They are emited from the streaming thread and therefore can be emitted
concurrently with the events on the main thread. This can cause crashes
when the hook list is iterated.
Instead, make those events into callbacks that are more efficient,
and threadsafe.
Add a control.ump port property. When true, the port wants UMP and the
mixer will convert to it. When false, the port supports both UMP and
Midi1 and no conversions will happen. When unset, the mixer will always
convert UMP to midi1.
Remove the CONTROL_types property from the filter. This causes problems
because this is the format negotiated with peers, which might not
support the types but can still be linked because the mixer will
convert.
The control.ump port property is supposed to be a temporary fix until we
can negotiate the mixer ports properly with the CONTROL_types.
Remove UMP handling from bluetooth midi, just use the raw Midi1 events
now that the mixer will give those and we are supposed to output our
unconverted format.
Fix midi events in-place in netjack because we can.
Update docs and pw-mididump to note that we are back to midi1 as the
default format.
With this, most of the midi<->UMP conversion should be gone again and we
should be able to avoid conversion problems in ALSA and PipeWire.
Fixes#5183
Avoid doing conversions in the nodes between Midi formats, just assume
the imput is what we expect and output what we naturally produce.
For ALSA this means we produce and consume Midi1 or Midi2 depending on the
configurtation.
All of the other modules (ffado, RTP, netjack and VBAN) really only
produce and consume MIDI1.
Set the default MIDI format to MIDI1 in ALSA.
Whith this change, almost everything now produces and consumes MIDI1
again (previously the buffer format was forced to MIDI2).
The problem is that MIDI2 to and from MIDI1 conversion has problems in
some cases in PipeWire and ALSA and breaks compatibility with some
hardware.
The idea is to let elements produce their prefered format and that the
control mixer also negotiates and converts to the node prefered format.
There is then a mix of MIDI2 and MIDI1 on ports but with the control
port adapting, this should not be a problem.
There is one remaining problem to make this work, the port format is
taken from the node port and not the mixer port, which would then expose
the prefered format on the port and force negotiation to it with the
peer instead of in the mixer.
See #5183
Don't accept absolute library paths that are not in the search path,
skip the ../ in paths to avoid opening arbitrary libraries from
unexpected places.
This reverts commit bb0efd777f.
It is unclear what the problem was before this commit. If there are any
pending operations, the suspend should simply cancel them.
See #5207
When the driver changes, the clock position can also change and there
would be a discont in the rtp_timestamp.
This is not usually a problem except in RAOP mode where the base rtp
timestamp is negotiated and anything that deviates too much is to be
discarded.
If we are not using direct_timestamp for the sender, make sure we always
keep the rtp_time aligned to avoid this problem.
See #5167
Don't close an -1 fd in clear_data.
If we let the client allocate buffer, set our fd and data to invalid
values. If the client decides to renegotiate before we get the buffer
data we might otherwise try to clear the mem_id (default 0) or
close the fd (also default 0).
Fixes#5162
We can only use non-shared memory when both nodes live in the same
process _and_ we can be sure the output port is never going to be linked
to a remote node because it is exclusive.
This fixes the case where a null-sink is loaded inside the process space
of the server and linked to the ALSA sink. This would create a link
without shared mem and then as soon as something else (out of process)
wants to link to the null-sink output, it would get a -22 EINVAL
negotiation error because the memory can't be shared.
Fixes#5159
We keep a mapping between the sndfile formats and the format we would
like to decode them to for encoded formats. Make sure we don't mix up
the sample widths between them.
Make sure we don't send encoded formats as raw.
Debug the uncompressed format name correctly.
Fixes#5155
Sink/Source pairs should not have the same link-group otherwise the
session manager will not be able to autoconnect them with a loopback or
some other internally linked stream.
Setting the current clock time when resending buffers is often wrong.
Especially for pseudo-live sources - the default mode - it discards
the original buffer time, which again is used by the base-class to
adjust the timestamps further, ultimately resulting in very wrong
timestamps.
Instead, try to calculate the delta between when we originally got the
buffer and now.
(cherry picked from commit efd1526423)
Buffer timestamps get adjusted by the base class, GstBaseSrc, even if we
take an additional ref. Arguably the base class should check if buffers
are writable (gst_buffer_make_writable()), which would trigger a buffer
copy. That is currently not the case, though, thus do so on our side.
Notes:
1. Usually a buffer copy doesn't copy the underlying memory, i.e.
copying is cheap.
2. The copy holds a reference to the copied buffer, preventing the
buffer from getting recycled as before.
(cherry picked from commit 49300d8ee0)
Fix path comparison in is_socket_unix() and don't unset LISTEN_FDS since
the function that uses it is called more than once and it was not unset
when sd_listen_fds() was used.
