Make methods to load_config and load/save state. For now the config
and state directories are the same but it might not be. Implement
the search path for all config/state files as:
$XDG_CONFIG_HOME/[$prefix]/$name
$HOME/.config/[$prefix]/$name
$PIPEWIRE_CONFIG_DIR/pipewire/[$prefix]/$name
/etc/pipewire/[$prefix]/$name
Make some config files for jack and RT clients. Make pw-cat use the
client-rt config.
Use core state and config management in media-session.
Move all session manager state and config files to the build dir and
set the PIPEWIRE_CONFIG_DIR to this build dir.
Use PIPEWIRE_PROPS to create and connect the context as well. This
makes it possible to pass configuration to the modules loaded by
the context such as:
PIPEWIRE_PROPS="context.modules.args={nice.level=-14}" jack_simple_client
To set the nice level of the jack app.
See #698
Make a port.extra property that contains API specific extra port
information that is passed in the global object properties.
Use this to pass the custom jack flags for a port. Carla uses this
to mark CV ports.
Refuse to run when we run in the daemon, we can't really do a blocking
connect to the daemon from its main thread.
Also refuse to run when we detect an old pipewire libaray. Blocks the
case where some app linked against 0.2 wants to use the replacement
libjack.so.
This is necessary on some 32-bit architectures that implement atomic
operations on 64-bit quantities as library calls, including Debian's
armel and mipsel ports.
Signed-off-by: Simon McVittie <smcv@debian.org>
Make sure we emit the buffer callback either from the calling thread
or from the thread_loop. Mark the callback as pending and don't call
the process function until the callback has been handled.
Only append app name and node name when they are different
Use / to separate app name from node name because : is used to
separate client name from port name.
Only emit callbacks when something changed.
When setting a value, first update the local store. This triggers
the emit from the calling thread (as expected by jack clients) and
when the update arrives from the server, it will not emit the callback
anymore because the value didn't change.
This is easier to handle in general and we should not use the name
as a unique id. If the session manager wants to save things, it can
use whatever fields it wants from the object to create a unique
persistent name.
Let the jack client make passive links when the PIPEWIRE_LINK_PASSIVE
variable is set. Makes it possible to start qsynth with passive links
so that it suspends when not in use.
This makes it easier to test PipeWire in its "as-installed" state,
for example in an OS distribution.
The .test metadata files in ${datadir}/installed-tests/${package} are
a convention taken from GNOME's installed-tests initiative, allowing a
generic test-runner like gnome-desktop-testing to discover and run tests
in an automatic way.
The installation path ${libexecdir}/installed-tests/${package} is also
a convention borrowed from GNOME's installed-tests initiative.
In addition to the automated tests, I've installed example executables
in the same place, for manual testing. They could be separated into
a different directory if desired, but they seem like they have more
similarities with the automated tests than differences: both are there
to test that PipeWire works correctly, and neither should be relied on
for production use. Some examples are installed in deeper subdirectories
to avoid name clashes.
Signed-off-by: Simon McVittie <smcv@debian.org>
On GNU/Linux systems, the literal string '${LIB}' in dynamic linker
paths expands to "lib", a biarch libQUAL directory such as "lib64", or
a Debian-style multiarch directory such as "lib/x86_64-linux-gnu".
If we're installing libraries to such a directory, and we have both
word-sizes' compatibility libraries available, then pw-pulse can
use LD_LIBRARY_PATH='/usr/${LIB}/pipewire-0.3/pulse' to make both
i386 and x86_64 programs load the correct version.
Signed-off-by: Simon McVittie <smcv@debian.org>
LD_LIBRARY_PATH="/path/to/lib:" is interpreted as equivalent to
LD_LIBRARY_PATH="/path/to/lib:.", loading libraries from the current
working directory, which could lead to malicious libraries being loaded
if the current working directory is untrusted. To avoid this, only add
the current LD_LIBRARY_PATH to the new LD_LIBRARY_PATH if it is
actually set.
