NEWS for #101698

Signed-off-by: Simon McVittie <smcv@collabora.com>

tests: Make tests fail if they try to connect to the real session bus

It is too easy for a developer working in an environment that has a
session bus to write tests that pass locally, but fail in minimal
environments. This is also risky because the tests might do
destructive things on the developer's real session bus. We can avoid
connecting to the session bus by consistently removing its address
from the environment, and replacing it with something that will
always fail.

Signed-off-by: Simon McVittie <smcv@collabora.com>
Reviewed-by: Philip Withnall <withnall@endlessm.com>
Bug: https://bugs.freedesktop.org/show_bug.cgi?id=101698

test/dbus-daemon: Unset DBUS_SESSION_BUS_ADDRESS

When we intend to exercise the default behaviour in the absence of
DBUS_SESSION_BUS_ADDRESS (but with an XDG_RUNTIME_DIR present), it would
help if we unset DBUS_SESSION_BUS_ADDRESS. Otherwise we'll just connect
to the real session bus, if there is one.

Signed-off-by: Simon McVittie <smcv@collabora.com>
Reviewed-by: Philip Withnall <withnall@endlessm.com>
Bug: https://bugs.freedesktop.org/show_bug.cgi?id=101698

Update NEWS for 1.10 branch

Signed-off-by: Simon McVittie <smcv@collabora.com>

name-test: Backport dbus-run-session wrapper from git master

test-pending-call-disconnected relies on being run under a session bus.
On master, the TESTS in this directory all get that treatment, but
in dbus-1.10 they do not. This caused test-pending-call-disconnected
to fail in minimal environments like travis-ci where there is no
developer-initiated session bus.

Backport part of commit ec6b220 "name-test: run most C tests directly,
not via run-test.sh" to wrap it in dbus-run-session. This is better
than putting it in run-test.sh because this way, its TAP output is
parsed directly by Automake.

It also has the side benefit of exercising dbus-run-session in the
automated tests.

Signed-off-by: Simon McVittie <smcv@collabora.com>
Reviewed-by: Philip Withnall <withnall@endlessm.com>
Bug: https://bugs.freedesktop.org/show_bug.cgi?id=101698

dbus_message_iter_open_container: Don't leak signature on failure

If we run out of memory while calling _dbus_type_writer_recurse()
(which is impossible for most contained types, but can happen for
structs and dict-entries), then the memory we allocated in the call to
_dbus_message_iter_open_signature() will still be allocated, and we
have to free it in order to return to the state of the world prior to
calling open_container().

One might reasonably worry that this change can break callers that use
this (incorrect) pattern:

    if (!dbus_message_iter_open_container (outer, ..., inner))
      {
        dbus_message_iter_abandon_container (outer, inner);
        goto fail;
      }
    /* now we know inner is open, and we must close it later */

However, testing that pattern with _dbus_test_oom_handling()
demonstrates that it already dies with a DBusString assertion failure
even before this commit.

This is all concerningly fragile, and I think the next step should be
to zero out DBusMessageIter instances when they are invalidated, so
that a "double-free" is always detected.

Signed-off-by: Simon McVittie <smcv@collabora.com>
Reviewed-by: Philip Withnall <withnall@endlessm.com>
Bug: https://bugs.freedesktop.org/show_bug.cgi?id=101568
(cherry picked from commit 031aa2ceb3dfff373e7b398dfc5d020d77262512)

dbus_message_iter_append_basic: Don't leak signature if appending fd fails

Signed-off-by: Simon McVittie <smcv@collabora.com>
Reviewed-by: Philip Withnall <withnall@endlessm.com>
Bug: https://bugs.freedesktop.org/show_bug.cgi?id=101568
(cherry picked from commit 8384e795516066960bb9fcfbfe138f569420edb9)

dbus_message_append_args_valist: Don't leak memory on inappropriate type

Found by source code inspection while trying to debug an unrelated
leak.

