Get the support for newer gupnp 1.2
Fixes
| configure: error: Package requirements (gupnp-1.2 >= 1.2.0) were not met:
|
| No package 'gupnp-1.2' found
Signed-off-by: Khem Raj <raj.khem@gmail.com>
- CMakeList.txt patch deleted because added to the new version.
Changelog:
* Miscellaneous bugs fixed.
* PCM plugin support.
* Add CMake build support.
* Add meson build support.
* tinyplay can now read from stdin.
* Improved versioning support for library.
* Improvements to pcm actions (prepare at open time and after overrun, etc.)
* Improvements/fixes to pcm_get_htimestamp().
* Fixes for the mixer percent functions.
Signed-off-by: alperak <alperyasinak1@gmail.com>
Signed-off-by: Khem Raj <raj.khem@gmail.com>
Changelog:
0.1.6
- Fixed an infinite loop when decoding some AMR-NB samples
- Fixed noise spikes when decoding non-voice frames for both AMR-NB and AMR-WB
0.1.5
- Fix an autotools issue with cross compiling from the 0.1.4 release
0.1.4
- Autotools cleanups
- Fixes for SID/DTX in the AMR-WB decoder, fixes for handling of bad
frames in both AMR-WB and AMR-NB
Signed-off-by: alperak <alperyasinak1@gmail.com>
Signed-off-by: Khem Raj <raj.khem@gmail.com>
Patches deleted because issues fixed in the new version.
Changelog:
ver 0.23.14 (2023/10/08)
* decoder
- flac: fix scanning files with non-ASCII names on Windows
- mad: fix calculation of LAME peak values
* mixer
- wasapi: fix problem setting volume
* more libfmt 10 fixes
* fix auto-detected systemd unit directory
* Android
- require Android 7 or newer
ver 0.23.13 (2023/05/22)
* input
- curl: fix busy loop after connection failed
- curl: hide "404" log messages for non-existent ".mpdignore" files
* archive
- zzip: fix crash bug
* database
- simple: reveal hidden songs after deleting containing CUE
* decoder
- ffmpeg: reorder to a lower priority than "gme"
- gme: require GME 0.6 or later
* output
- pipewire: fix corruption bug due to missing lock
* Linux
- shut down if parent process dies in --no-daemon mode
- determine systemd unit directories via pkg-config
* support libfmt 10
Signed-off-by: alperak <alperyasinak1@gmail.com>
Signed-off-by: Khem Raj <raj.khem@gmail.com>
Changelog:
- security fixes
- fixed generation of pkgconfig file with autotools
- fixed crashes recently detected by fuzzing
- read HDR parameters from VUI header
- fix reading matrix_coeffs from VUI header
- build fixes for vcpkg
- add API to read VUI color information from bitstream
- fix race in static mutex initialization
- fix missing headers to install in CMakeLists.txt
Signed-off-by: alperak <alperyasinak1@gmail.com>
Signed-off-by: Khem Raj <raj.khem@gmail.com>
Changelog
-Make mbedtls hmac less restrictive in line with RFC and library capabilities
-Allow overriding stream list implementation
-update min openssl version to 1.1.0
-Openssl 3 support for EVP MAC
-Add android build to CI
-add CIFuzz Github action to workflows
-Add policy_set_aes_cm_192_xxx symbols to def file
-Use current directions in stream update test
-Call `set_aad` and `get_tag` in AEAD performance tests
-Create alias for srtp2 as libSRTP::srtp2
-Remove compatibility code for legacy OpenSSL to fix LibreSSL build
-Export CMake Targets
-Avoid initializing local char table when converting nibble to hex
-EVP_CIPHER_CTX_cleanup() is deprecated in OpenSSL 1.1.0
-Correct null cipher key sizes and be more defensive
-Include directory should point to 'include' not to 'include/srtp2'
-Count blocks instead of bytes in AES-ICM limit computation
-Rtp decoder support ssrc and roc
-Use a full-length key even with null ciphers
-Fix set ROC functionality with gcm
Signed-off-by: alperak <alperyasinak1@gmail.com>
Signed-off-by: Khem Raj <raj.khem@gmail.com>
Switch to using libsoup-3 and meson for build system
Disable media-export plugin, since it fails to build
rygel-media-export-dvd-parser.vala:91.27-91.39: error: The name `uri_copy' does not exist in the context of `Soup' (libsoup-3.0)
| 91 | var new_uri = Soup.uri_copy(file_uri,
| | ^~~~~~~~~~~~~
| libsoup-3.0.vapi:4.1-4.14: note: `Soup' was declared here
| 4 | namespace Soup {
| | ^~~~~~~~~~~~~~
Signed-off-by: Khem Raj <raj.khem@gmail.com>
This fixes:
| Run-time dependency gi-docgen found: YES 2023.1
| Program gi-docgen found: NO
|
| ../gssdp-1.4.0.1/doc/meson.build:4:11: ERROR: Program 'gi-docgen' not found or
not executable
Signed-off-by: Markus Volk <f_l_k@t-online.de>
Signed-off-by: Khem Raj <raj.khem@gmail.com>
No longer used in generating packages
Also creates a possible confusion with the recipe maintainer
name.