This commit also single-quotes the interpolated @VARIABLES@ so that
their values can contain shell special characters (other than single
quotes).
Signed-off-by: Simon McVittie <smcv@debian.org>
Protect the global metadata with a lock because we update this from
multiple clients. Avoid updating the metadata if it didn't change.
Add a simple lock to protect the session objects, they could be
accessed from the main thread or data thread. Use the simple lock
in methods that just read.
Use the new data loop invoke to make sure we sync the data update
with the data thread.
Stop the data loop when our position io is removed.
Don't use a special name for the replacement libraries but install
them into the modules directory by default. Add an option to install
them into another location.
This way, we don't need to set up symlinks in development, distros can
choose to install them where they want and/or we can use symlinks or
LD_LIBRARY path to select the replacement versions.
Rework the metadata implementation without pw_properties to make
it easier to delete all subjects and implement the metadata API.
Remove metadata from all objects when they are destroyed.
Don't load and use rt-kit by default because it can cause sigkill in
the app, which is not good when it happens in a compositor.
Make the module profile a comma-separated list of profiles and make
it possible to explicitly load rtkit as well.
load the rtkit profile as well in jack.
JACK can handle 2 buffers at most, make the default buffer allocator
allocate 2 buffers when no params are given.
Prefer 2 buffers, it allows some form of async fill/consume
Don't unlink the mix->link when freeing, it is only linked when
in the free pool.
Protect against invalid number of buffers that could corrupt our
state.
Don't allocate the links in an array because they might be moved
when the array is resized. Instead just use calloc and add them
to a list.
Make sure we update the list used in the real-time thread from a
safe context.
Some plugins (calfjackhost) only call jack_port_get_buffer()
once and cache the pointer. This behaviour has been deprecated
in JACK2 but still works.
Add a workaround for this by getting a buffer ourselves and memcpy
into it from the scratch buffer.
Add a asprintf helper function that handles errors correctly.
Use this in places where we use asprintf to avoid warnings when we
don't check the return value.
Use the DSP media subtype to describe DSP formats. DSP formats
don't include the rate, channels and channel position in the
format and must use the rate and duration from the position io. This
makes it possible to later change the samplerate dynamically without
having to renegotiate the graph.
The same goes for the video DSP format, which uses the io_video_size
from the io_position to get the size/stride. Set this up in the node
based on the defaults from the context.
Make it possible to define defaults in the daemon config file, such
as samplerate, quantum, video size and framerate. This is stored in
the context and used for the DSP formats.
This is more in line with wayland and it allows us to create new
interfaces in modules without having to add anything to the type
enum. It also removes some lookups to map type_id to readable
name in debug.
Make the thread_loop alloc its own loop by default to simplify
some core. Add extra new_full method to pass a custom pw_loop.
Make other loop implementations ready to support custom loops
if we want that later.
The proxy API is the one that we would like to expose for applications
and the other API is used internally when implementing modules or
factories.
The current pw_core object is really a context for all objects so
name it that way. It also makes it possible to rename pw_core_proxy
to pw_proxy later.
The pw_remote object is really a wrapper around the pw_core_proxy.
The events it emits are also available in the core proxy and are
generally awkward to use.
With some clever new pw_core_proxy_* methods and a pw_core_connect
to create the core_proxy, we can convert all code away from pw_remote.
This is a first step in this conversion, using the pw_remote behind
the scenes. It leaks into some places because it really needs to become
its own struct in a next step.
Change node.priority to priority.session to indicate that this
is the priority that the session manager uses to select the node.
Add another priority.master that the core uses to select a master
driver. Keep the driver nodes sorted by master priority.
Let jack always prefer to connect to the master driver nodes.
Send suspend to the node when suspending. This is usually the same
as puse for all nodes.
Implement negotiation when we Start audioadapter. This makes it
easier that to track the ports that are negotiated for now.
