Skip to contentSkip to conversation

package/dbus-broker: new package

PATCH Applied Norbert Lange 2 participants 11 messages Jan 9, 2022 – Jul 27, 2022

lore
Share & tools
lore
#0 lore raw

PATCH v6 1/4 package/dbus-broker: new package

dbus-broker is an alternate implementation of a dbus dameon. It can be used as a drop-in replacement for the system bus daemon, as well as the session bus daemon.

dbus-broker is (basically, and as far as we're concerned in Buildroot) split in two components:

- the actual message bus daemon, that relays messages across clients

- a launcher, which is responsible for setting various aspects of the bus, like setting the policy et al. and opening the socket(s) the message bus daemon will have to listen on...

The launcher can only be used in a systemd setup (it makes heavy use of systemd facilities), while the message bus is generic. However, the message bus daemon is useless without a launcher. There does not exist a non-systemd launcher, which makes dbus-broker actually a systemd-only package; this can be revisited when/if a non-systemd launcher appears.

Note, however, that libdbus is not provided by dbus-borker. People who want to use dbus-broker as the bus daemon, and need libdbus, will have to enable both, see below.

There are two cases:

1. original dbus disabled

Here, we install the config files and systemd socket activation units; dbus-broker provides the system and sessions bus daemons.

In this case, libdbus is not available.

2. original dbus enabled

In this case, we do not install the config files and systemd socket activation units, or define a user: they all are provided by the original dbus, and we piggy-back on those.

In this situation, the default system and sessions message bus are the original dbus, and libdbus is available; dbus-broker is not enabled.

So far, we believe it would be over-engineered to provide a way to allow choosing the bus daemon in the configuraiton. However, users may opt-in to use dbus-broker in a few ways:
- at build-time: by calling systemctl enable/disable from a post-build script (preferred), or by providing drop-in units or presets in an overlay (less preferred) or custom skeleton (as a last resort),
- at runtime (on a RW filesystem): by calling systemctl enable/disable

Note about the user: the path to the system bus socket is a so-called "well-known location": it is expected to be there, by spec. Moving it elsewhere is going to break existing programs. So, the user running the system bus daemon must be able to create that socket.

As we may have two packages providing a system bus daemon, they have to be both able to create the socket, and thus must both be able to write in the directory containing the socket. And since they can be switched at runtime, they must be running as the same user.

We can't just reference the original dbus user, so we duplicate the entry. What is important, is that the user be named 'dbus', as that's what we use in both cases.

Finally, the licensing terms are pretty trivial for dbus-broker itself, but it makes use of third-party code that it inherits as git submodules (that are bundled in the release archive). Thus the licensing is a bit convoluted... The third-party codes claim to be licensed as "Apache-2.0 and LGP-2.1+" in their AUTHORS files, but at the same time claim "**Apache-2.0** OR **LGPL-2.1-or-later**" in their README files. The individual source files (that are used) do not seem to have any licensing header to clarify the situation. So we represent the situation with "Apache-2.0 and/or LGPL-2.1+".

[[email protected]:
- don't select systemd; depend on it instead
- only install config files and systemd units without original dbus
- install a user to run the message bus as
- fix licensing info
- entirely reword and extend the commit log
- add myself to DEVELOPERS as well
]

---
Changes v5 -> v6 (Norbert, respinning after Yann):
- bump version to v29
- Linux 4.17 is now a hard requirement, drop passing option
- sync dbus_broker user to dbus package ( home = /run/dbus )

Changes v4 -> v5 (Yann, after review by Norbert):
- define the user to run as directly in system.conf
- as a consequence, drop the unit drop-in
- add myself to DEVELOPERS as well

Changes v3 -> v4 (Yann, respining after review by Norbert):
- drop the non-systemd case
- drop the launcher option
- reinstate BR2_COREUTILS_HOST_DEPENDENCY and ln --relative
- reinstate the user, explain it

Changes v2 -> v3 (Norbert, respinning after Yann):
- add an own config entry for dbus-broker-launch enabled by default if systemd init is used
- undo BR2_COREUTILS_HOST_DEPENDENCY
- undo adding dbus user - never used by this package
- add condtional audit dependency
- cleanup conditional logic a bit

Changes v1 -> v2 (Yann):
- make launcher conditional
- don't select systemd; don't depend on it either
- don't install systemd units without systemd
- only install config files and systemd units wihtout original dbus
- rename hooks with meaningful names
- fix licensing info
- entirely reword and extend the commit log

---
 DEVELOPERS                           |   2 +
 package/Config.in                    |   1 +
 package/dbus-broker/Config.in        |  23 +++++
 package/dbus-broker/dbus-broker.hash |   3 +
 package/dbus-broker/dbus-broker.mk   |  71 ++++++++++++++++
 package/dbus-broker/dbus.socket      |   5 ++
 package/dbus-broker/session.conf     |  65 ++++++++++++++
 package/dbus-broker/system.conf      | 123 +++++++++++++++++++++++++++
 8 files changed, 293 insertions(+)
 create mode 100644 package/dbus-broker/Config.in
 create mode 100644 package/dbus-broker/dbus-broker.hash
 create mode 100644 package/dbus-broker/dbus-broker.mk
 create mode 100644 package/dbus-broker/dbus.socket
 create mode 100644 package/dbus-broker/session.conf
 create mode 100644 package/dbus-broker/system.conf
DEVELOPERS +2 -0
index ed65c74319..02ba98e410 100644
@@ -2111,6 +2111,7 @@ F:	package/tpm-tools/
 F:	package/trousers/
 
 N:	Norbert Lange <[email protected]>
+F:	package/dbus-broker/
 F:	package/systemd/
 F:	package/tcf-agent/
 
@@ -2914,6 +2915,7 @@ F:	package/asterisk/
 F:	package/cegui/
 F:	package/dahdi-linux/
 F:	package/dahdi-tools/
+F:	package/dbus-broker/
 F:	package/dtc/
 F:	package/dtv-scan-tables/
 F:	package/dvb-apps/
package/Config.in +1 -0
index 5e01187b83..ebc64a25f2 100644
@@ -459,6 +459,7 @@ endmenu
 	source "package/dahdi-tools/Config.in"
 	source "package/davinci-bootcount/Config.in"
 	source "package/dbus/Config.in"
+	source "package/dbus-broker/Config.in"
 	source "package/dbus-cpp/Config.in"
 	source "package/dbus-glib/Config.in"
 	source "package/dbus-python/Config.in"
package/dbus-broker/Config.in new file +23 -0
new file mode 100644
index 0000000000..c7206232da
@@ -0,0 +1,23 @@
+config BR2_PACKAGE_DBUS_BROKER
+	bool "dbus-broker"
+	depends on BR2_USE_MMU
+	depends on BR2_TOOLCHAIN_HAS_THREADS
+	depends on BR2_TOOLCHAIN_HEADERS_AT_LEAST_4_17
+	depends on BR2_PACKAGE_SYSTEMD
+	select BR2_PACKAGE_EXPAT
+	help
+	  Linux D-Bus Message Broker.
+
+	  The dbus-broker project is an implementation of a message bus
+	  as defined by the D-Bus specification. Its aim is to provide
+	  high performance and reliability, while keeping compatibility
+	  to the D-Bus reference implementation.
+
+	  It is exclusively written for Linux systems, and makes use of
+	  many modern features provided by recent linux kernel releases.
+
+	  https://github.com/bus1/dbus-broker/wiki
+
+comment "dbusbroker needs systemd and a toolchain w/ threads"
+	depends on BR2_USE_MMU
+	depends on !BR2_TOOLCHAIN_HAS_THREADS || !BR2_PACKAGE_SYSTEMD
package/dbus-broker/dbus-broker.hash new file +3 -0
new file mode 100644
index 0000000000..26ebab7ac1
@@ -0,0 +1,3 @@
+# Locally calculated
+sha256  4eca425db52b7ab1027153e93fea9b3f11759db9e93ffbf88759b73ddfb8026a  dbus-broker-29.tar.xz
+sha256  3cda3630283eda0eab825abe5ac84d191248c6b3fe1c232a118124959b96c6a4  LICENSE
package/dbus-broker/dbus-broker.mk new file +71 -0

71 lines not on this page · raw message

package/dbus-broker/dbus.socket new file +5 -0
new file mode 100644
index 0000000000..5c373cf450
@@ -0,0 +1,5 @@
+[Unit]
+Description=D-Bus System Message Bus Socket
+
+[Socket]
+ListenStream=/run/dbus/system_bus_socket
package/dbus-broker/session.conf new file +65 -0

65 lines not on this page · raw message

package/dbus-broker/system.conf new file +123 -0

123 lines not on this page · raw message

Signature · 6 lines
2.34.1

_______________________________________________
buildroot mailing list
[email protected]
https://lists.buildroot.org/mailman/listinfo/buildroot
#1 lore raw

PATCH v6 2/4 package/systemd: do not force dbus if dbus-broker is available

From: "Yann E. MORIN" <[email protected]>

dbus-broker fits the bill as a message bus daemon, so only enable the original dbus if dbus-broker is not enabled.