Signed-off-by: Simon McVittie <smcv@collabora.com>
Reviewed-by: Philip Withnall <withnall@endlessm.com>
Bug: https://bugs.freedesktop.org/show_bug.cgi?id=101568
(cherry picked from commit 6b7bdb105b120b3db312de93af94af1bb6a2a474)

Start on 1.10.22

Signed-off-by: Simon McVittie <smcv@debian.org>

1.10.20

Signed-off-by: Simon McVittie <smcv@collabora.com>

Update NEWS for 1.10.x branch

Signed-off-by: Simon McVittie <smcv@collabora.com>

transport: Don't pile up errors for semicolon-separated components

If we somehow get an autolaunch address with multiple
semicolon-separated components, and one of them fails, then we will
hit an assertion failure when we try the next one.

Signed-off-by: Simon McVittie <smcv@collabora.com>
Reviewed-by: Philip Withnall <withnall@endlessm.com>
Bug: https://bugs.freedesktop.org/show_bug.cgi?id=101257
(cherry picked from commit ecdcb86bff42d2bb9cac617bf79f0aa3d47676d9)

Doxyfile.in: do not put timestamps in HTML

The build timestamp is not particularly useful (the version number of
the package is already present in the HTML), and it prevents the build
from being reproducible. See <https://reproducible-builds.org/> for more
information.

Signed-off-by: Simon McVittie <smcv@debian.org>
Reviewed-by: Philip Withnall <withnall@endlessm.com>
Bug: https://bugs.freedesktop.org/show_bug.cgi?id=100692
(cherry picked from commit 0310ead0022b3537392869cc2ed3296ba1a7c17d)

Fix missing dbus_message_unref() in error reply path

The error message was leaked when blocking on a pending call after
the connection was disconnected.

Reviewed-by: Philip Withnall <withnall@endlessm.com>
[smcv: re-word commit message]
Reviewed-by: Simon McVittie <smcv@collabora.com>
Bug: https://bugs.freedesktop.org/show_bug.cgi?id=101481

HACKING: Update git URIs to sync up with cgit.freedesktop.org

git.freedesktop.org no longer offers anonymous checkouts; that facility
moved to anongit.freedesktop.org.

Signed-off-by: Simon McVittie <smcv@collabora.com>
Bug: https://bugs.freedesktop.org/show_bug.cgi?id=100715

Start changelog for 1.10.20

1.10.18

Signed-off-by: Simon McVittie <smcv@debian.org>

test: Fix a couple of memory leaks in test-corrupt

Spotted while testing bug #100568.

Signed-off-by: Philip Withnall <withnall@endlessm.com>
Reviewed-by: Simon McVittie <smcv@collabora.com>
Bug: https://bugs.freedesktop.org/show_bug.cgi?id=100568

test: Fix reading off the end of an array in test-corrupt

One level of pointer indirection too many when passing the arguments to
dbus_message_append_args().

Bug: https://bugs.freedesktop.org/show_bug.cgi?id=100568
Signed-off-by: Philip Withnall <withnall@endlessm.com>
Reviewed-by: Simon McVittie <smcv@collabora.com>

refer to Debian bug in NEWS

NEWS

Initialize SELinux and Apparmor after capabilities are set

avc_init() in the SELinux code path is creating a new thread, we need to
set to capabilities before it gets created so it has the permission to
send audit messages.

It also make more sense to open the audit netlink before the different
logging callbacks are set.

Bug: https://bugs.freedesktop.org/show_bug.cgi?id=92832
[smcv: add comments explaining why initialization must happen in this
specific order]
Reviewed-by: Simon McVittie <simon.mcvittie@collabora.co.uk>
Bug-Debian: https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=857660
Reviewed-by: Stephen Smalley <sds@tycho.nsa.gov>
(cherry picked from commit a3a5935a0a038c3b44c61ce5719f0f7e647b96c6)

NEWS: fix typo

Signed-off-by: Simon McVittie <simon.mcvittie@collabora.co.uk>

Start 1.10.18

Signed-off-by: Simon McVittie <simon.mcvittie@collabora.co.uk>

Prepare 1.10.16

activation test: Fix time-of-check/time-of-use bug waiting to happen

Creating a directory is atomic, stat'ing it to see whether to remove
it is very much not.

Bug: https://bugs.freedesktop.org/show_bug.cgi?id=99828
Signed-off-by: Simon McVittie <simon.mcvittie@collabora.co.uk>
Reviewed-by: Philip Withnall <withnall@endlessm.com>

Change _dbus_create_directory to fail for existing directories

If we don't trap EEXIST and its Windows equivalent, we are unable to
detect the situation where we create an ostensibly unique
subdirectory in a shared /tmp, but an attacker has already created it.
This affects dbus-nonce (the nonce-tcp transport) and the activation
reload test.

