If we are already in the loop thread and flushing, this means we
added a new invoke item on the list from a callback. Place the
item on the queue and let the flush code take care of it after the
callback completes.
Required to fix some issues with draining in pulse where a stream
is destroyed from the drained callback which then invokes a pause.
Because the signal can't be removed from the callback we can
simply iterate backwards and then forwards.
The first added hook (the unlock/lock pair) is called last before
going into the poll and first when leaving. This executes all other
callbacks inside a locked situation. And removing them with the lock
is not going to cause problems.
Use a safer version of the before and after hooks. First call
all before hooks and save them in reverse order in a save list.
Then call the after event for the ones remaining in the save list
and move them back to the hook list.
This makes it possible to remove the hooks from one the callbacks or
even from other threads with the right locks. Found as a solution to
the following problem as observed in vlc:
main thread thread_loop
pw_thread_loop_lock() before hook: lock suspend thread
pw_context_destroy()
- removes before hook to flush clients
pw_thread_loop_unlock()
before hook: lock acquired, resume
before hook: flush client hook executed
*crash*
pw_thread_loop_stop()
pw_thread_loop_destroy()
Any of the safer cursor methods (like spa_hook_list_call()) would also
work but are more expensive and don't reverse the before/after
order.
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.
Timestamps have usec precision and the seconds are limited
to 9 digits. Usually what matters in these messages is to spot
delays between printouts and not really what is the absolute
time of the system.
FreeBSD doesn't provide timerfd and eventfd functions. These are implemented in
3rd party library called epoll-shim. Link targets requiring these functions to
this library.
When we do invoke from the loop thread, first process the pending
invoke items before handling ours. This can be used to sync the
invoke queue before cleanup.
Add system functions that can execute on the evenless Real-Time
subsystem. https://evlproject.org/
Enable this plugin as the system library for the data thread and
all poll/timerfd/eventfd will execute on the evl subsystem.
Define a set of standard factory names and document what they
contain. This makes it possible to change the implementation by
mapping the factory-name to a different shared library.
Make a set-prop command to set a property from the config file
into a pw_properties. Pass this to the pw_core_new() and the
main-loop to tweak some stuff.
Move some warns to errors
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.