---
 package/systemd/Config.in | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
package/systemd/Config.in +1 -1
index cc0736561e..1c227bb07f 100644
@@ -26,7 +26,7 @@ menuconfig BR2_PACKAGE_SYSTEMD
 	depends on BR2_TOOLCHAIN_GCC_AT_LEAST_5
 	depends on BR2_HOST_GCC_AT_LEAST_5 # host-systemd
 	select BR2_PACKAGE_HAS_UDEV
-	select BR2_PACKAGE_DBUS # runtime dependency only
+	select BR2_PACKAGE_DBUS if !BR2_PACKAGE_DBUS_BROKER # runtime
 	select BR2_PACKAGE_LIBCAP
 	select BR2_PACKAGE_UTIL_LINUX
 	select BR2_PACKAGE_UTIL_LINUX_LIBS
Signature · 6 lines
2.34.1

_______________________________________________
buildroot mailing list
[email protected]
https://lists.buildroot.org/mailman/listinfo/buildroot
#2 lore raw

PATCH v6 3/4 support/testsuite: de-duplicate the systemd runtime tests

In reply to Norbert Lange

From: "Yann E. MORIN" <[email protected]>

Of all the systemd init tests, only one does some additional tests, and for just this lone wolf, we duplicate the test function.

---
 support/testing/tests/init/test_systemd.py | 32 ++++++++--------------
 1 file changed, 12 insertions(+), 20 deletions(-)
support/testing/tests/init/test_systemd.py +12 -20
index f0cc52bac8..80c6776f59 100644
@@ -21,8 +21,9 @@ class InitSystemSystemdBase(InitSystemBase):
         # BR2_TARGET_ROOTFS_TAR is not set
         """.format(infra.filepath("conf/binfmt-misc-kernel-fragment.config"))
 
-    def check_init(self):
-        super(InitSystemSystemdBase, self).check_init("/lib/systemd/systemd")
+    def check_systemd(self, fs):
+        self.start_emulator(fs, "zImage", "vexpress-v2p-ca9")
+        self.check_init("/lib/systemd/systemd")
 
         # Test all units are OK
         output, _ = self.emulator.run("systemctl --no-pager --failed --no-legend")
@@ -35,6 +36,9 @@ class InitSystemSystemdBase(InitSystemBase):
         output, _ = self.emulator.run("journalctl --no-pager --lines 1 --quiet")
         self.assertEqual(len(output), 1)
 
+        # Check the network is up
+        self.check_network("eth0")
+
 
 class TestInitSystemSystemdRoNetworkd(InitSystemSystemdBase):
     config = InitSystemSystemdBase.config + \
@@ -46,9 +50,7 @@ class TestInitSystemSystemdRoNetworkd(InitSystemSystemdBase):
         """.format(infra.filepath("tests/init/systemd-factory"))
 
     def test_run(self):
-        self.start_emulator("squashfs", "zImage", "vexpress-v2p-ca9")
-        self.check_init()
-        self.check_network("eth0")
+        self.check_systemd("squashfs")
 
         # This one must be executed on the target, to check that
         # the factory feature works as expected
@@ -65,9 +67,7 @@ class TestInitSystemSystemdRwNetworkd(InitSystemSystemdBase):
         """
 
     def test_run(self):
-        self.start_emulator("ext2", "zImage", "vexpress-v2p-ca9")
-        self.check_init()
-        self.check_network("eth0")
+        self.check_systemd("ext2")
 
 
 class TestInitSystemSystemdRoIfupdown(InitSystemSystemdBase):
@@ -80,9 +80,7 @@ class TestInitSystemSystemdRoIfupdown(InitSystemSystemdBase):
         """
 
     def test_run(self):
-        self.start_emulator("squashfs", "zImage", "vexpress-v2p-ca9")
-        self.check_init()
-        self.check_network("eth0")
+        self.check_systemd("squashfs")
 
 
 class TestInitSystemSystemdRwIfupdown(InitSystemSystemdBase):
@@ -94,9 +92,7 @@ class TestInitSystemSystemdRwIfupdown(InitSystemSystemdBase):
         """
 
     def test_run(self):
-        self.start_emulator("ext2", "zImage", "vexpress-v2p-ca9")
-        self.check_init()
-        self.check_network("eth0")
+        self.check_systemd("ext2")
 
 
 class TestInitSystemSystemdRoFull(InitSystemSystemdBase):
@@ -125,9 +121,7 @@ class TestInitSystemSystemdRoFull(InitSystemSystemdBase):
         """
 
     def test_run(self):
-        self.start_emulator("squashfs", "zImage", "vexpress-v2p-ca9")
-        self.check_init()
-        self.check_network("eth0")
+        self.check_systemd("squashfs")
 
 
 class TestInitSystemSystemdRwFull(InitSystemSystemdBase):
@@ -155,6 +149,4 @@ class TestInitSystemSystemdRwFull(InitSystemSystemdBase):
         """
 
     def test_run(self):
-        self.start_emulator("ext2", "zImage", "vexpress-v2p-ca9")
-        self.check_init()
-        self.check_network("eth0")
+        self.check_systemd("ext2")
Signature · 6 lines
2.34.1

_______________________________________________
buildroot mailing list
[email protected]
https://lists.buildroot.org/mailman/listinfo/buildroot
#3 lore raw

Re: [Buildroot] [PATCH v6 3/4] support/testsuite: de-duplicate the systemd runtime tests

Quoted 6 lines · Norbert Lange

From: "Yann E. MORIN" <[email protected]>

Of all the systemd init tests, only one does some additional tests, and for just this lone wolf, we duplicate the test function.

Applied to master, thanks.

Regards,
Arnout

Quoted 86 lines
---
  support/testing/tests/init/test_systemd.py | 32 ++++++++--------------
  1 file changed, 12 insertions(+), 20 deletions(-)
support/testing/tests/init/test_systemd.py +11 -17
index f0cc52bac8..80c6776f59 100644
@@ -21,8 +21,9 @@ class InitSystemSystemdBase(InitSystemBase):
          # BR2_TARGET_ROOTFS_TAR is not set
          """.format(infra.filepath("conf/binfmt-misc-kernel-fragment.config"))
 
-    def check_init(self):
-        super(InitSystemSystemdBase, self).check_init("/lib/systemd/systemd")
+    def check_systemd(self, fs):
+        self.start_emulator(fs, "zImage", "vexpress-v2p-ca9")
+        self.check_init("/lib/systemd/systemd")
 
          # Test all units are OK
          output, _ = self.emulator.run("systemctl --no-pager --failed --no-legend")
@@ -35,6 +36,9 @@ class InitSystemSystemdBase(InitSystemBase):
          output, _ = self.emulator.run("journalctl --no-pager --lines 1 --quiet")
          self.assertEqual(len(output), 1)
 
+        # Check the network is up
+        self.check_network("eth0")
+
 
  class TestInitSystemSystemdRoNetworkd(InitSystemSystemdBase):
      config = InitSystemSystemdBase.config + \
@@ -46,9 +50,7 @@ class TestInitSystemSystemdRoNetworkd(InitSystemSystemdBase):
          """.format(infra.filepath("tests/init/systemd-factory"))
 
      def test_run(self):
-        self.start_emulator("squashfs", "zImage", "vexpress-v2p-ca9")
-        self.check_init()
-        self.check_network("eth0")
+        self.check_systemd("squashfs")
 
          # This one must be executed on the target, to check that
          # the factory feature works as expected
@@ -65,9 +67,7 @@ class TestInitSystemSystemdRwNetworkd(InitSystemSystemdBase):
          """
 
      def test_run(self):
-        self.start_emulator("ext2", "zImage", "vexpress-v2p-ca9")
-        self.check_init()
-        self.check_network("eth0")
+        self.check_systemd("ext2")
 
 
  class TestInitSystemSystemdRoIfupdown(InitSystemSystemdBase):
@@ -80,9 +80,7 @@ class TestInitSystemSystemdRoIfupdown(InitSystemSystemdBase):
          """
 
      def test_run(self):
-        self.start_emulator("squashfs", "zImage", "vexpress-v2p-ca9")
-        self.check_init()
-        self.check_network("eth0")
+        self.check_systemd("squashfs")
 
 
  class TestInitSystemSystemdRwIfupdown(InitSystemSystemdBase):
@@ -94,9 +92,7 @@ class TestInitSystemSystemdRwIfupdown(InitSystemSystemdBase):
          """
 
      def test_run(self):
-        self.start_emulator("ext2", "zImage", "vexpress-v2p-ca9")
-        self.check_init()
-        self.check_network("eth0")
+        self.check_systemd("ext2")
 
 
  class TestInitSystemSystemdRoFull(InitSystemSystemdBase):
@@ -125,9 +121,7 @@ class TestInitSystemSystemdRoFull(InitSystemSystemdBase):
          """
 
      def test_run(self):
-        self.start_emulator("squashfs", "zImage", "vexpress-v2p-ca9")
-        self.check_init()
-        self.check_network("eth0")
+        self.check_systemd("squashfs")
 
 
  class TestInitSystemSystemdRwFull(InitSystemSystemdBase):

support/testing/tests/init/test_systemd.py class TestInitSystemSystemdRwFull(InitSystemSystemdBase): · line 152