Add a new _dbus_ensure_directory() for the one case where we want it to
succeed even on EEXIST: the DBUS_COOKIE_SHA1 keyring, which we know
we are creating in our own trusted "official" $HOME. In the new
transient service support on Bug #99825, ensure_owned_directory()
would need the same treatment.

We are not treating this as a serious security problem, because the
nonce-tcp transport is rarely enabled on Unix and there are multiple
mitigations.

The nonce-tcp transport creates a new unique file with O_EXCL and 0600
(private to user) permissions, then overwrites the requested filename
via atomic-overwrite, so the worst that could happen there is that an
attacker could place a symbolic link matching the name of a directory
we are going to create, causing a dbus-daemon configured for nonce-tcp
to traverse the symlink and atomically overwrite a file named "nonce"
in a directory of the attacker's choice, with new random contents that
are not known to the attacker. This seems unlikely to be exploitable
for anything worse than denial of service in practice. In mainline
Linux since 3.6, this attack is also defeated by the
fs.protected_symlinks sysctl, which many distributions enable by default.

The activation reload test suffers from a classic symlink attack
due to time-of-check/time-of-use errors in its implementation, but as
part of the developer-only "embedded tests" that are only intended
to be run on a trusted machine, it is not treated as security-sensitive.
That code path will be fixed in a subsequent commit.

Bug: https://bugs.freedesktop.org/show_bug.cgi?id=99828
Signed-off-by: Simon McVittie <simon.mcvittie@collabora.co.uk>
Reviewed-by: Philip Withnall <withnall@endlessm.com>

Travis-CI: Get new autoconf-archive from Ubuntu

Hopefully this has better uptime than snapshot.debian.org, which is
really an archival service rather than a production component.

This particular autoconf-archive version was in Ubuntu 16.10, so it
should stay around for a while.

Signed-off-by: Simon McVittie <smcv@debian.org>
(cherry picked from commit 9935a5b7d177b2368506fa48e7fdfa94ea0151fa)

Update NEWS for 1.10 branch

ci-install, ci-build: add brief documentation for parameter variables

Signed-off-by: Simon McVittie <simon.mcvittie@collabora.co.uk>
Bug: https://bugs.freedesktop.org/show_bug.cgi?id=98889

ci-install.sh: make ci_in_docker follow the yes/no convention too

Signed-off-by: Simon McVittie <simon.mcvittie@collabora.co.uk>
Bug: https://bugs.freedesktop.org/show_bug.cgi?id=98889

travis-ci: Add and use infrastructure to build and test in Docker

Debian stable, Debian testing and Ubuntu LTS provide a reasonable
spectrum of old and new distributions. I'm only doing one build on
each to avoid a combinatorial explosion of options.

The Docker images don't have any deb-src apt sources set up, so don't
use `apt-get build-dep`; just include dependencies manually.

Signed-off-by: Simon McVittie <simon.mcvittie@collabora.co.uk>
Reviewed-by: Philip Withnall <philip.withnall@collabora.co.uk>
Bug: https://bugs.freedesktop.org/show_bug.cgi?id=98889

travis-ci: add an install script instead of open-coding it in .travis.yml

Reviewed-by: Philip Withnall <philip.withnall@collabora.co.uk>
[smcv: move comment to install script as suggested]
Signed-off-by: Simon McVittie <simon.mcvittie@collabora.co.uk>
Bug: https://bugs.freedesktop.org/show_bug.cgi?id=98889

travis-ci: introduce maybe_fail_tests() to make test failure more obvious

Taken from the version I added to OSTree.

Signed-off-by: Simon McVittie <simon.mcvittie@collabora.co.uk>
Reviewed-by: Philip Withnall <philip.withnall@collabora.co.uk>
Bug: https://bugs.freedesktop.org/show_bug.cgi?id=98889

travis-ci: run in bash, with the "unofficial strict mode"

set -u forces us to set all variables that we use (for example with the
${foo:=bar} syntax to take an existing value or set a default), or use the
${foo:-bar} syntax to make it explicit that the variable might be unset.

set -o pipefail (which is a bash feature) detects failure in non-last
elements of a pipeline.

