The check added by 4c27749b8c breaks
booting an arm64 image on x86 using qemu-bin-fmt, so remove it.
Without it, the image built with mkosi --architecture=aarch64
boots fine in nspawn.
Interface=/MACVLAN=/IPVLAN= nspawn options take a _list_ of interface
names - this was recently enhanced by 2f091b1b49 to support interface
pairs. Unfortunately, this also introduced a regression where we don't
parse the list as a list, but just as a single value. For example,
having `Interface=sd-shared1 sd-shared2` in an nspawn config file would
throw:
systemd-nspawn[898]: Network interface, interface name not valid: sd-shared1 sd-shared2
systemd-nspawn[898]: /run/systemd/nspawn/testsuite-13.nspawn-settings.1po.nspawn:41: Failed to parse file: Invalid argument
Follow-up to 2f091b1b49.
Otherwise hilarity ensues:
AddressSanitizer:DEADLYSIGNAL
=================================================================
==722==ERROR: AddressSanitizer: SEGV on unknown address 0xffffffff00000000 (pc 0x7f8d50ca9ffb bp 0x7fff11b0d4a0 sp 0x7fff11b0cc30 T0)
==722==The signal is caused by a READ memory access.
#0 0x7f8d50ca9ffb in __interceptor_strcmp.part.0 (/lib64/libasan.so.8+0xa9ffb)
#1 0x7f8d4f9cf5a1 in strcmp_ptr ../src/fundamental/string-util-fundamental.h:33
#2 0x7f8d4f9cf5f8 in streq_ptr ../src/fundamental/string-util-fundamental.h:46
#3 0x7f8d4f9d74d2 in free_and_strdup ../src/basic/string-util.c:948
#4 0x49139a in free_and_strdup_warn ../src/basic/string-util.h:197
#5 0x4923eb in oci_absolute_path ../src/nspawn/nspawn-oci.c:139
#6 0x7f8d4f6bd359 in json_dispatch ../src/shared/json.c:4395
#7 0x4a8831 in oci_hooks_array ../src/nspawn/nspawn-oci.c:2089
#8 0x7f8d4f6bd359 in json_dispatch ../src/shared/json.c:4395
#9 0x4a8b56 in oci_hooks ../src/nspawn/nspawn-oci.c:2112
#10 0x7f8d4f6bd359 in json_dispatch ../src/shared/json.c:4395
#11 0x4aa298 in oci_load ../src/nspawn/nspawn-oci.c:2197
#12 0x446cec in load_oci_bundle ../src/nspawn/nspawn.c:4744
#13 0x44ffa7 in run ../src/nspawn/nspawn.c:5477
#14 0x4552fb in main ../src/nspawn/nspawn.c:5920
#15 0x7f8d4e04a50f in __libc_start_call_main (/lib64/libc.so.6+0x2750f)
#16 0x7f8d4e04a5c8 in __libc_start_main@GLIBC_2.2.5 (/lib64/libc.so.6+0x275c8)
#17 0x40d284 in _start (/usr/bin/systemd-nspawn+0x40d284)
AddressSanitizer can not provide additional info.