Quoted 7 lines
support/testing/tests/init/test_systemd.py +1 -3
@@ -155,6 +149,4 @@ class TestInitSystemSystemdRwFull(InitSystemSystemdBase):
          """
 
      def test_run(self):
-        self.start_emulator("ext2", "zImage", "vexpress-v2p-ca9")
-        self.check_init()
-        self.check_network("eth0")
+        self.check_systemd("ext2")

_______________________________________________
buildroot mailing list
[email protected]
https://lists.buildroot.org/mailman/listinfo/buildroot

#4 lore raw

PATCH v6 4/4 support/run-test: add test for systemd using dbus-broker

In reply to Norbert Lange

From: "Yann E. MORIN" <[email protected]>

Add four new tests for systemd (rw and ro in each case):
- use dbus-broker instead of the original dbus
- use the original dbus, with dbus-broker installed

The first two extend the existing IfUpDown test cases by just enabling dbus-broker; the second ones extend this further, by explicitly enabling the original dbus.

For one of the tests, we overload the test_run() function to test that the dbus-broker daemon is indeed running as root. We need not replicate that check in the other dbus-broker-only test, and it does not make sense to test that in tests that have the original dbus enabled.

Presence of the original dbus and dbus-broker on the same system is
valid: the original dbus is used as the default system bus daemon. We do not test switching between the two at runtime, though as this is really too corner-case specific. We just test to ensure the original dbus system bus daemon is not impacted by the presence of dbus-broker.

Note: the 'full' test-case enables all systemd options, and some of them do pull the original dbus package, so we can't use that to test the integration of dbus-broker; instead, we extend the ifupdown case, which does not enable the original dbus.

---
 support/testing/tests/init/test_systemd.py | 37 ++++++++++++++++++++++
 1 file changed, 37 insertions(+)
support/testing/tests/init/test_systemd.py +37 -0
index 80c6776f59..80d172e5e2 100644
@@ -83,6 +83,29 @@ class TestInitSystemSystemdRoIfupdown(InitSystemSystemdBase):
         self.check_systemd("squashfs")
 
 
+class TestInitSystemSystemdRoIfupdownDbusbroker(TestInitSystemSystemdRoIfupdown):
+    config = TestInitSystemSystemdRoIfupdown.config + \
+        """
+        BR2_PACKAGE_DBUS_BROKER=y
+        """
+
+    def test_run(self):
+        # Parent class' test_run() method does exactly that, no more:
+        self.check_systemd("squashfs")
+
+        # Check that the dbus-broker daemon is running as non-root
+        cmd = "find /proc/$(pidof dbus-broker) -maxdepth 1 -name exe -user dbus"
+        out, _ = self.emulator.run(cmd)
+        self.assertEqual(len(out), 1)
+
+
+class TestInitSystemSystemdRoIfupdownDbusbrokerDbus(TestInitSystemSystemdRoIfupdownDbusbroker):
+    config = TestInitSystemSystemdRoIfupdownDbusbroker.config + \
+        """
+        BR2_PACKAGE_DBUS=y
+        """
+
+
 class TestInitSystemSystemdRwIfupdown(InitSystemSystemdBase):
     config = InitSystemSystemdBase.config + \
         """
@@ -95,6 +118,20 @@ class TestInitSystemSystemdRwIfupdown(InitSystemSystemdBase):
         self.check_systemd("ext2")
 
 
+class TestInitSystemSystemdRwIfupdownDbusbroker(TestInitSystemSystemdRwIfupdown):
+    config = TestInitSystemSystemdRwIfupdown.config + \
+        """
+        BR2_PACKAGE_DBUS_BROKER=y
+        """
+
+
+class TestInitSystemSystemdRwIfupdownDbusbrokerDbus(TestInitSystemSystemdRwIfupdownDbusbroker):
+    config = TestInitSystemSystemdRwIfupdownDbusbroker.config + \
+        """
+        BR2_PACKAGE_DBUS=y
+        """
+
+
 class TestInitSystemSystemdRoFull(InitSystemSystemdBase):
     config = InitSystemSystemdBase.config + \
         """
Signature · 6 lines
2.34.1

_______________________________________________
buildroot mailing list
[email protected]
https://lists.buildroot.org/mailman/listinfo/buildroot
#5 lore raw
In reply to Norbert Lange

Norbert, Yann,

As usual, apologies for taking so long to review this. It's because it's a bit
complicated. I actually tried to look at it just after our hackaton in January but it was too complicated to quickly get my head around.

Quoted 42 lines · Norbert Lange

dbus-broker is an alternate implementation of a dbus dameon. It can be used as a drop-in replacement for the system bus daemon, as well as the session bus daemon.

dbus-broker is (basically, and as far as we're concerned in Buildroot) split in two components:

- the actual message bus daemon, that relays messages across clients

- a launcher, which is responsible for setting various aspects of the bus, like setting the policy et al. and opening the socket(s) the message bus daemon will have to listen on...

The launcher can only be used in a systemd setup (it makes heavy use of systemd facilities), while the message bus is generic. However, the message bus daemon is useless without a launcher. There does not exist a non-systemd launcher, which makes dbus-broker actually a systemd-only package; this can be revisited when/if a non-systemd launcher appears.

Note, however, that libdbus is not provided by dbus-borker. People who want to use dbus-broker as the bus daemon, and need libdbus, will have to enable both, see below.

There are two cases:

1. original dbus disabled

Here, we install the config files and systemd socket activation units; dbus-broker provides the system and sessions bus daemons.

In this case, libdbus is not available.

2. original dbus enabled

In this case, we do not install the config files and systemd socket activation units, or define a user: they all are provided by the original dbus, and we piggy-back on those.

In this situation, the default system and sessions message bus are the original dbus, and libdbus is available; dbus-broker is not enabled.

Quoted 9 lines

So far, we believe it would be over-engineered to provide a way to allow choosing the bus daemon in the configuraiton. However, users may opt-in to use dbus-broker in a few ways:
- at build-time: by calling systemctl enable/disable from a post-build script (preferred), or by providing drop-in units or presets in an overlay (less preferred) or custom skeleton (as a last resort),
- at runtime (on a RW filesystem): by calling systemctl enable/disable

I don't know... To me it seems logical that if both are enabled, that you use
dbus-broker for the bus and only use original dbus for libdbus.

So to me the proper approach seems to change dbus itself so the daemon is
optional, i.e. only libdbus is installed if the daemon is not enabled, similar to e.g. libcurl. BR2_PACKAGE_DBUS_DAEMON should default to y because that's what most people want. We'd have to go through existing packages to check if they depend on libdbus or on the daemon - mostly it will be the first. If there is a package that really needs the daemon (I don't actually think so), then we might need to introduce a virtual package for the daemon. Regardless, dbus-broker should depend on !BR2_PACKAGE_DBUS_DAEMON.

Quoted 5 lines

Note about the user: the path to the system bus socket is a so-called "well-known location": it is expected to be there, by spec. Moving it elsewhere is going to break existing programs. So, the user running the system bus daemon must be able to create that socket.

Quoted 8 lines

As we may have two packages providing a system bus daemon, they have to be both able to create the socket, and thus must both be able to write in the directory containing the socket. And since they can be switched at runtime, they must be running as the same user.

We can't just reference the original dbus user, so we duplicate the entry. What is important, is that the user be named 'dbus', as that's what we use in both cases.

If it's a virtual package, then the user could be created by the virtual
package itself.

Quoted 3 lines

Finally, the licensing terms are pretty trivial for dbus-broker itself, but it makes use of third-party code that it inherits as git submodules (that are bundled in the release archive).

Ugh, bundling...