Signed-off-by: Simon McVittie <simon.mcvittie@collabora.co.uk>
Reviewed-by: Philip Withnall <philip.withnall@collabora.co.uk>
Bug: https://bugs.freedesktop.org/show_bug.cgi?id=98889

travis-ci: consistently use yes/no instead of yes/empty

Signed-off-by: Simon McVittie <simon.mcvittie@collabora.co.uk>
Reviewed-by: Philip Withnall <philip.withnall@collabora.co.uk>
Bug: https://bugs.freedesktop.org/show_bug.cgi?id=98889

travis-ci: consistently use ci_* for parameter variables

This aligns it with the more generic script based on this one that
I sent to OSTree.

Signed-off-by: Simon McVittie <simon.mcvittie@collabora.co.uk>
Reviewed-by: Philip Withnall <philip.withnall@collabora.co.uk>
Bug: https://bugs.freedesktop.org/show_bug.cgi?id=98889

ci-build: retab with 4-space indentation

This realigns it with the script loosely based on this one that I
sent to OSTree.

Signed-off-by: Simon McVittie <simon.mcvittie@collabora.co.uk>
Reviewed-by: Philip Withnall <philip.withnall@collabora.co.uk>
Bug: https://bugs.freedesktop.org/show_bug.cgi?id=98889

travis-ci: add an explicit copyright/license statement

Signed-off-by: Simon McVittie <simon.mcvittie@collabora.co.uk>
Reviewed-by: Philip Withnall <philip.withnall@collabora.co.uk>
Bug: https://bugs.freedesktop.org/show_bug.cgi?id=98889

Rename distro-style CI build from "release" to "production"

This avoids confusion with the meaning of "release" used by
AX_IS_RELEASE. AX_IS_RELEASE is about facts about the source tree,
namely the distinction between releases (tags) and random snapshots.
The build variants in .travis.yml are about facts about the build
being done, namely the distinction between production and
debug/developer builds.

Production builds are sometimes referred to as "release builds",
for example in typical CMake and MSVC build environments, but a
different term seems better here.

Signed-off-by: Simon McVittie <smcv@debian.org>
Bug: https://bugs.freedesktop.org/show_bug.cgi?id=97357
[smcv: cherry-picked from master to dbus-1.10 to get the Travis-CI setup
consistent between the two branches]

Update autoconf-archive, and use snapshot.debian.org

This way the link won't expire in future.

Signed-off-by: Simon McVittie <smcv@debian.org>
[smcv: cherry-picked from master to dbus-1.10 to get the Travis-CI setup
consistent between the two branches; it is not strictly needed on dbus-1.10]

Travis-CI: fetch a newer autoconf-archive from Debian

Signed-off-by: Simon McVittie <simon.mcvittie@collabora.co.uk>
[smcv: cherry-picked from master to dbus-1.10 to get the Travis-CI setup
consistent between the two branches; it is not strictly needed on dbus-1.10]

Travis-CI: install new autoconf-archive build-dependency

Signed-off-by: Simon McVittie <simon.mcvittie@collabora.co.uk>
[smcv: cherry-picked from master to dbus-1.10 to get the Travis-CI setup
consistent between the two branches; it is not strictly needed on dbus-1.10]

Start developing 1.10.16

Signed-off-by: Simon McVittie <simon.mcvittie@collabora.co.uk>

dbus 1.10.14

Signed-off-by: Simon McVittie <simon.mcvittie@collabora.co.uk>

Suppress -Wmisleading-indentation for this stable branch

We are not going to fix the inconsistent tab/space indentation in a
stable branch just to keep gcc happy.

Signed-off-by: Simon McVittie <simon.mcvittie@collabora.co.uk>

Make uid 0 immune to pending_fd_timeout limit

This is a workaround for
<https://bugs.freedesktop.org/show_bug.cgi?id=95263>. If a service
sends a file descriptor sufficiently frequently that its queue of
messages never goes down to 0 fds pending, then it will eventually be
disconnected. logind is one such service.

We do not currently have a good solution for this: the proposed
patches either don't work, or reintroduce a denial of service
security vulnerability (CVE-2014-3637). Neither seems desirable.
However, we can avoid the worst symptoms by trusting uid 0 not to be
malicious.

