not all functions in glib returns a value for the used log fields
in this case a 'g_return_val_if_fail(..) will only set 4 fields an
set the other field to NULL which leads to a segfault
Jul 22 13:41:37 Cynap-Pro-01121508 systemd-coredump[7247]: Process 496 (wireplumber) of user 4242 dumped core.
Stack trace of thread 496:
#0 0x00007f6b2ce58e99 __strlen_avx2 (libc.so.6 + 0x150e99)
#1 0x00007f6b2cf3ebe5 g_log_writer_journald (libglib-2.0.so.0 + 0x5fbe5)
#2 0x00007f6b2d0a9316 wp_log_fields_write_to_journal (libwireplumber-0.5.so.0 + 0x2b316)
#3 0x00007f6b2d0a96f5 wp_log_writer_default (libwireplumber-0.5.so.0 + 0x2b6f5)
#4 0x00007f6b2cf3cf3e g_log_structured_array (libglib-2.0.so.0 + 0x5df3e)
#5 0x00007f6b2cf3d0ae g_log_structured_array (libglib-2.0.so.0 + 0x5e0ae)
#6 0x00007f6b2cf3d4b7 g_logv (libglib-2.0.so.0 + 0x5e4b7)
#7 0x00007f6b2cf3d7ef g_log (libglib-2.0.so.0 + 0x5e7ef)
#8 0x00007f6b29efafef get_node_by_id (libwireplumber-module-cb-ipc.so + 0x24fef)
#9 0x00007f6b29efb032 wp_cb_ipc_get_app_properties (libwireplumber-module-cb-ipc.so + 0x25032)
A profile is a list of features set to required/optional/disabled
which governs which components are getting loaded, given a static
components list with well-defined dependencies
These tests are meant to catch wireplumber issues, so the libpipewire
logs are not very interesting and just make it hard to go through
the interesting bits
Each component can now list required and optional dependencies,
using the component feature names to match other components.
In addition, each component feature can be declared as required, optional
or disabled, making optional components easier to deal with.
The component flags (ifexists, nofail) have been removed.
Using virtual components, this system also allows easier customization
of which components should be loaded for a specific configuration,
without requiring the user to copy the list of components and edit it.
Also bump the required glib version to 2.68 for g_assert_cmpstrv()
This allows registering arbitrary objects on the core's registry and
finding them later, without having to add API for each and every object.
I think this is useful enough to have it public, even though it's
probably not going to be used that much... The rationale here is to
allow registering custom component loaders without having to make them
subclass WpPlugin or to create custom API for registering component
loaders specifically.
Also, remove the wp_plugin_register() and wp_si_factory_register()
functions, since they are not going to be used much in the future.
The idea is to let the component loader do the registration under the
scenes, as the component is getting loaded.
Each component can optionally "provide" a feature, which is basically
a string that describes the feature (ex. "support.dbus"). If the
component loads successfully, the feature is marked as provided and
can be tested for its presence with wp_core_test_feature()
Now that we have proper module load order, we can have this shared
dbus connection in a module instead of the library. The module has
to be loaded before any other modules that need it, obviously.
Some special characters like '\v' are encoded using 6 characters, which
currently does not work because the VLA size asumes a maximum of 4 characters
per encoded special character. This patch refactors this logic to avoid using
VLAs at all and encodes the string directly into the builder.
See #471
Similar to wp_spa_json_builder_add_property(), we need to make sure the dst
array in wp_spa_json_builder_add_string() has room for the null character
because builder_add() expects it.
Fixes#471
Disable bluez5.auto-connect.
It makes PW to try connect to all locally known BT audio devices on
start, which is in general not a good thing to do.
Some BT adapters are not capable of simultaneous audio streaming and
scanning for connections (e.g Raspberry Pi builtin BT, probably more),
so audio stutters until adapter/kernel give up trying to connect to
devices that cannot be reached.
Disabling it makes devices to not reconnect on user VT switch, but also
that we should only do for devices that were connected when logind
switched.
After a timeout, it's possible that something else may have
destroyed the node already. This can happen for example with virtual
sinks that were created from another process.
Fixes: #467
libcamera uses namespaces to sandbox IPA (Image Processing Algorithm)
modules because they are sometimes proprietary binary blobs. The
RestrictNamespaces option in Wireplumber's systemd service breaks this
sandboxing when libcamera is loaded via the libcamera SPA module, so
cameras requiring an IPA do not work.
This commit removes RestrictNamespaces so that the sandboxing works
again. I've confirmed that after this change wireplumber works with
libcamera with an IPA module.
Resolves#466
There is no reason to return the component object... all components
are supposed to be long-lived objects that are referenced by the
registry and there is API to find them. The caller is only interested
in the success or failure of the operation.
Do this in wp_core_load_component() and let the component loaders worry
only about creating the object.
Also run the main loop in tests while loading components, to ensure
that the async operation finishes before continuing execution. GTask
makes sure to make the operation async always, by emitting the callback
from an idle GSource.
Regarding the core parameter, the case used to be that WpComponentLoader
was a WpPlugin, so it had a reference to the core internally, but since
this is no longer a requirement, we need to pass this explicitly