Quoted 7 lines

Thus the licensing is a bit
convoluted... The third-party codes claim to be licensed as "Apache-2.0 and LGP-2.1+" in their AUTHORS files, but at the same time claim "**Apache-2.0** OR **LGPL-2.1-or-later**" in their README files. The individual source files (that are used) do not seem to have any licensing header to clarify the situation. So we represent the situation with "Apache-2.0 and/or LGPL-2.1+".

That's rather f**d than convoluted :-)

Quoted 11 lines

[[email protected]:
- don't select systemd; depend on it instead
- only install config files and systemd units without original dbus
- install a user to run the message bus as
- fix licensing info
- entirely reword and extend the commit log
- add myself to DEVELOPERS as well
]

[snip]

Quoted 19 lines
package/dbus-broker/Config.in new file +13 -0
new file mode 100644
index 0000000000..c7206232da
@@ -0,0 +1,23 @@
+config BR2_PACKAGE_DBUS_BROKER
+	bool "dbus-broker"
+	depends on BR2_USE_MMU
+	depends on BR2_TOOLCHAIN_HAS_THREADS
+	depends on BR2_TOOLCHAIN_HEADERS_AT_LEAST_4_17
+	depends on BR2_PACKAGE_SYSTEMD
+	select BR2_PACKAGE_EXPAT
+	help
+	  Linux D-Bus Message Broker.
+
+	  The dbus-broker project is an implementation of a message bus
+	  as defined by the D-Bus specification. Its aim is to provide
+	  high performance and reliability, while keeping compatibility

package/dbus-broker/Config.in line 23

Quoted 10 lines
package/dbus-broker/Config.in new file +10 -0
+	  to the D-Bus reference implementation.
+
+	  It is exclusively written for Linux systems, and makes use of
+	  many modern features provided by recent linux kernel releases.
+
+	  https://github.com/bus1/dbus-broker/wiki
+
+comment "dbusbroker needs systemd and a toolchain w/ threads"
+	depends on BR2_USE_MMU
+	depends on !BR2_TOOLCHAIN_HAS_THREADS || !BR2_PACKAGE_SYSTEMD

It's a nitpick, but: I think since dbus-broker is so tightly integrated with
systemd, there's no point really to show the comment if systemd is not enabled.

Quoted 15 lines
package/dbus-broker/dbus-broker.hash new file +3 -0
new file mode 100644
index 0000000000..26ebab7ac1
@@ -0,0 +1,3 @@
+# Locally calculated
+sha256  4eca425db52b7ab1027153e93fea9b3f11759db9e93ffbf88759b73ddfb8026a  dbus-broker-29.tar.xz
+sha256  3cda3630283eda0eab825abe5ac84d191248c6b3fe1c232a118124959b96c6a4  LICENSE
package/dbus-broker/dbus-broker.mk new file +0 -0
new file mode 100644
index 0000000000..547b79eb84

package/dbus-broker/dbus-broker.mk line 7

Quoted 7 lines
package/dbus-broker/dbus-broker.mk new file +7 -0
@@ -0,0 +1,71 @@
+################################################################################
+#
+# dbus-broker
+#
+################################################################################
+
+DBUS_BROKER_VERSION = 29

By now we're at 31 already.

The new version uses meson's wrapgit feature instead of submodules, but
fortunately the release tarball still include them.

Quoted 47 lines
package/dbus-broker/dbus-broker.mk new file +47 -0
+DBUS_BROKER_SOURCE = dbus-broker-$(DBUS_BROKER_VERSION).tar.xz
+DBUS_BROKER_SITE = https://github.com/bus1/dbus-broker/releases/download/v$(DBUS_BROKER_VERSION)
+
+# For the third-party code, the licensing legla-info is inconsistent between
+# the AUTHORS and README, so keep both
+DBUS_BROKER_LICENSE = \
+	Apache-2.0, \
+	Apache-2.0 and/or LGPL-2.1+ (c-dvar, c-ini, c-list, c-rbtree, c-shquote, c-stdaux, c-utf8)
+DBUS_BROKER_LICENSE_FILES = \
+	LICENSE \
+	subprojects/c-dvar/AUTHORS subprojects/c-dvar/README.md \
+	subprojects/c-ini/AUTHORS subprojects/c-ini/README.md \
+	subprojects/c-list/AUTHORS subprojects/c-list/README.md \
+	subprojects/c-rbtree/AUTHORS subprojects/c-rbtree/README.md \
+	subprojects/c-shquote/AUTHORS subprojects/c-shquote/README.md \
+	subprojects/c-stdaux/AUTHORS subprojects/c-stdaux/README.md \
+	subprojects/c-utf8/AUTHORS subprojects/c-utf8/README.md
+
+DBUS_BROKER_DEPENDENCIES = expat systemd
+DBUS_BROKER_CONF_OPTS = -Dlauncher=true
+
+ifeq ($(BR2_PACKAGE_AUDIT),y)
+DBUS_BROKER_DEPENDENCIES += audit
+DBUS_BROKER_CONF_OPTS += -Daudit=true
+else
+DBUS_BROKER_CONF_OPTS += -Daudit=false
+endif
+
+ifeq ($(BR2_PACKAGE_LIBSELINUX),y)
+DBUS_BROKER_DEPENDENCIES += libselinux
+DBUS_BROKER_CONF_OPTS += -Dselinux=true
+else
+DBUS_BROKER_CONF_OPTS += -Dselinux=false
+endif
+
+# We must be using the same user as the original dbus, so we can share
+# the home directory and create a socket there. As a consequence, the
+# username and groupname must be dbus:dbus, and they both need to have
+# the same home.
+define DBUS_BROKER_USERS
+	dbus -1 dbus -1 * /run/dbus - dbus DBus messagebus user
+endef
+
+# Only install units for system bus daemon socket if original dbus is not present
+# Only install config and service files if original dbus is not present
+#
+# Note: BR2_COREUTILS_HOST_DEPENDENCY to be able to use ln --relative

package/dbus-broker/dbus-broker.mk line 64

Quoted 10 lines
package/dbus-broker/dbus-broker.mk new file +10 -0
+ifeq ($(BR2_PACKAGE_DBUS),)
+DBUS_BROKER_DEPENDENCIES += $(BR2_COREUTILS_HOST_DEPENDENCY)
+
+define DBUS_BROKER_INSTALL_INIT_SYSTEMD
+	$(INSTALL) -D -m 0644 $(DBUS_BROKER_PKGDIR)/session.conf \
+		$(TARGET_DIR)/usr/share/dbus-1/session.conf
+	$(INSTALL) -D -m 0644 $(DBUS_BROKER_PKGDIR)/system.conf \
+		$(TARGET_DIR)/usr/share/dbus-1/system.conf
+	$(INSTALL) -D -m 0644 $(DBUS_BROKER_PKGDIR)/dbus.socket \
+		$(TARGET_DIR)/usr/lib/systemd/system/dbus.socket

I'm surprised that these files either isn't needed for original dbus, or that
it's included in original dbus but not in dbus-broker...

package/dbus-broker/dbus-broker.mk line 67

Quoted 3 lines
package/dbus-broker/dbus-broker.mk new file +3 -0
+	$(HOST_MAKE_ENV) ln -sf --relative \
+		$(TARGET_DIR)/usr/lib/systemd/system/dbus.socket \
+		$(TARGET_DIR)/usr/lib/systemd/system/sockets.target.wants/dbus.socket

Is it really useful too include the coreutils dependency and HOST_MAKE_ENV,
just to avoid having to put

	ln -sf ../dbus-socket \
		$(TARGET_DIR)/usr/lib/systemd/system/sockets.target.wants/dbus.socket

I'm going to work more on and hopefully merge this series later today. I have
some local changes already (spelling mistakes), so please don't resend just yet.

Regards,
Arnout

package/dbus-broker/dbus-broker.mk line 71

Quoted 4 lines
package/dbus-broker/dbus-broker.mk new file +4 -0
+endef
+endif # !BR2_PACKAGE_DBUS
+
+$(eval $(meson-package))

[snip]
_______________________________________________
buildroot mailing list
[email protected]
https://lists.buildroot.org/mailman/listinfo/buildroot

#6 lore raw

Am Mi., 27. Juli 2022 um 13:08 Uhr schrieb Arnout Vandecappelle <[email protected]>:

Quoted 60 lines

Norbert, Yann,

As usual, apologies for taking so long to review this. It's because it's a bit
complicated. I actually tried to look at it just after our hackaton in January but it was too complicated to quickly get my head around.

Quoted 42 lines · Norbert Lange

dbus-broker is an alternate implementation of a dbus dameon. It can be used as a drop-in replacement for the system bus daemon, as well as the session bus daemon.

dbus-broker is (basically, and as far as we're concerned in Buildroot) split in two components:

- the actual message bus daemon, that relays messages across clients

- a launcher, which is responsible for setting various aspects of the bus, like setting the policy et al. and opening the socket(s) the message bus daemon will have to listen on...