Bug: https://bugs.freedesktop.org/show_bug.cgi?id=95263
Bug-Ubuntu: https://bugs.launchpad.net/ubuntu/+source/systemd/+bug/1591411
Reviewed-by: Łukasz Zemczak
Tested-by: Ivan Kozik
Tested-by: Finn Herpich
Tested-by: autostatic
Tested-by: Ben Parafina
Signed-off-by: Simon McVittie <simon.mcvittie@collabora.co.uk>
(cherry picked from commit d5fae1db789d741295ca4746b84915d4bec591fd)
[smcv: omit the test/dbus-daemon.c part, which does not apply unless
a363822f5f58e5513e30dc2f84a30ae03cd91e07 is also applied]

Log to syslog when pending_fd_timeout is exceeded

This is either a denial-of-service attempt, a pathological performance
problem or a dbus-daemon bug. Sysadmins should be told about any of
these.

Bug: https://bugs.freedesktop.org/show_bug.cgi?id=86442
[smcv: add units to timeout: it is in milliseconds]
Signed-off-by: Simon McVittie <smcv@debian.org>
(cherry picked from commit 05cb619f0a6a4c9997832cb4acbb26f7a0be66c3)

Install mingw build-dependencies in a different order

This avoids installing the build-dependencies for dbus and its tests,
then uninstalling them all because they rely on libraries whose versions
are older than the ones needed by wine:i386 (and apparently apt prefers
to remove those libraries rather than upgrade them). Doing it this way
round seems to convince apt to do the right thing.

Signed-off-by: Simon McVittie <simon.mcvittie@collabora.co.uk>

ci-build: run our copy of config.guess

Signed-off-by: Simon McVittie <simon.mcvittie@collabora.co.uk>

NEWS

Signed-off-by: Simon McVittie <simon.mcvittie@collabora.co.uk>

Don't test X11 autolaunching if it was disabled at compile time

Signed-off-by: Simon McVittie <simon.mcvittie@collabora.co.uk>
Reviewed-by: Philip Withnall <philip.withnall@collabora.co.uk>
Bug: https://bugs.freedesktop.org/show_bug.cgi?id=98665

1.10.12

Disable deprecation warnings for stable branch

We're not going to replace deprecated functions here.

Signed-off-by: Simon McVittie <smcv@debian.org>
Reviewed-by: Colin Walters <walters@verbum.org>
Bug: https://bugs.freedesktop.org/show_bug.cgi?id=98157

Ignore ActivationFailure if not using systemd activation

This isn't security-related, just defensive programming: if
dbus-daemon wasn't run with --systemd-activation, then there is no
reason why systemd would legitimately send us this signal, and if it
does we should just ignore it.

Signed-off-by: Simon McVittie <smcv@debian.org>
Reviewed-by: Colin Walters <walters@verbum.org>
Bug: https://bugs.freedesktop.org/show_bug.cgi?id=98157

bus_driver_handle_message: reject ActivationFailure if unprivileged

Specifically, this will allow ActivationFailure messages from our
own uid or from root, but reject them otherwise, even if the bus
configuration for who can own org.freedesktop.systemd1 is entirely
wrong due to something like CVE-2014-8148.

Signed-off-by: Simon McVittie <smcv@debian.org>
Reviewed-by: Colin Walters <walters@verbum.org>
Bug: https://bugs.freedesktop.org/show_bug.cgi?id=98157

dbus_activation_systemd_failure: do not use non-literal format string

In principle this could lead to arbitrary memory overwrite via
a format string attack in the message received from systemd,
resulting in arbitrary code execution.

This is not believed to be an exploitable security vulnerability on the
system bus in practice: it can only be exploited by the owner of the
org.freedesktop.systemd1 bus name, which is restricted to uid 0, so
if systemd is attacker-controlled then the system is already doomed.
Similarly, if a systemd system unit mentioned in the activation failure
message has an attacker-controlled name, then the attacker likely already
has sufficient access to execute arbitrary code as root in any case.

However, prior to dbus 1.8.16 and 1.9.10, due to a missing check for
systemd's identity, unprivileged processes could forge activation
failure messages which would have gone through this code path.
We thought at the time that this was a denial of service vulnerability
(CVE-2015-0245); this bug means that it was in fact potentially an
arbitrary code execution vulnerability.