Use Suspend to clear the audioadapter negotiation.
Reorganize some things, let the clients update the segment info
in their own activation, then let the server merge it. This avoids
clients stepping on eachother. When looping through the clients,
copy the segment info when we encounter its owner.
Remove the list of segment owners to the activation. This is better
than in the activation because we can then just keep one list of
owners.
Remove the NONBLOCK flag from the eventfd so that we can do blocking
reads as well.
Just keep a reposition owner in the driver activation. This points
to the node that has the reposition info. This avoid complicated
synchronization to keep multiple nodes from stepping on eachother.
Now they can just prepare the reposition info in their activation and
set themselves as the reposition owner. The last one who succeeds
wins.
When the node latency property is changed, trigger a graph recalc
to set the new quantum if needed.
Also update the driver quantum when unassigned nodes are assigned
to a driver.
This makes it easier to keep track of who is responsible for what.
Also remove the valid fields and move them to flags in the segment
info. That way, the owner can update the flags without having to
worry about concurrency.
Keep separate info for the reposition information. We need to do this
to make it possible to seek in other formats than the frame.
Clear out the owner field when the node is destroyed or removed from
the driver.
Place the requested sync and position update flag in the node
activation. This way we can use our existing loop to update the node
sync states and check if the node is ready.
Implement sync timeout, when the client can't start or seek within the
timeout, we start RUNNING anyway and hope the client catches up.
Sync is enabled when clients need time to move to a new location.
It's a bit like GStreamer preroll after a seek. Clients that need
time, increment the sync_total. Whenever a seek is done, the server
waits in the Starting state until the sync_pending is 0 (or timeout
later).
Improve atomic operations
Add an offset to apply to the clock time before we can compare to the
segment values. This way we can keep the segment start independent of the
clock values and we only need to adjust the offset when paused. It's
like the base_time in GStreamer to calculate the running time.
Move fields from the io_position to io_segment. The segment contains
the mapping between raw clock time and stream time in various
formats. We keep an array of pending segments available in the
io_position field so clients can anticipate changes.
Make looping a flag in the segment instead of a state.
Prepare for segment masters. These will be registered clients that
are responsible for updating parts of the extended segment info.
Add namespace to some defines.
Move some things around. Move the duration of the current cycle
to the clock. Also add the estimated next timeout to the clock.
Add a generic media specific counter to the clock.
Clean up the position_bar info. We can do with only a double beat
value and make the signature in floats.
Flesh out the io_position info. This has now the information needed
to convert a raw clock time into a stream time. It basically has
the same kind of features as GStreamer segments such as looping,
variable rate playback etc.. It also contains the state of the
timeline (paused/playing) and it can be used to update the position
and state from clients.
There is also extended information in the position field that
clients can update when they can.
Plugins basically only update the clock info they get (and use
the position info to check if they are slaved or not).
Before each cycle, check if there is a pending position update and
apply it.
Remove the parent_id from the global event. Remove the parent
and owner from the global object.
Use properties instead to mark parents and owners of objects.
Properties are easier to control for client exported objects and
usually a simple parent/child is not enough. For example, a client
exported node has the client as a parent but also the factory that
created the node.
Implement per channel volume on channelmix. Extend control on stream to
take an array of values when possible.
Remove name argument from pw_node_new and pw_device_new. We can pass
this as a property instead.
Improve properties on nodes to more closely match what pulseaudio does.
Don't let the monitor do too much with the udev properties but let the
session manager set the description and icon-names.
Remove some change_mask flags for things that don't change in
introspect. Use the flags to mark changes in -cli and -monitor.
Remove the CAN_USE_BUFFERS flag, it is redundant. We can know this
because of the IO params and buffer params.
Add flags to the port_use_buffer call. We also want this call to
replace port_alloc_buffer. Together with a new result event we can
ask the node to (a)synchronously fill up the buffer data for us. This
is part of a plan to let remote nodes provide buffer data.