Signed-off-by: Markus Fuchs <mklntf@gmail.com>
Signed-off-by: Khem Raj <raj.khem@gmail.com>
The "update" in commit 60726d77d5 added a SRCREV rather than updating
SRCREV_base. Change the recipe to actually use SRCREV so that it is
compatible with future AUH updates.
Also specify the version in the recipe file name, and correct somw
whitespace.
Signed-off-by: Peter Kjellerstedt <peter.kjellerstedt@axis.com>
Signed-off-by: Khem Raj <raj.khem@gmail.com>
The code fails to compile.
/mnt/b/yoe/master/build/tmp/work/cortexa15t2hf-neon-yoe-linux-gnueabi/aom/3.7.0/git/aom_dsp/arm/highbd_subpel_variance_neon.c:257:1: error: call to undeclared function 'aom_highbd_8_variance16x16_neon'; ISO C99 and later do not support implicit function declarations [-Wimplicit-function-declaration]
Signed-off-by: Khem Raj <raj.khem@gmail.com>
PipeWire 0.3.84 (2023-11-02)
This is the fourth 1.0 release candidate that is API and ABI compatible
with previous 0.3.x releases.
Highlights
- Fix a regression with openal because the queued buffers in the stream
were not reported correctly.
- Fix a bug in port busy counters that could cause random silent links.
- Fix a regression in echo-cancel because it was not reporting its
streams as ASYNC.
- Fix a JACK regression where not all ports were enumerated in all cases.
- Many more fixes and improvements.
PipeWire
- pw_stream now reports the queued buffers more accurately. This fixes
a regression when using openal. (#3592)
- The port busy counters were not updated correctly in some cases. This
could lead to negotiation errors and silent links. (#3547)
- Ignore latency maximum when forcing rate/quantum. (#3613)
- Nodes can now be added to multiple groups and link-groups. (#3612)
Modules
- The filter-chain now also handles notify port dependencies
correctly. (#3596)
- Filter-chain has support for new linear, clamp, recip, exp, log, mult,
sine builtin plugins.
- The echo-cancel module now correctly reports its playback and capture
streams as ASYNC to avoid running out of buffers. (#3593)
- It is now possible to specify an array of remote names to connect to
with the native protocol.
- module-rtp-sap and module-rtp-sink now try to bind to the specified
interface.
SPA
- The alsa plugin now removes the runtime properties such as period-num,
period-size and max-latency when suspended. (#3613)
Bluetooth
- BAP Locations/Context is now set on endpoints as required by new bluez.
- Improve selection of BAP leader.
JACK
- Add a jack_set_sample_rate() extension function.
- Make sure we get the info of all nodes/ports before completing the
jack_client_open() operation so that we can enumerate the ports
correctly in all cases. (#3618)
GStreamer
- Fix types of metadata in pipewiresink.
- Also copy metadata in buffers in all cases.
- Fix size allocation in bufferpool for compressed formats.
- Don't stop streaming thread when unlinked. (#3620)
ALSA
- The ALSA plugin now handles NULL values from mmap_areas. (#3600)
Signed-off-by: Markus Volk <f_l_k@t-online.de>
Signed-off-by: Khem Raj <raj.khem@gmail.com>
PipeWire 0.3.83 (2023-10-19)
This is the third 1.0 release candidate that is API and ABI compatible
with previous 0.3.x releases.
Highlights
- A quantum change regression was fixed.
- Use a 2 socket server now for the manager and the applications
with (when wireplumber is updated) different permissions.
- Reduce memory usage a little in audioconvert and use fewer buffers.
- Some JACK deadlocks were fixed.
- More bugfixes and improvements.
PipeWire
- Fix quantum change regression. (#3574)
- Use a 2 socket server by default. One for the session-manager and one
for applications.
- Fix a potential use-after-free in node and device cleanup. (#3588)
modules
- Some hardcoded buffer size limits were removed.
- Fix ASYNC flag on combined-streams.
- Add support for on-demand combined-streams using metadata.
SPA
- alsa-udev will now ignore PCMs with the ACP_IGNORE udev environment
variable. (#3570)
- The audioadapter now uses at least 2 buffers when the follower is
async.
- The number of buffers used by plugins was tweaked a little. Most
plugins now only ask 1 buffer.
- Memory usage in audioconvert was reduced.
- Fix some unaligned reads and writes and undefined left shifts reported
by ASAN. (#3572)
- Rework vulkan dependency checking.
