t where necessary.
In older kernels IPoIB network devices expose the port number via
the sysfs attribute 'dev_id', which is not intended to be used this way.
Let's support both options for a while.
An InfiniBand network address is 20 bytes long. Only the least
significant 8 bytes can be interpreted as a persistent hardware unit
identifier; the other 12 are transiently derived at runtime from metadata
specific to the protocol stack.
However, since the network interface name length is hard-capped by
IFNAMSIZ at 16 chars and the 2-byte type prefix with '\0' at the end
leave us only at 13, we cannot squeeze a descriptive representation of a
HW address into an interface name. Thus, it makes the most sense to drop
the scheme for IPoIB interfaces entirely.
Currently udev just gets confused and does what it has been taught
to do: fetches the first six bytes and puts them into a permanent
device attribute.
We've long neglected IP-over-InfiniBand network interfaces, let's treat
them the same way we treat anyone else.
IPoIB interfaces will retain the 'ib' prefix; otherwise the naming scheme
is the same one we use for other network interfaces. E.g. a IPoIB network
device provided by a PCI card at bus 21 slot 0 function 6 will be named
'ibp21s0f6'.
Quoting https://github.com/systemd/systemd/issues/10074:
> detect_vm_uml() reads /proc/cpuinfo with read_full_file()
> read_full_file() has a file max limit size of READ_FULL_BYTES_MAX=(4U*1024U*1024U)
> Unfortunately, the size of my /proc/cpuinfo is bigger, approximately:
> echo $(( 4* $(cat /proc/cpuinfo | wc -c)))
> 9918072
> This causes read_full_file() to fail and the Condition test fallout.
Let's just read line by line until we find an intersting line. This also
helps if not running under UML, because we avoid reading as much data.
optind may be used in each verb, e.g., udevadm. So, let's initialize
optind before calling verbs.
Without this, e.g., udevadm -d hwdb --update causes error in parsing arguments.
Since the commit torvalds/linux@fdb5c4531c
the syscall BPF_PROG_ATTACH return EBADF when CONFIG_CGROUP_BPF is
turned off and as result the bpf_firewall_supported() returns the
incorrect value.
This commmit replaces the syscall BPF_PROG_ATTACH with BPF_PROG_DETACH
which is still work as expected.
Resolvesopenbmc/linux#159
See also systemd/systemd#7054
Signed-off-by: Alexander Filippov <a.filippov@yadro.com>
Without this, log shows meaningless error message 'No anode', e.g.,
===
Failed to unshare the mount namespace: Operation not permitted
foo.service: Failed to set up mount namespacing: No anode
foo.service: Failed at step NAMESPACE spawning /usr/bin/test: No anode
===
Follow-up for 1beab8b0d0.
Both SO_SNDBUFFORCE and SO_RCVBUFFORCE requires capability 'net_admin'.
If this capability is not granted to the service the first attempt to increase
the recv/snd buffers (via sd_notify()) with SO_RCVBUFFORCE/SO_SNDBUFFORCE will
fail, even if the requested size is lower than the limit enforced by the
kernel.
If apparmor is used, the DENIED logs for net_admin will show up. These log
entries are seen as red warning light, because they could indicate that a
program has been hacked and tries to compromise the system.
It would be nicer if they can be avoided without giving services (relying on
sd_notify) net_admin capability or dropping DENIED logs for all such services
via their apparmor profile.
I'm not sure if sd_notify really needs to forcibly increase the buffer sizes,
but at least if the requested size is below the kernel limit, the capability
(hence the log entries) should be avoided.
Hence let's first ask politely for increasing the buffers and only if it fails
then ignore the kernel limit if we have sufficient privileges.
When $XDG_DATA_DIRS is unset, then, the default value
'/usr/local/share:/usr/share' is used.
When $XDG_DATA_DIRS contain the default paths but the order
is inverted: '/usr/share:/usr/local/share', then test-path-lookup fails.
Fixes#10002.
Back in 08318a2c5a, value "false" was enabled for
'-Dtests=', but various tests were not conditionalized properly. So even with
-Dtests=false -Dslow-tests=false we'd run 120 tests. Let's make this consistent.
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.
Actually check the return code from logind_schedule_shutdown() and proceed to
immediate shutdown if that fails. Negative return codes can be returned if
systemctl is compiled without logind support, or if logind otherwise failed
(either too old, disabled/masked, or it is incomplete
systemd-shim/systemd-service implementation).
An assertion in dhcp_network_bind_raw_socket() is triggered when
starting an sd_dhcp_client without setting a MAC address first.
- sd_dhcp_client_start()
- client_start()
- client_start_delayed()
- dhcp_network_bind_raw_socket()
In that case, the arp-type and MAC address is still unset. Note that
dhcp_network_bind_raw_socket() already checks for a valid arp-type
and MAC address below, so we should just gracefully return -EINVAL.
Maybe sd_dhcp_client_start() should fail earlier when starting without
MAC address. But the failure here will be correctly propagated and
the start aborted.
Fixes: 76253e73f9
When a network namespace is needed, /sys is mounted as tmpfs (see commit
d8fc6a000f for details).
But in this case mode 755 was used as initial permissions for /sys whereas the
default mode for sysfs is 555.
In practice using 755 doesn't have any impact because /sys is mounted read-only
too but for consistency, let's use the correct mode.
Fixes: #10050