On Linux, there are two classes of AF_UNIX socket, which D-Bus refers
to as unix:path=... (portable to non-Linux systems) and unix:abstract=...
(not portable).
Back in 2003 when dbus gained support for abstract Unix sockets on Linux,
everyone thought they were better in every way than path-based Unix
sockets: if a DBusServer crashes or is terminated abnormally, there's
no detritus left in the filesystem. What's not to like? As a result,
since commit a70b042f (2003-06-04), when a DBusServer listens on a
unix:tmpdir=... address on Linux, the default is for the result to be
a unix:abstract=... address, with unix:path=... addresses only used on
non-Linux platforms.
However, the world has changed in the last 19 years, and namespace-based
Linux containers (which didn't exist in 2003) are now very popular. This
makes abstract sockets problematic.
Abstract sockets are tied to the network namespace, which is
all-or-nothing: if a container is to access the Internet without using
some sort of proxy or intermediary (like slirp4netns) then it needs to
share the network namespace with the host system, and that implies
sharing all abstract sockets with the host system. If the well-known
session bus is listening on an abstract socket, then it's a sandbox
escape route for any sandboxed or containerized app running under the
same uid. Conversely, if a container is *not* sharing the network
namespace with the host system, then it cannot access a session bus that
is listening on an abstract socket without using some sort of proxy
(like xdg-dbus-proxy), even if it isn't intended to impose a security
boundary and giving it direct access to the session bus would have been
more desirable.
Path-based sockets do not have this problem because they exist in the
filesystem (part of the "everything is a file" Unix philosophy),
allowing mount namespaces and bind-mounts to be used to share or
unshare them selectively.
On systems with `systemd --user` where dbus has been configured with
`--enable-user-session`, in general the session bus will already be
using a path-based socket for the "user bus", disregarding the listening
address specified in /usr/share/dbus-1/session.conf. The default in many
recent Linux distributions is either to use dbus-daemon in this way, or
to use dbus-broker, a reimplementation of the message bus service which
has similar "user bus" behaviour.
However, the <listen> address in session.conf is used when dbus-launch(1)
or dbus-run-session(1) is used to start a session bus, either manually,
via autolaunching, or via system integration glue in operating systems
that are not using `systemd --user`. This will occur particularly often
in operating systems that boot using a non-systemd init system.
Making unix:tmpdir=/tmp equivalent to unix:dir=/tmp ensures that the
well-known session bus listens on a path-based socket, allowing container
and sandboxing frameworks to mediate access to it in the same way they
would for the user bus. The D-Bus Specification already allows (but does
not require) this behaviour, because it is the only thing that was
implementable on non-Linux systems such as *BSD.
This change has the potential to cause regressions. If a container
framework enters a chroot or unshares the mount namespace but does not
unshare the network namespace, and is relying on the ability for a
process inside a container to access the session bus outside the
container via its abstract socket, then that assumption will be broken
by this change. Some use cases of schroot(1) are likely to suffer from
this. However, container frameworks with that assumption would already
have found that it does not hold when using the user bus, and it is
necessary to break that assumption if we want it to be possible to apply
application-level sandboxing in a secure way.
Another potential regression from this change is that if a dbus-daemon
is terminated abnormally, it will leave a socket in /tmp. Distributors
of operating systems where heavy use of dbus-launch(1) is expected might
wish to run dbus-cleanup-sockets(1) periodically.
This partially reverts commit a70b042f.
Resolves: https://gitlab.freedesktop.org/dbus/dbus/-/issues/416
Signed-off-by: Simon McVittie <smcv@collabora.com>
I am building DBus targeting the Arm Morello board and the "no padding"
layout assertion fails here since pointers require 16-byte alignment, and
therefore we have to add two additional ints to the DBusMessageIter struct.
As this is a new architecture, where DBus previously failed to compiled
we do not have any layout backwards compatibility requirements, so we can
simplify the DBusMessageIter structure to allocate space for 16 pointers
(which should give us a lot of space for any further changes).