- Don't try to link ALSA devices when prepare fails. This fixes some
crashes.
- Fix a stall when the allowed codecs are changed in ALSA.
- Improve ALSA rate control for sources to avoid xruns. (#3584)
- Try to fix IEC958 TrueHD and DTS playback. (#2284)
Bluetooth
- Improve fallback SCO mtu when the kernel doesn't tell us.
JACK
- The fixed buffer size limit was removed.
- Add an option to make input buffers writable (default true).
- A potential deadlock was fixed when applications lock the process
function. (#3585)
- Use a separate thread to dispatch notifications to avoid deadlocks.
(#3585)
- Potentially fix silent export in ardour in some cases. (#3514)
Older versions:
PipeWire 0.3.82 (2023-10-13)
This is the second 1.0 release candidate that is API and ABI compatible
with previous 0.3.x releases.
Highlights
- Fix a regression in some devices when the Pro-Audio profile was selected.
Only enable the IRQ based scheduling and device linking in specific
safe cases. (#3556)
- Improve rate switching. In some cases the graph rate would not switch
correctly. (#2929)
- Fix regression in alsa wakeups that would cause silence in VMs.
- Fix a leak in the SBC codecs for SCO.
- More improvements to the RAOP module.
- Other small improvements and fixes.
PipeWire
- Improve client property checks.
- Allow non-power-of-2 quantums when forced.
- Improve rate switching. In some cases the graph rate would not switch
correctly. (#2929)
- The PIPEWIRE_QUANTUM env variable now forces the size and rate in the
graph for the duration of the application. The softer PIPEWIRE_LATENCY
and PIPEWIRE_RATE can still be used to merely suggest a maximum latency
and a rate.
modules
- Remove the RTSP FLUSH request in RAOP because it does not seem necessary.
- The RAOP module now uses the common RTP stream functions.
- Add sockets option to protocol-native to make pipewire listen on multiple
sockets.
SPA
- Clean up some of the log functions.
- Add an option in ALSA to disable linking devices together.
- Only link pcms together when 1 capture and 1 playback pcm. For more complex
devices we can't be sure which ones can be linked. (#3556)
- disable tsched only when using linked devices.
- Add some extra checks in ALSA to avoid segfaults. (#3554)
- Add Tag support to alsa-sink and alsa-source.
- Use dynamic pod builder when we can.
- Set priority.driver on midi-bridge to allow it as a fallback driver. (#3562)
- Fix regression in alsa wakeups. (#3565)
- The PTP clock can now be found from the interface in node-driver.
pulse-server
- Some small cleanups and internal improvements.
- Add some memory debugging messages.
- Add Tag messages to streams.
Bluetooth
- Fix a leak in the SBC codecs for SCO.
JACK
- Patch up midi events in the destination buffer instead of writing to the
source buffer. (#3580)
- Group all jack clients together to avoid transport issues. (#3562)
ALSA-plugins
- Add also.deny option to block alsa clients from opening the PCM.
Signed-off-by: Markus Volk <f_l_k@t-online.de>
Signed-off-by: Khem Raj <raj.khem@gmail.com>
Additions:
- A new "DSP policy" module has been added; its purpose is to automatically
load a filter-chain when a certain hardware device is present, so that
audio always goes through this software DSP before reaching the device.
This is mainly to support Apple M1/M2 devices, which require a software
DSP to be always present
- WpImplModule now supports loading module arguments directly from a SPA-JSON
config file; this is mainly to support DSP configuration for Apple M1/M2
and will likely be reworked for 0.5
- Added support for automatically combining Bluetooth LE Audio device sets
(e.g. pairs of earbuds) (!500)
- Added command line options in ``wpctl`` to display device/node names and
nicknames instead of descriptions
- Added zsh completions file for ``wpctl``
- The device profile selection policy now respects the ``device.profile``
property if it is set on the device; this is useful to hand-pick a profile
based on static configuration rules (alsa_monitor.rules)
Changes/Fixes:
- Linking policy now sends an error to the client before destroying the node,
if it determines that the node cannot be linked to any target; this fixes
error reporting on the client side
- Fixed a crash in suspend-node that could happen when destroying virtual
sinks that were loaded from another process such as pw-loopback (#467)
- Virtual machine default period size has been bumped to 1024 (#507)
- Updated bluez5 default configuration, using ``bluez5.roles`` instead of
``bluez5.headset-roles`` now (!498)
- Disabled Bluetooth autoconnect by default (!514)
- Removed ``RestrictNamespaces`` option from the systemd services in order to
allow libcamera to load sandboxed IPA modules (#466)
- Fixed a JSON encoding bug with empty strings (#471)
- Lua code can now parse strings without quotes from SPA-JSON
- Added some missing `\since` annotations and made them show up in the
generated gobject-introspection file, to help bindings generators
Signed-off-by: Markus Volk <f_l_k@t-online.de>
Signed-off-by: Khem Raj <raj.khem@gmail.com>
In order to pass reproducible tests, recipes that use the
useradd class must have static ids configured.