Bug found using -Wsuggest-attribute=format and -Wformat-security.

Signed-off-by: Simon McVittie <simon.mcvittie@collabora.co.uk>
Reviewed-by: Colin Walters <walters@verbum.org>
Bug: https://bugs.freedesktop.org/show_bug.cgi?id=98157

NEWS

DBusMessage: Fix UB (misaligned access) in call to _dbus_header_set_field_basic()

The const void* 'value' pointer that is passed the address of a
uint32_t here eventually ends up in _dbus_marshal_write_basic(), which
casts it to a DBusBasicValue, a union type that has an alignment of
eight on 64-bit platforms and is therefore more-aligned than the
uint32.

The read of a value of a more-aligned type through a pointer to a less
-aligned type is undefined behaviour.

Fix by storing the uint32 in a DBusBasicValue and passing that instead.

Found by UBSan:

  dbus/dbus/dbus-marshal-basic.c:832:14: runtime error: member access within misaligned address 0x7fdb8dac3a04 for type 'const union DBusBasicValue', which requires 8 byte alignment
  0x7fdb8dac3a04: note: pointer points here
    4a 87 b5 71 01 00 00 00  40 7d 01 00 00 61 00 00  10 3b ac 8d db 7f 00 00  2c 2a 3e 94 db 7f 00 00
                ^
    #0 0x7fdb9444a2c3 in _dbus_marshal_write_basic dbus/dbus/dbus-marshal-basic.c:832
    #1 0x7fdb943d22fb in _dbus_type_writer_write_basic_no_typecode dbus/dbus/dbus-marshal-recursive.c:1605
    #2 0x7fdb943d64e9 in _dbus_type_writer_write_basic dbus/dbus/dbus-marshal-recursive.c:2327
    #3 0x7fdb943c52a6 in write_basic_field dbus/dbus/dbus-marshal-header.c:318
    #4 0x7fdb943c919e in _dbus_header_set_field_basic dbus/dbus/dbus-marshal-header.c:1321
    #5 0x7fdb943e1349 in dbus_message_set_reply_serial dbus/dbus/dbus-message.c:1173

Signed-off-by: Marc Mutz <marc@kdab.net>
Reviewed-by: Simon McVittie <simon.mcvittie@collabora.co.uk>
Bug: https://bugs.freedesktop.org/show_bug.cgi?id=98035

NEWS for 1.10.x

autogen.sh: fail cleanly if autoconf fails

Signed-off-by: Simon McVittie <simon.mcvittie@collabora.co.uk>
(cherry picked from commit 0bbfca6def7a39c03b7a0a4cba166fdf89f3c630)

1.10.11

1.10.10

Signed-off-by: Simon McVittie <smcv@debian.org>

Add a regression test for dbus-launch in X11

[Modified for the dbus-1.10 branch: DBUS_USE_TEST_BINARY is needed.]

Signed-off-by: Simon McVittie <smcv@debian.org>

test-dbus-daemon-fork: exercise closed stdin, stdout, stderr

Signed-off-by: Simon McVittie <smcv@debian.org>

test: expand dbus-launch-eval test to cover stdin being closed

Signed-off-by: Simon McVittie <smcv@debian.org>

Add a simple test for dbus-daemon --fork

[Modified for the dbus-1.10 branch: DBUS_USE_TEST_BINARY is needed.]

Signed-off-by: Simon McVittie <smcv@debian.org>

test: add a regression test for eval "$(dbus-launch --sh-syntax)"

[Modified for the dbus-1.10 branch: DBUS_USE_TEST_BINARY is needed.]

Signed-off-by: Simon McVittie <smcv@debian.org>

dbus-daemon, dbus-launch: cope with callers having closed standard fds

In Debian bug <https://bugs.debian.org/829348>, lightdm appears to
have been starting dbus-launch with at least one of the three
standard fds 0, 1, 2 (stdin, stdout, stderr) closed. This resulted
in the dbus-daemon's epoll_create1() returning a fd less than 3.
Later, _dbus_become_daemon() replaces fds 0-2 with /dev/null. As a
result, a subsequent call to _dbus_loop_add_watch() for the reload
pipe resulted in calling epoll_ctl on the non-epoll fd pointing to
/dev/null, which fails with EINVAL, resulting in the dbus-daemon
exiting unsuccessfully.