The launcher can only be used in a systemd setup (it makes heavy use of systemd facilities), while the message bus is generic. However, the message bus daemon is useless without a launcher. There does not exist a non-systemd launcher, which makes dbus-broker actually a systemd-only package; this can be revisited when/if a non-systemd launcher appears.

Note, however, that libdbus is not provided by dbus-borker. People who want to use dbus-broker as the bus daemon, and need libdbus, will have to enable both, see below.

There are two cases:

1. original dbus disabled

Here, we install the config files and systemd socket activation units; dbus-broker provides the system and sessions bus daemons.

In this case, libdbus is not available.

2. original dbus enabled

In this case, we do not install the config files and systemd socket activation units, or define a user: they all are provided by the original dbus, and we piggy-back on those.

In this situation, the default system and sessions message bus are the original dbus, and libdbus is available; dbus-broker is not enabled.

Quoted 9 lines

So far, we believe it would be over-engineered to provide a way to allow choosing the bus daemon in the configuraiton. However, users may opt-in to use dbus-broker in a few ways:
- at build-time: by calling systemctl enable/disable from a post-build script (preferred), or by providing drop-in units or presets in an overlay (less preferred) or custom skeleton (as a last resort),
- at runtime (on a RW filesystem): by calling systemctl enable/disable

Quoted 2 lines

I don't know... To me it seems logical that if both are enabled, that you use
dbus-broker for the bus and only use original dbus for libdbus.

Would be fine by me, back then most distros required you to manually enable dbus-broker, now it seems they shift to using it by default if installed.

Quoted 8 lines

So to me the proper approach seems to change dbus itself so the daemon is
optional, i.e. only libdbus is installed if the daemon is not enabled, similar to e.g. libcurl. BR2_PACKAGE_DBUS_DAEMON should default to y because that's what most people want. We'd have to go through existing packages to check if they depend on libdbus or on the daemon - mostly it will be the first. If there is a package that really needs the daemon (I don't actually think so), then we might need to introduce a virtual package for the daemon. Regardless, dbus-broker should depend on !BR2_PACKAGE_DBUS_DAEMON.

Debian does something similar, including splitting of the xml config files and actual daemons into their own packages.

see [1].

Quoted 10 lines
Quoted 10 lines

Note about the user: the path to the system bus socket is a so-called "well-known location": it is expected to be there, by spec. Moving it elsewhere is going to break existing programs. So, the user running the system bus daemon must be able to create that socket.

As we may have two packages providing a system bus daemon, they have to be both able to create the socket, and thus must both be able to write in the directory containing the socket. And since they can be switched at runtime, they must be running as the same user.

Quoted 6 lines
Quoted 3 lines

We can't just reference the original dbus user, so we duplicate the entry. What is important, is that the user be named 'dbus', as that's what we use in both cases.

If it's a virtual package, then the user could be created by the virtual
package itself.

with systemd you probably dont even need a real user for the isolation benefits.
So maybe this would just be necessary if dbus utilities are used, you can also just let systemd create that user dynamically:

[Service]
...
User=dbus
Group=dbus
DynamicUser=true ExecStart=!/usr/bin/dbus-broker-launch --scope system --audit ExecReload=!/usr/bin/busctl call org.freedesktop.DBus /org/freedesktop/DBus org.freedesktop.DBus ReloadConfig

Quoted 142 lines
Quoted 3 lines

Finally, the licensing terms are pretty trivial for dbus-broker itself, but it makes use of third-party code that it inherits as git submodules (that are bundled in the release archive).

Ugh, bundling...

Quoted 7 lines

Thus the licensing is a bit
convoluted... The third-party codes claim to be licensed as "Apache-2.0 and LGP-2.1+" in their AUTHORS files, but at the same time claim "**Apache-2.0** OR **LGPL-2.1-or-later**" in their README files. The individual source files (that are used) do not seem to have any licensing header to clarify the situation. So we represent the situation with "Apache-2.0 and/or LGPL-2.1+".

That's rather f**d than convoluted :-)

Quoted 11 lines

[[email protected]:
- don't select systemd; depend on it instead
- only install config files and systemd units without original dbus
- install a user to run the message bus as
- fix licensing info
- entirely reword and extend the commit log
- add myself to DEVELOPERS as well
]

[snip]

Quoted 19 lines
package/dbus-broker/Config.in new file +13 -0
new file mode 100644
index 0000000000..c7206232da
@@ -0,0 +1,23 @@
+config BR2_PACKAGE_DBUS_BROKER
+     bool "dbus-broker"
+     depends on BR2_USE_MMU
+     depends on BR2_TOOLCHAIN_HAS_THREADS
+     depends on BR2_TOOLCHAIN_HEADERS_AT_LEAST_4_17
+     depends on BR2_PACKAGE_SYSTEMD
+     select BR2_PACKAGE_EXPAT
+     help
+       Linux D-Bus Message Broker.
+
+       The dbus-broker project is an implementation of a message bus
+       as defined by the D-Bus specification. Its aim is to provide
+       high performance and reliability, while keeping compatibility

package/dbus-broker/Config.in line 23

Quoted 10 lines
package/dbus-broker/Config.in new file +10 -0
+       to the D-Bus reference implementation.
+
+       It is exclusively written for Linux systems, and makes use of
+       many modern features provided by recent linux kernel releases.
+
+       https://github.com/bus1/dbus-broker/wiki
+
+comment "dbusbroker needs systemd and a toolchain w/ threads"
+     depends on BR2_USE_MMU
+     depends on !BR2_TOOLCHAIN_HAS_THREADS || !BR2_PACKAGE_SYSTEMD

It's a nitpick, but: I think since dbus-broker is so tightly integrated with
systemd, there's no point really to show the comment if systemd is not enabled.

Quoted 15 lines
package/dbus-broker/dbus-broker.hash new file +3 -0
new file mode 100644
index 0000000000..26ebab7ac1
@@ -0,0 +1,3 @@
+# Locally calculated
+sha256  4eca425db52b7ab1027153e93fea9b3f11759db9e93ffbf88759b73ddfb8026a  dbus-broker-29.tar.xz
+sha256  3cda3630283eda0eab825abe5ac84d191248c6b3fe1c232a118124959b96c6a4  LICENSE
package/dbus-broker/dbus-broker.mk new file +0 -0
new file mode 100644
index 0000000000..547b79eb84

package/dbus-broker/dbus-broker.mk line 7

Quoted 7 lines
package/dbus-broker/dbus-broker.mk new file +7 -0
@@ -0,0 +1,71 @@
+################################################################################
+#
+# dbus-broker
+#
+################################################################################
+
+DBUS_BROKER_VERSION = 29

By now we're at 31 already.

The new version uses meson's wrapgit feature instead of submodules, but
fortunately the release tarball still include them.

Quoted 52 lines
package/dbus-broker/dbus-broker.mk new file +52 -0
+DBUS_BROKER_SOURCE = dbus-broker-$(DBUS_BROKER_VERSION).tar.xz
+DBUS_BROKER_SITE = https://github.com/bus1/dbus-broker/releases/download/v$(DBUS_BROKER_VERSION)
+
+# For the third-party code, the licensing legla-info is inconsistent between
+# the AUTHORS and README, so keep both
+DBUS_BROKER_LICENSE = \
+     Apache-2.0, \
+     Apache-2.0 and/or LGPL-2.1+ (c-dvar, c-ini, c-list, c-rbtree, c-shquote, c-stdaux, c-utf8)
+DBUS_BROKER_LICENSE_FILES = \
+     LICENSE \
+     subprojects/c-dvar/AUTHORS subprojects/c-dvar/README.md \
+     subprojects/c-ini/AUTHORS subprojects/c-ini/README.md \
+     subprojects/c-list/AUTHORS subprojects/c-list/README.md \
+     subprojects/c-rbtree/AUTHORS subprojects/c-rbtree/README.md \
+     subprojects/c-shquote/AUTHORS subprojects/c-shquote/README.md \
+     subprojects/c-stdaux/AUTHORS subprojects/c-stdaux/README.md \
+     subprojects/c-utf8/AUTHORS subprojects/c-utf8/README.md
+
+DBUS_BROKER_DEPENDENCIES = expat systemd
+DBUS_BROKER_CONF_OPTS = -Dlauncher=true
+
+ifeq ($(BR2_PACKAGE_AUDIT),y)
+DBUS_BROKER_DEPENDENCIES += audit
+DBUS_BROKER_CONF_OPTS += -Daudit=true
+else
+DBUS_BROKER_CONF_OPTS += -Daudit=false
+endif
+
+ifeq ($(BR2_PACKAGE_LIBSELINUX),y)
+DBUS_BROKER_DEPENDENCIES += libselinux
+DBUS_BROKER_CONF_OPTS += -Dselinux=true
+else
+DBUS_BROKER_CONF_OPTS += -Dselinux=false
+endif
+
+# We must be using the same user as the original dbus, so we can share
+# the home directory and create a socket there. As a consequence, the
+# username and groupname must be dbus:dbus, and they both need to have
+# the same home.
+define DBUS_BROKER_USERS
+     dbus -1 dbus -1 * /run/dbus - dbus DBus messagebus user
+endef
+
+# Only install units for system bus daemon socket if original dbus is not present
+# Only install config and service files if original dbus is not present
+#
+# Note: BR2_COREUTILS_HOST_DEPENDENCY to be able to use ln --relative
+ifeq ($(BR2_PACKAGE_DBUS),)
+DBUS_BROKER_DEPENDENCIES += $(BR2_COREUTILS_HOST_DEPENDENCY)
+
+define DBUS_BROKER_INSTALL_INIT_SYSTEMD
+     $(INSTALL) -D -m 0644 $(DBUS_BROKER_PKGDIR)/session.conf \
Quoted 8 lines