Signed-off-by: Fabien Thomas <fabien.thomas@smile.fr>
Reviewed-by: Yoann Congal <yoann.congal@smile.fr>
Signed-off-by: Khem Raj <raj.khem@gmail.com>
This is the first 1.0 release candidate that is API and ABI compatible
with previous 0.3.x releases.
- jackdbus support is now enabled by default.
- IRQ based scheduling in ALSA was improved and enabled by default for
Pro-Audio profile. It will also link the pcms together to get lower
latency. This now matches what JACK does and gives equal latency to
PipeWire for Pro-Audio profiles.
- Support both old and new versions of webrtc-audio-processing to make
the transition easier.
- Forced quantum changes by nodes or metadata will now also force a
suspend and resume of the graph, like the rate changes to make sure all
nodes adapt to the new quantum. This is important for Pro-Audio nodes
that need to reconfigure the hardware to a new period in IRQ based
scheduling.
- Fix a regression in regex parsing.
- Many bugfixes and improvements.
- jackdbus is by default enabled now. The idea is that when jackdbus is
installed, the real libjack.so is in the path and we can become a
real JACK client.
- Forces quantum changes by nodes or metadata will now also force a
suspend and resume in the graph, like the rate changes to make sure all
nodes adapt to the new quantum. This is important for Pro-Audio nodes
that need to reconfigure the hardware to a new period.
- The stream now has an EARLY_PROCESS option that can be used to implement
custum buffer fill levels. (#3480)
- Fix a regression in regex parsing. (#3528)
- Fix a bug in position reporting in the driver node. (#3189) (#3544)
- Destroying a link will now recalculate the graph correctly.
- Fix the rate comparison for finding the best rate in the graph.
- Use malloc_trim() when available to release memory. (#1840)
- pw-cat now supports DFF DSD files.
- pw-cli avoid some NULL derefs in some cases.
- The RAOP sink has seen some cleanups and improvements. It will now ask
for feedback every 2 seconds to keep some devices alive.
- A bug in filter-chain was fixed where it would fail to apply the gain
when mixing just one source.
- The filter-chain can now pass the stream volume to a control in the
filter-chain graph. (#3434)
- Improve volume handling in RAOP sink.
- Some cleanup in the pending_stream handling.
- Fix a regression in the event emission code where it failed to emit
a changed event when a node was linked. (#3522)
- Lower the realtime priority of pulseaudio clients.
- Set pulse.module.id on the echo-cancel streams. (#3541)
- Support both old and new versions of webrtc-audio-processing to make
the transition easier.
- The ALSA driver now does the sync of all followers directly from the
wakeup event. This results in more stable rate matching.
- IRQ based scheduling in ALSA was improved and enabled by default for
Pro-Audio profile. It will also link the pcms together to get lower
latency. This now matches what JACK does and gives equal latency to
PipeWire for Pro-Audio profiles.
- GNU/Hurd support was added.
- Some improvements to passthrough handling.
- Improvements to the codec handling when PipeWire is used as Audio
Gateway.
- Adapt to new Bluez API for BAP devices.
- When the jack library is set in the default library path, avoid using
LD_LIBRARY_PATH because this can cause confusion.
- Handle clearing the latency on a port.
- jack_property now always manages to actually change the metadata because
it waits for a roundtrip before exiting.
Signed-off-by: Markus Volk <f_l_k@t-online.de>
Signed-off-by: Khem Raj <raj.khem@gmail.com>
These are test images to build all recipes in layer. Renaming them makes
them refect what they are. Moreover we can rename the ptest images to
match OE-Core naming conventions for meta-oe/meta-perl/meta-python
Signed-off-by: Khem Raj <raj.khem@gmail.com>
These were essentially duplicates of core-image-minimal, however
core-image-base is a better baseline for upper layers, so switched the
consumers of these images to use core-image-base
Signed-off-by: Khem Raj <raj.khem@gmail.com>
The signature recalculation task can fail:
```
/.../build/tmp/work/armv8a-poky-linux/libcamera/1_0.0.4-r0/temp/run.do_recalculate_ipa_signatures_package.937421: line 154: /.../build/tmp/work/armv8a-poky-linux/libcamera/1_0.0.4-r0/git/src/ipa/ipa-sign-install.sh: No such file or directory
```
This happens whenever ${S} is removed and packages are retrieved from
sstate cache.
Instead of trying to fix the task, avoid the issue altogether by
performing the signature recalculation directly in do_package.
Signed-off-by: Tom Hochstein <tom.hochstein@nxp.com>
Acked-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
Signed-off-by: Khem Raj <raj.khem@gmail.com>