Add a tag field when creating a memmap so that we can do lookup on it.
This makes it easier to implement the tracking of mappings for io areas.
Remove custom io memory tracking and use the tags.
Add flags to spa_chunk to make data corrupted. The flags on the buffer
stay constant for the life of the buffer. Add flags to mark memory
readable and writable. Mark memory readonly in v4l2-source.
Pass the daemon activation area to the client in the transport event.
This never changes and need to be handled differently from the other
activation areas.
Use the right flags when importing memory.
Add the (desired) memory type to mempool_alloc.
improve some debug.
Add a memory pool to manage blocks of memory. Make it possible
to allocate and import blocks.
Add add_mem and remove_mem to the core events to signal a client
of a block of memory. Remove the client-node add_mem.
Make a global pool for memory and a per client pool where we
import and share the memory we need with the client.
Use the new memory pool to track and map memory in clients.
Move the epoll functions to the system functions and make the loop
use those. Use simple mask for events instead of enum.
Add the used system api in pw_loop.
Add System API to spa_support and use it where possible.
Pass the system API used in the realtime loops in spa_support as
well and use this in the realtime paths.
Improve bootstrapping, load only the log and cpu interfaces because
those can/need to be shared between instances. Let the core load
the other interfaces.
Add keys to configure the System and Loop implementations used in
pw_loop.
The interface struct has the type,version and methods of the
interface.
Make spa interfaces extend from spa_interface and make a
separate structure for the methods.
Pass a generic void* as the first argument of methods, like
we don in PipeWire.
Bundle the methods + implementation in a versioned inteface
and use that to invoke methods. This way we can do version
checks on the methods.
Make resource and proxy interfaces that we can can call. We
can then make the core interfaces independent on proxy/resource and
hide them in the lower layers.
Add add_listener method to methods of core interfaces, just
like SPA.
Simplify the scheduling by using simple lists and removing the
subgraphs etc..
Make the driver node trigger all nodes it manages and when they
complete, trigger the driver node to finish the graph.
Pass the complete spa_node_info to update node information. Remove
the redundant max/min port info.
Update the remote node based on port and node update events.
Add a new struct spa_param_info that lists the available params on
a node/port and if they are readable/writable/updated. We can use
this to replace and improve the PARAM_List and also to notify
property change and updates.
Update elements and code to deal with this new param stuff. Add
port and node info to most elements and signal changes.
Use async enum_params in -inspect and use the param info to know
which ones to enumerate.
Use the port info to know what parameters to update in the
remote-node.
Make the bind function a callback instead of an event. We can then
get a return value and use that to clean up the pending proxy and
generate an error.
Don't use special callback in node to receive the results. Instead,
use a generic result callback to receive the result. This makes things
a bit more symetric and generic again because then you can choose how
to match the result to the request and you have a generic way to handle
both the sync and async case. We can then also remove the wait method.
This also makes the remote interface and spa interface to objects very
similar.
Make a helper object to receive and dispatch results. Use this in the
helper for enum_params.
Make device use the same result callbacks.
Remove the done and error callbacks. The error callback is in an
error message. The done callback is replace with spa_pending.
Make enum_params take a callback and data for the results. This allows
us to push the results one after another to the app and avoids ownership
issues of the passed data. We can then extend this to handle the async
case by doing a _wait call with a spa_pending+callback+data that will
be called when the _enum_params returns and async result.
Add a sync method.
All methods can now return SPA_RESULT_IS_ASYNC return values and you
can use spa_node_wait() to register a callback when they complete
with optional extra parameters. This makes it easier to sync and
handle the reply.
Make helper methods to simulate the sync enum_params behaviour for
sync nodes.
Let the transport generate the sequence number for pw_resource_sync()
and pw_proxy_sync(). That way we don't need to keep track of numbers
ourselves and we can match the reply to the request easily.