package/dbus-broker/dbus-broker.mk line 64

Quoted 5 lines
package/dbus-broker/dbus-broker.mk new file +5 -0
+             $(TARGET_DIR)/usr/share/dbus-1/session.conf
+     $(INSTALL) -D -m 0644 $(DBUS_BROKER_PKGDIR)/system.conf \
+             $(TARGET_DIR)/usr/share/dbus-1/system.conf
+     $(INSTALL) -D -m 0644 $(DBUS_BROKER_PKGDIR)/dbus.socket \
+             $(TARGET_DIR)/usr/lib/systemd/system/dbus.socket

I'm surprised that these files either isn't needed for original dbus, or that
it's included in original dbus but not in dbus-broker...

Not sure I understand what you are saying. They are needed for the original dbus,
and not included in dbus-broker as the IPC daemon could (theoretically) run without.
See [2].

Quoted 10 lines

package/dbus-broker/dbus-broker.mk line 67

Quoted 4 lines
package/dbus-broker/dbus-broker.mk new file +3 -0
+     $(HOST_MAKE_ENV) ln -sf --relative \
+             $(TARGET_DIR)/usr/lib/systemd/system/dbus.socket \
+             $(TARGET_DIR)/usr/lib/systemd/system/sockets.target.wants/dbus.socket

Is it really useful too include the coreutils dependency and HOST_MAKE_ENV,
just to avoid having to put

        ln -sf ../dbus-socket \
                $(TARGET_DIR)/usr/lib/systemd/system/sockets.target.wants/dbus.socket

Yann is very determined about that [3].

Quoted 7 lines

I'm going to work more on and hopefully merge this series later today. I have
some local changes already (spelling mistakes), so please don't resend just yet.

Regards,
Arnout

Quoted 6 lines
Quoted 4 lines
package/dbus-broker/dbus-broker.mk new file +4 -0
+endef
+endif # !BR2_PACKAGE_DBUS
+
+$(eval $(meson-package))

[snip]

Norbert

[1] - https://packages.debian.org/source/sid/dbus.
[2] - https://github.com/bus1/dbus-broker/issues/262
[3] - http://lists.busybox.net/pipermail/buildroot/2020-June/587178.html
_______________________________________________
buildroot mailing list
[email protected]
https://lists.buildroot.org/mailman/listinfo/buildroot

#7 lore raw
Quoted 62 lines · Norbert Lange

Am Mi., 27. Juli 2022 um 13:08 Uhr schrieb Arnout Vandecappelle <[email protected]>:

Quoted 60 lines

Norbert, Yann,

As usual, apologies for taking so long to review this. It's because it's a bit
complicated. I actually tried to look at it just after our hackaton in January but it was too complicated to quickly get my head around.

Quoted 51 lines · Norbert Lange

dbus-broker is an alternate implementation of a dbus dameon. It can be used as a drop-in replacement for the system bus daemon, as well as the session bus daemon.

dbus-broker is (basically, and as far as we're concerned in Buildroot) split in two components:

- the actual message bus daemon, that relays messages across clients

- a launcher, which is responsible for setting various aspects of the bus, like setting the policy et al. and opening the socket(s) the message bus daemon will have to listen on...

The launcher can only be used in a systemd setup (it makes heavy use of systemd facilities), while the message bus is generic. However, the message bus daemon is useless without a launcher. There does not exist a non-systemd launcher, which makes dbus-broker actually a systemd-only package; this can be revisited when/if a non-systemd launcher appears.

Note, however, that libdbus is not provided by dbus-borker. People who want to use dbus-broker as the bus daemon, and need libdbus, will have to enable both, see below.

There are two cases:

1. original dbus disabled

Here, we install the config files and systemd socket activation units; dbus-broker provides the system and sessions bus daemons.

In this case, libdbus is not available.

2. original dbus enabled

In this case, we do not install the config files and systemd socket activation units, or define a user: they all are provided by the original dbus, and we piggy-back on those.

In this situation, the default system and sessions message bus are the original dbus, and libdbus is available; dbus-broker is not enabled.

So far, we believe it would be over-engineered to provide a way to allow choosing the bus daemon in the configuraiton. However, users may opt-in to use dbus-broker in a few ways:
- at build-time: by calling systemctl enable/disable from a post-build script (preferred), or by providing drop-in units or presets in an overlay (less preferred) or custom skeleton (as a last resort),
- at runtime (on a RW filesystem): by calling systemctl enable/disable

Quoted 6 lines
Quoted 2 lines

I don't know... To me it seems logical that if both are enabled, that you use
dbus-broker for the bus and only use original dbus for libdbus.

Would be fine by me, back then most distros required you to manually enable dbus-broker, now it seems they shift to using it by default if installed.

Since we anyway already have the infrastructure to automatically enable it if
dbus is not included, it should be easy to do that even if dbus is included.

Quoted 8 lines
Quoted 8 lines

So to me the proper approach seems to change dbus itself so the daemon is
optional, i.e. only libdbus is installed if the daemon is not enabled, similar to e.g. libcurl. BR2_PACKAGE_DBUS_DAEMON should default to y because that's what most people want. We'd have to go through existing packages to check if they depend on libdbus or on the daemon - mostly it will be the first. If there is a package that really needs the daemon (I don't actually think so), then we might need to introduce a virtual package for the daemon. Regardless, dbus-broker should depend on !BR2_PACKAGE_DBUS_DAEMON.

Looking more detailed at it, this seems too complicated. dbus doesn't offer a
way to install only the library. Also in terms of rootfs size, the only difference is the dbus-daemon and dbus-launch binaries, plus a few config files. Not really sufficient to make it worth it.

What I will do, I think, is to include logic in dbus to remove the activation
units (both the service and the socket) when dbus-broker is enabled. It is slightly messy because dbus.mk and dbus-broker.mk become fairly tightly woven together, but I think it's worth it from a usability perspective.

Quoted 21 lines

Debian does something similar, including splitting of the xml config files and actual daemons into their own packages.

see [1].

Quoted 16 lines
Quoted 5 lines

Note about the user: the path to the system bus socket is a so-called "well-known location": it is expected to be there, by spec. Moving it elsewhere is going to break existing programs. So, the user running the system bus daemon must be able to create that socket.

Quoted 8 lines

As we may have two packages providing a system bus daemon, they have to be both able to create the socket, and thus must both be able to write in the directory containing the socket. And since they can be switched at runtime, they must be running as the same user.

We can't just reference the original dbus user, so we duplicate the entry. What is important, is that the user be named 'dbus', as that's what we use in both cases.

If it's a virtual package, then the user could be created by the virtual
package itself.

It won't be a virtual package, so no longer relevant.

Quoted 9 lines

with systemd you probably dont even need a real user for the isolation benefits.
So maybe this would just be necessary if dbus utilities are used, you can also just let systemd create that user dynamically:

[Service]
...
User=dbus
Group=dbus > DynamicUser=true

If we do this both in the dbus and the dbus-broker units, then indeed we can
remove the dbus user.

If I add this just to dbus-broker, but dbus still creates a static user, is
this going to give a problem for systemd? I.e. does systemd support a DynamicUser=true if the user exists already statically? I think so...