These static assertions fail on CHERI-enabled architectures such as Arm
Morello, where pointers are 128 bits. Architectures with 128-bit pointers
were not supported in DBus 1.10, so we can skip the checks for DBus 1.10
structure layout compatibility for architectures with pointer size > 64 bit.
When building for Arm Morello (where pointers are 16 bytes), I hit the
static assertion that sizeof (DBusMessageRealIter) <= sizeof (DBusMessageIter)
inside _dbus_message_iter_init_common() otherwise. This can be fixed by
moving the pointers to the beginning of the struct to remove padding.
This is required e.g. for CHERI-enabled targets such as Arm Morello where
aligning to sizeof(long) is not sufficient to load/store pointers (which
need 16 byte alignment instead of 8 bytes).
As we can't depend on C11 yet, this commit adds a max_align_t emulation
to dbus-internals.h.
When targeting CHERI-enabled architectures such as Arm Morello, performing
a bitwise and with uintptr_t values can result in an ambiguous operation
compiler warning. Fix this warning by telling compiler which operand is
(potentially) a pointer and which one is an integer by changing the
boundary type to size_t. This change has no functional effect on other
architectures but is required to build with -Werror for Morello.
Example warning message:
```
warning: binary expression on capability types 'unsigned __intcap' and 'unsigned __intcap'; it is not clear which should be used as the source of provenance; currently provenance is inherited from the left-hand side [-Wcheri-provenance]
_dbus_assert (_DBUS_ALIGN_VALUE (insert_at, 8) == (unsigned) insert_at);
```
FreeBSD 13.0 has reached EOL and it appears packages
are built against a newer baseline now, so we end up
with missing symbol errors:
ld-elf.so.1: /lib/libc.so.7: version FBSD_1.7 required by /usr/local/lib/libpython3.9.so.1.0 not found
It also appears that the fdpass test still fails on 13.1, so update the
condition to less than 14.0
I am trying to run cross-compiled tests in QEMU with the build directory
mounted via smbfs, and therefore creating the sockets in the CWD does not
work. Using DBUS_TEST_SOCKET_DIR (/tmp by default) allows me to run the
tests successfully.
If the elements field has a fixed nonzero size, accessing elements
beyond that size is technically undefined behaviour, which is caught
by some options of the undefined behaviour sanitizer. Try to use a C99
flexible array, or failing that, a zero-length array (which is a popular
non-standard syntax to achieve the same thing).
dbus 1.15.x has C99 as a requirement, but this commit avoids assuming
C99 in order to make this change backportable to 1.14.x if it becomes
necessary to do so (for example to be able to run tests or fuzzers
against 1.14.x, or if compilers' defaults become more strict).
Signed-off-by: Simon McVittie <smcv@collabora.com>
Similar to dbus/dbus!286, but more so: just use the package names,
ignoring their version numbers completely.
pcre2 is not strictly needed at the moment, but it'll be a dependency
for GLib >= 2.73.x (older versions used pcre). For a bit of
future-proofing, download both pcre and pcre2.
Signed-off-by: Simon McVittie <smcv@collabora.com>
Debug messages in a background thread can corrupt the machine-readable
TAP output, and in particular GWin32AppInfo emits debug messages from
a background thread when we link to libgio.
Resolves: https://gitlab.freedesktop.org/dbus/dbus/-/issues/414
Signed-off-by: Simon McVittie <smcv@collabora.com>
With internal DBus checks disabled, but with assertions enabled, the
function would be ifdef'ed out. This is problematic, since the function
is called from within an assertion statement in _dbus_variant_write().
Fixes#412.
Co-authored-by: Simon McVittie <smcv@collabora.com>
The /odd-limit/at test passes on 13.1 and 14.0 images, but fails on 13.1.
Debugging has not given me any useful hints why this may be the case, so
disable this test on 13.0 for now.
This allows us to drop the ci_test_fatal: "no" override which will ensure
that any FreeBSD regressions are caught.
The Makefile.am files contain % pattern rules that are not supported by
`make` (bmake) on FreeBSD. Since the replacing the patterns is non-trivial,
this commit updates the CI script to use GNU make when building on FreeBSD.
