|
Number of commits found: 7
| Commit History - (may be incomplete: for full details, see links to repositories near top of page) |
| Commit | Credits | Log message |
|---|
0.9.6
13 Jul 2026 05:39:53
     |
Kevin Bowling (kbowling)  |
audio/pipewire-spa-oss-ng: Update to 0.9.6
pin EnumFormat stereo last so Pulse volume width stays stable after idle
(device close and reopen shows volume slider on Plasma6). |
0.9.5
08 Jul 2026 05:55:40
|
Kevin Bowling (kbowling) |
audio/pipewire-spa-oss-ng: Update to 0.9.5
Retune in place across graph quantum changes |
0.9.4
05 Jul 2026 07:20:47
|
Kevin Bowling (kbowling) |
audio/pipewire-spa-oss-ng: Update to 0.9.4
Capture (source) fixes for cramped rings and fat strides
Fixes a field report of skipping audio when recording and playing back
at once on a fat-stride USB capture device (20-channel S32, 80
bytes/frame): FreeBSD clamps the soft ring at CHN_2NDBUFMAXSIZE, so it
holds only ~1.6 graph periods there, and the capture path had no
headroom for arrival jitter. |
0.9.3
04 Jul 2026 08:38:54
|
Kevin Bowling (kbowling) |
audio/pipewire-spa-oss-ng: Update to 0.9.3
Fixes a field report of periodic drops with crackles on USB audio: the
sink's near-full fill branch keyed on small driver-forced fragments as a
proxy for a small total buffer, which uaudio breaks (1 ms soft fragments,
large grant) - near-full targeting added 100+ ms of latency and left one
fragment of overshoot headroom, so every normal servo excursion
short-wrote and dropped the tail.
Validated by a discrete-event simulation of the servo (exact DLL and
bandwidth-adaptation ports) that reproduces the field failure with the
old target and shows zero drops with the new one across +/-2000 ppm
drift, jitter storms and 4-hour runs, plus the live battery on hdspe(4). |
0.9.2
03 Jul 2026 22:32:48
|
Kevin Bowling (kbowling) |
audio/pipewire-spa-oss-ng: Update to 0.9.2
robustness fixes, driver-clock overhaul, hardware volume control |
0.9.1
02 Jul 2026 11:43:33
|
Kevin Bowling (kbowling) |
audio/pipewire-spa-oss-ng: Update to 0.9.1
Threading / robustness:
- Serialize all reconfiguration onto the data loop and stop the driver
timer cleanly on pause/suspend (closes the renegotiation window behind
a PipeWire audioconvert crash).
- Detach loop sources and close the timerfd on clear().
- Fail negotiation instead of aborting when the device won't open or
rejects the format/channels/rate; validate peer-provided formats and
buffers; return errors from all unhandled SPA entry points.
- Run without devd (hotplug disabled instead of aborting).
Audio / OSS:
- Probe device caps and advertise real formats, rates and channel
layouts (mono through 7.1 in FreeBSD kernel interleave order, plus (Only the first 15 lines of the commit message are shown above  ) |
0.9.0
26 Jun 2026 07:44:23
|
Kevin Bowling (kbowling) |
audio/pipewire-spa-oss-ng: New port
PipeWire SPA plugin implementing a FreeBSD OSS backend, tracking the
pw-oss fork at https://github.com/kev009/pw-oss.
Changes over upstream shkhln/pw-oss:
Audio / OSS:
- Size the OSS buffer to the device and fix data-loop aborts.
- Adapt the fill target to the device's real fragment size.
- Report clock delay and rate; require MemPtr buffers.
- Reset the channel before close so close() doesn't block draining.
- Close the device when the format is cleared.
- Advertise and re-emit the port Format param (ALSA-sink pattern).
- Don't abort the process on an oversized chunk or a bad/unpositioned (Only the first 15 lines of the commit message are shown above ) |
Number of commits found: 7
|
As an Amazon Associate I earn from qualifying purchases.