Regards,
Arnout

Quoted 3 lines

ExecStart=!/usr/bin/dbus-broker-launch --scope system --audit ExecReload=!/usr/bin/busctl call org.freedesktop.DBus /org/freedesktop/DBus org.freedesktop.DBus ReloadConfig

[snip]
_______________________________________________
buildroot mailing list
[email protected]
https://lists.buildroot.org/mailman/listinfo/buildroot

#8 lore raw

Arnout Vandecappelle <[email protected]> schrieb am Mi., 27. Juli 2022, 19:05:

Quoted 108 lines · Norbert Lange
Quoted 6 lines

Am Mi., 27. Juli 2022 um 13:08 Uhr schrieb Arnout Vandecappelle <[email protected]>:

Quoted 4 lines

Norbert, Yann,

As usual, apologies for taking so long to review this. It's because

it's a bit

Quoted 1 line
Quoted 1 line

complicated. I actually tried to look at it just after our hackaton in

January

Quoted 11 lines
Quoted 11 lines

but it was too complicated to quickly get my head around.

Quoted 8 lines · Norbert Lange

dbus-broker is an alternate implementation of a dbus dameon. It can be used as a drop-in replacement for the system bus daemon, as well as the session bus daemon.

dbus-broker is (basically, and as far as we're concerned in Buildroot) split in two components:

- the actual message bus daemon, that relays messages across

clients

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

- a launcher, which is responsible for setting various aspects of

the

Quoted 6 lines
Quoted 6 lines
Quoted 6 lines

bus, like setting the policy et al. and opening the socket(s) the message bus daemon will have to listen on...

The launcher can only be used in a systemd setup (it makes heavy use of systemd facilities), while the message bus is generic. However, the message bus daemon is useless without a launcher. There does not exist

a

Quoted 11 lines
Quoted 11 lines
Quoted 11 lines

non-systemd launcher, which makes dbus-broker actually a systemd-only package; this can be revisited when/if a non-systemd launcher appears.

Note, however, that libdbus is not provided by dbus-borker. People who want to use dbus-broker as the bus daemon, and need libdbus, will have to enable both, see below.

There are two cases:

1. original dbus disabled

Quoted 8 lines
Quoted 8 lines
Quoted 8 lines

Here, we install the config files and systemd socket activation units; dbus-broker provides the system and sessions bus daemons.

In this case, libdbus is not available.

2. original dbus enabled

In this case, we do not install the config files and systemd

socket

Quoted 4 lines
Quoted 4 lines
Quoted 4 lines

activation units, or define a user: they all are provided by the original dbus, and we piggy-back on those.

In this situation, the default system and sessions message bus

are

Quoted 13 lines
Quoted 13 lines
Quoted 12 lines

the original dbus, and libdbus is available; dbus-broker is not enabled.

So far, we believe it would be over-engineered to provide a way to allow choosing the bus daemon in the configuraiton. However, users may opt-in to use dbus-broker in a few ways:
- at build-time: by calling systemctl enable/disable from a post-build script (preferred), or by providing drop-in units or presets in an overlay (less preferred) or custom skeleton (as a last resort),
- at runtime (on a RW filesystem): by calling systemctl enable/disable

Quoted 1 line
Quoted 1 line

I don't know... To me it seems logical that if both are enabled,

that you use

Quoted 5 lines
Quoted 1 line

dbus-broker for the bus and only use original dbus for libdbus.

Would be fine by me, back then most distros required you to manually enable dbus-broker, now it seems they shift to using it by default if installed.

Since we anyway already have the infrastructure to automatically enable
it if
dbus is not included, it should be easy to do that even if dbus is included.

Quoted 1 line
Quoted 1 line

So to me the proper approach seems to change dbus itself so the

daemon is

Quoted 1 line
Quoted 1 line

optional, i.e. only libdbus is installed if the daemon is not enabled,

similar

Quoted 1 line
Quoted 1 line

to e.g. libcurl. BR2_PACKAGE_DBUS_DAEMON should default to y because

that's what

Quoted 1 line
Quoted 1 line

most people want. We'd have to go through existing packages to check if

they

Quoted 1 line
Quoted 1 line

depend on libdbus or on the daemon - mostly it will be the first. If

there is a

Quoted 1 line
Quoted 1 line

package that really needs the daemon (I don't actually think so), then

we might

Quoted 1 line
Quoted 1 line

need to introduce a virtual package for the daemon. Regardless,

dbus-broker

Quoted 1 line
Quoted 1 line

should depend on !BR2_PACKAGE_DBUS_DAEMON.

Looking more detailed at it, this seems too complicated. dbus doesn't
offer a
way to install only the library. Also in terms of rootfs size, the only difference is the dbus-daemon and dbus-launch binaries, plus a few config files.
Not really sufficient to make it worth it.

Quoted 7 lines

What I will do, I think, is to include logic in dbus to remove the
activation
units (both the service and the socket) when dbus-broker is enabled. It is slightly messy because dbus.mk and dbus-broker.mk become fairly tightly woven
together, but I think it's worth it from a usability perspective.

Sounds messy too,
I would rather have a dbus-common package that includes the socket an config files (and creates the user if necessary).

That package would download dbus, but not build it, instead just copy over those files.
Kinda like util-linux-libs

Quoted 32 lines
Quoted 1 line

Debian does something similar, including splitting of the xml config

files

Quoted 19 lines

and actual daemons into their own packages.

see [1].

Quoted 15 lines
Quoted 5 lines

Note about the user: the path to the system bus socket is a so-called "well-known location": it is expected to be there, by spec. Moving it elsewhere is going to break existing programs. So, the user running the system bus daemon must be able to create that socket.

Quoted 8 lines

As we may have two packages providing a system bus daemon, they have to be both able to create the socket, and thus must both be able to write in the directory containing the socket. And since they can be switched at runtime, they must be running as the same user.

We can't just reference the original dbus user, so we duplicate the entry. What is important, is that the user be named 'dbus', as that's what we use in both cases.

If it's a virtual package, then the user could be created by the

virtual

Quoted 1 line
Quoted 1 line

package itself.

It won't be a virtual package, so no longer relevant.

Quoted 5 lines

with systemd you probably dont even need a real user for the isolation benefits.
So maybe this would just be necessary if dbus utilities are used, you can also just let systemd create that user dynamically:

Quoted 9 lines
Quoted 4 lines

[Service]
...
User=dbus
Group=dbus > DynamicUser=true

If we do this both in the dbus and the dbus-broker units, then indeed we
can
remove the dbus user.

AFAIR the reason for the dbus to be a system user is that dbus-launch can setuid into it.
So it only works without system user if dbus-launch is not used. Dbus-broker delegates this functionality to system (dbus activation)

Quoted 5 lines

If I add this just to dbus-broker, but dbus still creates a static user,
is
this going to give a problem for systemd? I.e. does systemd support a DynamicUser=true if the user exists already statically? I think so...

Think so too.

Quoted 9 lines

Regards,
Arnout

Quoted 3 lines

ExecStart=!/usr/bin/dbus-broker-launch --scope system --audit ExecReload=!/usr/bin/busctl call org.freedesktop.DBus /org/freedesktop/DBus org.freedesktop.DBus ReloadConfig

[snip]

Norbert

Quoted 1 line

_______________________________________________
buildroot mailing list
[email protected]
https://lists.buildroot.org/mailman/listinfo/buildroot

#9 lore raw
Quoted 9 lines · Norbert Lange

Arnout Vandecappelle <[email protected] <mailto:[email protected]>> schrieb am Mi., 27. Juli 2022, 19:05:

    On 27/07/2022 15:47, Norbert Lange wrote:
     > Am Mi., 27. Juli 2022 um 13:08 Uhr schrieb Arnout Vandecappelle
     > <[email protected] <mailto:[email protected]>>:

[snip]

Quoted 28 lines

>>    So to me the proper approach seems to change dbus itself so the daemon is >> optional, i.e. only libdbus is installed if the daemon is not enabled,
similar
>> to e.g. libcurl. BR2_PACKAGE_DBUS_DAEMON should default to y because
that's what
>> most people want. We'd have to go through existing packages to check if they >> depend on libdbus or on the daemon - mostly it will be the first. If
there is a
>> package that really needs the daemon (I don't actually think so), then
we might
>> need to introduce a virtual package for the daemon. Regardless, dbus-broker >> should depend on !BR2_PACKAGE_DBUS_DAEMON.

  Looking more detailed at it, this seems too complicated. dbus doesn't
offer a
way to install only the library. Also in terms of rootfs size, the only difference is the dbus-daemon and dbus-launch binaries, plus a few config files.
Not really sufficient to make it worth it.

  What I will do, I think, is to include logic in dbus to remove the
activation
units (both the service and the socket) when dbus-broker is enabled. It is slightly messy because dbus.mk <http://dbus.mk> and dbus-broker.mk
<http://dbus-broker.mk> become fairly tightly woven
together, but I think it's worth it from a usability perspective.

Quoted 1 line

Sounds messy too,

Agreed. But the only unmessy thing that I found so far is to not use
dbus-broker at all if original dbus is selected, and I think that that pretty much defeats the purpose of including dbus-broker.