SUMMARY: AddressSanitizer: SEGV (/lib64/libasan.so.8+0xa9ffb) in __interceptor_strcmp.part.0
==722==ABORTING
Use the returned errno even though we are going to ignore it, otherwise
the log message is just confusing:
config.json:119:13: Failed to resolve device node 4:2, ignoring: Success
When merging the settings we take the pointer to the array of extra
devices, but don't reset the array counter to zero. This later leads to
a NULL pointer dereference, where device_node_array_free() attempts to
loop over a NULL pointer:
+ systemd-nspawn --oci-bundle=/var/lib/machines/testsuite-13.oci-bundle.Npo
../src/nspawn/nspawn-settings.c:118:29: runtime error: member access within null pointer of type 'struct DeviceNode'
#0 0x4b91ee in device_node_array_free ../src/nspawn/nspawn-settings.c:118
#1 0x4ba42a in settings_free ../src/nspawn/nspawn-settings.c:161
#2 0x410b79 in settings_freep ../src/nspawn/nspawn-settings.h:249
#3 0x446ce8 in load_oci_bundle ../src/nspawn/nspawn.c:4733
#4 0x44ff42 in run ../src/nspawn/nspawn.c:5476
#5 0x455296 in main ../src/nspawn/nspawn.c:5919
#6 0x7f0cb7a4a50f in __libc_start_call_main (/lib64/libc.so.6+0x2750f)
#7 0x7f0cb7a4a5c8 in __libc_start_main@GLIBC_2.2.5 (/lib64/libc.so.6+0x275c8)
#8 0x40d284 in _start (/usr/bin/systemd-nspawn+0x40d284)
SUMMARY: UndefinedBehaviorSanitizer: undefined-behavior ../src/nspawn/nspawn-settings.c:118:29 in
Also, add an appropriate assert to catch such issues in the future.
Let's make use of the new DelegateSubgroup= feature and delegate the
/supervisor/ subcgroup already to nspawn, so that moving the supervisor
process there is unnecessary.
When we're checking if /etc/resolv.conf exists so we can bind mount
on top of it, we care about whether the symlink itself exists if
/etc/resolv.conf exists and not the file it points to, so add
CHASE_NOFOLLOW to make sure we check existence of the symlink and
not the file it points to.
systemd-nspawn now optionally supports colon-separated pair of
host interface name and container interface name for --network-macvlan, --network-ipvlan and --network-interface options.
Also supported in .nspawn configuration files (i.e Interface=, MACVLAN=, IPVLAN= parameters).
man page changed for ntwk interface naming
Previously, when the NULL (all zero) machine ID is configured in the
container, nspawn refused to execute.
Now id128_get_machine() is used, so NULL machine ID is refused with
-ENOMEDIUM, and fallback to specified UUID or randomly generated one.
Otherwise, if getopt() and friends are used before parse_argv(), then
the GNU extensions may be ignored.
This should not change any behavior at least now, as we usually use
getopt_long() only once per invocation. But in the next commit,
getopt_long() will be used for other arrays, hence this change will
become necessary.
Chasing symlinks is a core function that's used in a lot of places
so it deservers a less verbose names so let's rename it to chase()
and chaseat().
We also slightly change the pattern used for the chaseat() helpers
so we get chase_and_openat() and similar.
Whenever we're going to close all file descriptors, we tend to close
the log and set it into open when needed mode. When this is done with
the logging target set to LOG_TARGET_AUTO, we run into issues because
for every logging call, we'll check if stderr is connected to the
journal to determine where to send the logging message. This check
obviously stops working when we close stderr, so we settle the log
target before we do that so that we keep using the same logging
target even after stderr is closed.
We bind mount two selected inodes from the host into our container.
Let's turn off propagation for that, since we just want those inodes,
nothing else.
With this change "grep master: /proc/self/mountinfo" should list only
the mount propagation "tunnel" dir, and nothing else anymore.
@brauner noticed that in invoked containers the root directory is set to
still receive mounts from the host. We should disable that, and
guarantee we live in our own world, because that's what an
(nspawn-style) container *is* after all: a whole new world.
This hence mounts the container subtree to MS_PRIVATE after getting the
root dir in place. Note that this will later be set to MS_SHARED again.
The MS_PRIVATE disconnects mounts from the host, the MS_SHARED then
establishes a new peer group for mount propagation events, so that
payload service managers (such as systemd) can take benefit of
propagation further down the tree.
Zbigniew Jędrzejewski-Szmek
Lennart Poettering
Daan De Meyer
Luca Boccassi
Yu Watanabe
David Tardon
Frantisek Sumsal
Thierry Martin
Jan Janssen