Unix programs are not normally expected to behave correctly when
launched with the standard fds not already open; but at the same time,
X11 autolaunching means that dbus-launch (and hence the dbus-daemon)
can get started from an arbitrarily precarious situation.

Bug: https://bugs.freedesktop.org/show_bug.cgi?id=97008
Signed-off-by: Simon McVittie <smcv@debian.org>
Reviewed-by: Thiago Macieira <thiago@kde.org>
(cherry picked from commit c8f73a2a3a9d9d10587f596a62ebb64e8963197e)

_dbus_ensure_standard_fds: new function to ensure std* fds are open

This function opens stdin, stdout, stderr pointing to /dev/null
if they aren't already open. Optionally, it can also replace
whatever is available on those fds with /dev/null.

To allow for use in contexts where only async-signal-safe functions
should be used, such as between fork() and a following exec(),
this function does not use conventional libdbus error handling
(which would require malloc). Instead, it sets errno and returns
an explanatory string.

Bug: https://bugs.freedesktop.org/show_bug.cgi?id=97008
Signed-off-by: Simon McVittie <smcv@debian.org>
Reviewed-by: Thiago Macieira <thiago@kde.org>
(cherry picked from commit 69123a6bd2adabbaec1f770fc4573fc3ed4ceca6)

NEWS for 1.10

Mark WaitingForOK state as unused

It should probably be used (see #97298) but the fact that it isn't
is breaking compatibility with gcc 6, so apply a quick workaround
while we look into what's wrong here.

Signed-off-by: Simon McVittie <simon.mcvittie@collabora.co.uk>
Bug: https://bugs.freedesktop.org/show_bug.cgi?id=97282
(cherry picked from commit 21d61180819c141e779d6ecf9919e62e768b6fd9)

update-activation-environment: produce better diagnostics on error

If dbus-daemon or systemd replied to our method call with an error,
we would report it as "invalid arguments" instead of the true error
name and message.

Same root cause as <https://bugs.freedesktop.org/show_bug.cgi?id=96653>.

Signed-off-by: Simon McVittie <smcv@debian.org>

NEWS for 1.10 branch

configure.ac: explicitely check stdint.h

Otherwise HAVE_STDINT_H will not be defined or the var will not be
picked up from cache so builds could fail with errors like:
| ../../dbus-1.10.8/dbus/dbus-internals.h:239:8: error: ‘uintptr_t’ undeclared (first use in this function)

Signed-off-by: Ioan-Adrian Ratiu <adrian.ratiu@ni.com>
[smcv: fix Autoconf underquoting]
Reviewed-by: Simon McVittie <simon.mcvittie@collabora.co.uk>

Update NEWS for merge of dbus-1.10-ci branch

Merge branch 'dbus-1.10-ci' into dbus-1.10

Update NEWS

Merge branch 'dbus-1.10' into dbus-1.10-ci

activation: set children oom_score_adj to 0

If dbus is running as systemd service, dbus daemon is running with
oom_score_adj -900 by OOMScoreAdjust=-900. And children will also have
same value with dbus daemon.
To avoid this, set the child itself values after fork () to 0.

Bug: https://bugs.freedesktop.org/show_bug.cgi?id=32851
Reviewed-by: Simon McVittie <simon.mcvittie@collabora.co.uk>

Use dbus_set_error_from_message() to check for an error

Do not rely on dbus_message_get_args() to turn an ERROR message
into DBusError.

Bug: https://bugs.freedesktop.org/show_bug.cgi?id=96653
Reviewed-by: Simon McVittie <simon.mcvittie@collabora.co.uk>

Quote the DBUS_SESSION_BUS_ADDRESS variable in the shell file

Some D-Bus daemon versions set multiple addresses separated by semi-colon,
which breaks sourcing of the file.

Bug: https://bugs.freedesktop.org/show_bug.cgi?id=94746
Signed-off-by: Thiago Macieira <thiago@kde.org>
Reviewed-by: Simon McVittie <simon.mcvittie@collabora.co.uk>

travis: compile with mingw in both production and debug flavours

This should detect anything that is only a build failure when embedded
tests or verbose mode are enabled, such as the missing include fixed
in commit 4858faf. I'm not enabling embedded tests and verbose mode
orthogonally because we don't want CI builds to take too long.