Quoted 1 line

I would rather have a dbus-common package that includes the socket an config files (and creates the user if necessary).

That is a good idea. However, it only works for the two dbus configuration
files, not for the systemd units, so it only solves half of the mess. And the added benefit is not that great. Also, they don't change very often any more upstream (the last change is from 12/21, before that it's 12/17). So in the end we decided not to go for dbus-common.

Quoted 39 lines

That package would download dbus, but not build it, instead just copy over those files. Kinda like util-linux-libs

> Debian does something similar, including splitting of the xml config files > and actual daemons into their own packages.
>
> see [1].
>
>>> Note about the user: the path to the system bus socket is a so-called >>> "well-known location": it is expected to be there, by spec. Moving it >>> elsewhere is going to break existing programs. So, the user running the >>> system bus daemon must be able to create that socket.
>>>
>>> As we may have two packages providing a system bus daemon, they have to >>> be both able to create the socket, and thus must both be able to write >>> in the directory containing the socket. And since they can be switched >>> at runtime, they must be running as the same user.
>>>
>>> We can't just reference the original dbus user, so we duplicate the >>> entry. What is important, is that the user be named 'dbus', as that's >>> what we use in both cases.
>>
>>    If it's a virtual package, then the user could be created by the virtual >> package itself.

  It won't be a virtual package, so no longer relevant.

> with systemd you probably dont even need a real user for the isolation > benefits.
> So maybe this would just be necessary if dbus utilities are used, > you can also just let systemd create that user dynamically:
>
> [Service]
> ...
> User=dbus
> Group=dbus > DynamicUser=true

Quoted 6 lines

  If we do this both in the dbus and the dbus-broker units, then indeed we can
remove the dbus user.

AFAIR the reason for the dbus to be a system user is that dbus-launch can setuid into it.
So it only works without system user if dbus-launch is not used. Dbus-broker delegates this functionality to system (dbus activation)

Well, the service file still calls dbus-broker-launch, which I guess does
setuid because the upstream service file doesn't have any User= or Group=

I think we'll just stick to what we have now, and keep the upstream service
file rather than hacking our own...

Regards,
Arnout

Quoted 13 lines

  If I add this just to dbus-broker, but dbus still creates a static user, is
this going to give a problem for systemd? I.e. does systemd support a DynamicUser=true if the user exists already statically? I think so...

Think so too.

  Regards,
  Arnout

Quoted 9 lines

> ExecStart=!/usr/bin/dbus-broker-launch --scope system --audit > ExecReload=!/usr/bin/busctl call org.freedesktop.DBus > /org/freedesktop/DBus org.freedesktop.DBus ReloadConfig

[snip]

Norbert

_______________________________________________
buildroot mailing list
[email protected]
https://lists.buildroot.org/mailman/listinfo/buildroot

#10 lore raw

Arnout Vandecappelle <[email protected]> schrieb am Mi., 27. Juli 2022, 21:09:

Quoted 135 lines · Norbert Lange
Quoted 3 lines

Arnout Vandecappelle <[email protected] <mailto:[email protected]>> schrieb

am Mi.,

Quoted 7 lines

27. Juli 2022, 19:05:

    On 27/07/2022 15:47, Norbert Lange wrote:
     > Am Mi., 27. Juli 2022 um 13:08 Uhr schrieb Arnout Vandecappelle
     > <[email protected] <mailto:[email protected]>>:

[snip]

Quoted 1 line

>> So to me the proper approach seems to change dbus itself so

the daemon is

Quoted 1 line

>> optional, i.e. only libdbus is installed if the daemon is not

enabled,

Quoted 2 lines

similar
>> to e.g. libcurl. BR2_PACKAGE_DBUS_DAEMON should default to y

because

Quoted 2 lines

that's what
>> most people want. We'd have to go through existing packages to

check if they

Quoted 1 line

>> depend on libdbus or on the daemon - mostly it will be the

first. If

Quoted 2 lines

there is a
>> package that really needs the daemon (I don't actually think

so), then

Quoted 2 lines

we might
>> need to introduce a virtual package for the daemon. Regardless,

dbus-broker

Quoted 3 lines

>> should depend on !BR2_PACKAGE_DBUS_DAEMON.

Looking more detailed at it, this seems too complicated. dbus

doesn't

Quoted 2 lines

offer a
way to install only the library. Also in terms of rootfs size, the

only

Quoted 1 line

difference is the dbus-daemon and dbus-launch binaries, plus a few

config

Quoted 6 lines

files.
Not really sufficient to make it worth it.

What I will do, I think, is to include logic in dbus to remove the
activation
units (both the service and the socket) when dbus-broker is enabled.

It is

Quoted 6 lines

slightly messy because dbus.mk <http://dbus.mk> and dbus-broker.mk
<http://dbus-broker.mk> become fairly tightly woven together, but I think it's worth it from a usability perspective.

Sounds messy too,

Agreed. But the only unmessy thing that I found so far is to not use
dbus-broker at all if original dbus is selected, and I think that that pretty
much defeats the purpose of including dbus-broker.

Quoted 1 line

I would rather have a dbus-common package that includes the socket an

config

Quoted 1 line

files (and creates the user if necessary).

That is a good idea. However, it only works for the two dbus
configuration
files, not for the systemd units, so it only solves half of the mess. And the
added benefit is not that great. Also, they don't change very often any more
upstream (the last change is from 12/21, before that it's 12/17). So in the end
we decided not to go for dbus-common.

Quoted 1 line

That package would download dbus, but not build it, instead just copy

over those

Quoted 5 lines

files.
Kinda like util-linux-libs

> Debian does something similar, including splitting of the xml

config files

Quoted 5 lines
     > and actual daemons into their own packages.
     >
     > see [1].
     >
     >>> Note about the user: the path to the system bus socket is a

so-called

Quoted 1 line

>>> "well-known location": it is expected to be there, by spec.

Moving it

Quoted 1 line

>>> elsewhere is going to break existing programs. So, the user

running the

Quoted 3 lines

>>> system bus daemon must be able to create that socket. >>>
>>> As we may have two packages providing a system bus daemon, they

have to

Quoted 1 line

>>> be both able to create the socket, and thus must both be able

to write

Quoted 1 line

>>> in the directory containing the socket. And since they can be

switched

Quoted 3 lines

>>> at runtime, they must be running as the same user. >>>
>>> We can't just reference the original dbus user, so we duplicate

the

Quoted 1 line

>>> entry. What is important, is that the user be named 'dbus', as

that's

Quoted 3 lines

>>> what we use in both cases.
>>
>> If it's a virtual package, then the user could be created by

the virtual

Quoted 6 lines

>> package itself.

It won't be a virtual package, so no longer relevant.

> with systemd you probably dont even need a real user for the

isolation

Quoted 10 lines

> benefits.
> So maybe this would just be necessary if dbus utilities are used, > you can also just let systemd create that user dynamically:
>
> [Service]
> ...
> User=dbus
> Group=dbus > DynamicUser=true

If we do this both in the dbus and the dbus-broker units, then

indeed we can

Quoted 4 lines

remove the dbus user.

AFAIR the reason for the dbus to be a system user is that dbus-launch

can setuid

Quoted 9 lines
Quoted 2 lines

into it.
So it only works without system user if dbus-launch is not used.

Dbus-broker

Quoted 1 line

delegates this functionality to system (dbus activation)

Well, the service file still calls dbus-broker-launch, which I guess
does
setuid because the upstream service file doesn't have any User= or Group=

I meant Dbus-launch, dbus-broker-launch *does not use setuid* which depends on filesystem/inode permissions and uid.
If you prefix a ! In the service file dbus-broker-launch will be run as root, and since this is the only way the executable is used (unlike dbus) the dynamic user uid is enough.

Hence dbus-broker-launch works with random/dynamic uids, while dbus-launch needs a stable one.

Quoted 4 lines

I think we'll just stick to what we have now, and keep the upstream
service
file rather than hacking our own...

Quoted 27 lines

Regards,
Arnout

Quoted 3 lines

If I add this just to dbus-broker, but dbus still creates a

static user, is

Quoted 19 lines

this going to give a problem for systemd? I.e. does systemd support a DynamicUser=true if the user exists already statically? I think so...

Think so too.

Regards,
Arnout

> ExecStart=!/usr/bin/dbus-broker-launch --scope system --audit > ExecReload=!/usr/bin/busctl call org.freedesktop.DBus > /org/freedesktop/DBus org.freedesktop.DBus ReloadConfig

[snip]

Norbert

_______________________________________________
buildroot mailing list
[email protected]
https://lists.buildroot.org/mailman/listinfo/buildroot