Fixes#5140
Roc-toolkit log records are captured via a callback and
written to PipeWire log with corresponding verbosity level.
The log.level config parameter limits record verbosity at
the roc-toolkit level.
Patch by Lairton Lelis da Fonseca Junior (@lairton)
Remove the hard skip for IPv4 link-local addresses and add an interface
binding (matching the existing IPv6 link-local behavior).
The host needs a link-local address on the interface (ip addr add
169.254.x.x/16 dev wlan0 or via NetworkManager +ipv4.addresses).
Fixes#4830
The stream should be streaming before the get_time call is meaningful.
Various places in the code already check this an fall back to a default
value, we just need to return an error here.
Add a container option to override the extension check and force a
container when saving.
Add some more formats that are supported by libsndfile.
Add some options to list all supported formats, extensions/containers,
layouts and channel names.
Fixes#5117
A call to `release_buffer()` may happen in a gstreamer thread concurrently
with the pipewire stream emitting the `remove_buffer` event in the thread
loop, which, in pipewiresink calls `gst_pipewire_pool_remove_buffer()`, which
in turn modifies the `GstPipeWirePoolData` object. Thus a data race occurs
when accessing its members, which can lead to `pw_stream_return_buffer()`
being called with a null pointer.
Fix that by locking the thread loop before checking the conditions.
Fixes: c0a6a7ea32 ("gst: handle flush event in pipewiresink")
Socket activation uses sd_listen_fds from libsystemd, and can only be
compiled on systems with systemd.
This is an issue for Alpine / postmarketOS, where upstream has no
systemd package, but downstream depends on upstream's pipewire package
and wants to rely on socket activation. This also prevents using
socket-activation on other non-systemd distributions, including
non-Linux.
Implement equivalent functionality without a dependency on libsystemd.
This can easily be overlooked if the RTP rate equals the clock rate, which
is fairly common (for example, RTP rate and clock rate both being 48 kHz).
And, if an ASRC is active, and converting the output of the RTP source
node, the resampler's delay need to be taken into the account as well.
Clear the ringbuffer in stream_stop() when processing stops to prevent old invalid packets
from being sent when processing resumes via rtp_audio_flush_packets().
This ensures a clean state when the stream restarts.
Clearing the ring buffer is important not only in the direct timestamp
mode, but also in the constant delay mode, since missed packets can lead
to gaps in the ring buffer. These gaps may have stale data inside if the
ringbuffer is not cleared after reading from it.
In corner cases where the read and write pointers are very close, it may
not be possible to read out all the wanted samples. This can for example
happen when there is a jump in the graph driver position. Currently, the
code reads the wanted number of samples out of the ring buffer regardless
of the write and read pointer positions. It does so even when the read
pointer is ahead of the write pointer (that is, an underrun occurs).
Fix this by checking the fill level and reading only the available amount
of samples if that amount is less than the wanted amount. The remaining
space in the target buffer is then filled with nullbytes.
This reverts commit dcdc19238b.
Reverting this because it caused big sync errors of ~62 ms in test setups.
Further discussions about this can be found here:
https://gitlab.freedesktop.org/pipewire/pipewire/-/merge_requests/2666
Followup commits modify the device delay application (by scaling it),
which is another reason why this needs to be reverted.
Add 35 sec timeout for PLAY_SAMPLE streams to start streaming, similar
to what we do with normal streams, and fail playback if they don't
start.
This avoids pending sample playback using up resources indefinitely if
streams fail to start for some reason, e.g. session manager is not
linking them.
If we get EPROTO, we likely have missed on some messages from the
server, and our state is now out of sync.
It's likely we can't recover (e.g. if error is due to fd limit hit), so
just drop the server connection in this case, similarly as if we got
EPIPE.
The simple formats contain some common mappings for other extensions such
as mp3.
Makes pw-record test.mp3 actually write an mp3 instead of a wav file.
Since commit a1f33a99df changed buffer handling to create new GstBuffers
instead of reusing pool buffers, the video crop metadata was silently
lost. The code used gst_buffer_get_video_crop_meta() which returns NULL
on a fresh buffer, so the crop values from PipeWire were never applied.
Change to gst_buffer_add_video_crop_meta() to actually attach the
metadata to the buffer.
Also remove the now-obsolete call in gst_pipewire_pool_wrap_buffer.
This was discovered when using the XDG Desktop Portal's RemoteDesktop
interface: the full desktop was being delivered instead of just the
selected window, because the crop region metadata was not being
propagated to the GStreamer buffer.
Fixes: a1f33a99df ("gst: dequeue a shared buffer instead of original
pool buffer"), from merge request !1258
CC: @jameshilliard