Signed-off-by: Simon McVittie <simon.mcvittie@collabora.co.uk>
Reviewed-by: Ralf Habacker <ralf.habacker@freenet.de>
(cherry picked from commit 8f6fe1451ead14b82475177f40bbc2c256cba876)

Merge branch 'dbus-1.10' into dbus-1.10-ci

test-segfault: add missing include of <stdio.h> on Windows

Needed for fprintf (stderr, ...).

Signed-off-by: Simon McVittie <simon.mcvittie@collabora.co.uk>
Reviewed-by: Ralf Habacker <ralf.habacker@freenet.de>
(cherry picked from commit 07b7dcd7178f927cd0b3a3282396b7f99c0b1d29)

Fix ambiguous setup of DBusBabySitter struct member child_handle on Windows.

Bug: https://bugs.freedesktop.org/show_bug.cgi?id=95191
Reviewed-by: Simon McVittie <simon.mcvittie@collabora.co.uk>

On Windows make access to member 'refcount' of struct DBusBabysitter thread safe.

Bug: https://bugs.freedesktop.org/show_bug.cgi?id=95191
Reviewed-by: Simon McVittie <simon.mcvittie@collabora.co.uk>

Eliminates a race condition accessing DBusBabysitter instance at startup of babysitter() on Windows.

Ensure that the babysitter thread already owns its one reference to the
babysitter when it starts up, and eliminates the race condition.
This patch requires that DBusBabysitter refcounting is thread-safe
and is based on an analysis and proposal of Simon Mc Vittie.

Bug: https://bugs.freedesktop.org/show_bug.cgi?id=95191
Reviewed-by: Simon McVittie <simon.mcvittie@collabora.co.uk>

Suppress Windows popups and jit debugger when app crashes with exception.

Based on a patch from Yiyang Fei.

Bug: https://bugs.freedesktop.org/show_bug.cgi?id=95155
Reviewed-by: Simon McVittie <simon.mcvittie@collabora.co.uk>

Fix crash in test-spawn unit test app on Windows.

Bug: https://bugs.freedesktop.org/show_bug.cgi?id=95160
Reviewed-by: Simon McVittie <simon.mcvittie@collabora.co.uk>

Fix assert in test-spawn caused by missing initialization of DBusError instance on gcc builds.

Bug: https://bugs.freedesktop.org/show_bug.cgi?id=95160
Reviewed-by: Simon McVittie <simon.mcvittie@collabora.co.uk>
(cherry picked from commit 9323a621e868d6a5b628b89696b1efe0300ff939)

Merge branch 'dbus-1.10' into dbus-1.10-ci

1.10.9

1.10.8

Signed-off-by: Simon McVittie <smcv@debian.org>

Disable travis-ci apt source for Google Chrome

This prevents adding i386 as a foreign architecture. We don't need
Chrome anyway.

Signed-off-by: Simon McVittie <smcv@debian.org>

Merge branch 'dbus-1.10' into dbus-1.10-ci

update NEWS

dbus-test-tool echo: fix sleep documentation

Bug: https://bugs.freedesktop.org/show_bug.cgi?id=94244
Reviewed-by: Simon McVittie <simon.mcvittie@collabora.co.uk>

Statically assert that the DBusMessageIter struct has no padding

Signed-off-by: Simon McVittie <simon.mcvittie@collabora.co.uk>
Reviewed-by: Thiago Macieira <thiago@kde.org>
Bug: https://bugs.freedesktop.org/show_bug.cgi?id=94136

DBusMessageIter: eliminate padding on 64-bit platforms

Previously, 64-bit (LP64 or LLP64) platforms would have had 32 bits
of padding between pad2 and pad3. We want to guarantee that an ISO C
compiler will copy the entire struct when assigning between structs,
but padding is not guaranteed to be copied, so we want to ensure that
the struct is "packed".

Statically assert that the old ABI is compatible with the new ABI.

Reviewed-by: Thiago Macieira <thiago@kde.org>
[smcv: change >= to == as Thiago requested]
Signed-off-by: Simon McVittie <simon.mcvittie@collabora.co.uk>
Bug: https://bugs.freedesktop.org/show_bug.cgi?id=94136