Age | Commit message (Collapse) | Author | Files | Lines |
|
Generated using xml2ucm at configuration at:
https://github.com/ford-prefect/xml2ucm/blob/master/examples/aries-l-config.xml
Tested on Android L-based Firefox OS.
|
|
This has earpiece, speaker and headphones invidually tested - no
combined devices tested yet. Also there's a glitch when speaker playback
starts that needs investigation.
|
|
This adds a low-latency modifier that can be used for voice and
potentially other low-latency streams.
|
|
This sets up the Speaker device for h/w volume control as an example.
This can be extended to other devices. For the Headphones device, we
would probably want to merge the HPHL and HPHR controls to provide a
single stereo control so things can just work.
|
|
This provides a volume mixer control and a mute switch for the mic that
allow standard ALSA controls to be used to control voice call playback
volume and capture muting.
|
|
This leaves device-specific configuration in the top-level repo and lets
the upstream build system not have to care about the actual device.
|
|
|
|
This starts/stops csd-daemon at VoiceCall verb enable/disable time.
|
|
|
|
|
|
Needed to cleanly change routing while allowing CPU-side playback to
work during a call.
|
|
|
|
Needed for building with NEON support, for example
|
|
The script needs a bit of cleanup (port selection shouldn't be here),
but it's convenient for testing.
|
|
Add csd-daemon/csd-client to the build, as well as the VoiceCall UCM.
|
|
This factors out device enable and voice start to share code, makes sure
that the start/stop-voice on probe doesn't trigger an error (we only
really call stop if start was really triggered by an enable-device).
This also removes the deferred disable-device logic. As it turns out, we
probably should be disabling the device immediately to follow how the
Android HAL does it (since the modem bits are so fragile). In order to
not trigger an error by doing a stop-voice while no device is enabled,
we do a fake enable-device if required before calling stop-voice. This
is seriously ugly, but I don't see a better way to do it other than
Qualcomm fixing their library/firmware.
|
|
Changes include:
* Move volume setup to verb enable time instead of individual devices
like the Android mixer XML
* Enable/disable tx mixer control on mic device enable/disable just as
we do with the rx mixer for playback devices
* Optimise mixer enable/disable by not re-enabling the mixer in
DisableSquence (it'll get enabled in the following EnableSequence on
device switch)
|
|
We need to disable some mixer controls before changing routing. We
disable and then enable these in the EnableSequences and only disable in
the DisableSequence as an optimisation to avoid two sets of enables when
switching devices.
|
|
A few changes here:
* Move volume initialisation to verb enable, similar to how the
Android XML works
* Drop Headphones from conflicting with itself
* Add a missing mixer controls
* Minor cosmetic documentation changes
|
|
|
|
Trying to manage them in PA while maintaining ordering with
libcsd-client calls is too fragile and seems to not be worth the effort.
|
|
disable-device should only be run before the next enable-device and not
before a stop-voice. We'd like to keep the disable-device in the
DisableSequence in UCM for cleanliness, so we'll hide this detail away
in the daemon.
|
|
Uses libcsd-client.so to talk to the modem firmware on mako. Needs to be
in a separate process as we issue the commands from UCM via Exec
commands, but the library needs to at least be open throughout the
duration of the call.
|
|
This installs the PA and UCM config files from data/ instead of
depending on modifying the top-level device/ make files.
We do assume that all devices are UCM-based, but this is a reasonable
assumption for now.
|
|
This isn't complete yet, probably needs quite a bit of tweaking, but
does get basic output going.
|
|
Our tree has tags now, so no more UNKNOWN.UNKNOWN
|
|
|
|
This avoids the need to have androgenizer in the path, etc. With this,
we just Just add the androgenizer repo to local manifest, and we're
done.
|
|
A number of other trees contain bits of alsa-utils, so let's make this
opt-in.
|
|
Seems to not required the PWD prefix any more.
|
|
The submodules will now be pulled using standard Android tools instead
(repo).
|
|
|
|
The earphone enable switch should only be switched on when specifically
needed (i.e. when the Earphone device is selected).
|
|
Now that we're using roles for modifiers, we use module-intended-roles
for role-based routing. We also make the suspend timeout shorter so that
modifiers are only enabled for the short period when they are needed.
|
|
|
|
|
|
|
|
At least 'Headset' capture and playback paths work with this.
|
|
|
|
This pulls in UCM updates from Linaro for the Pandaboard ES and adapts
it for Tuna usage. Some tweaking might be necessary for input paths, at
least in the VoiceCall verb.
|
|
This makes it easier to quickly build and install amxier/aplay without
mucking around with the build system.
|
|
This includes a PulseAudio-side changing adding and Android policy
module, and configuration changes to use this.
|
|
|
|
Added a fix for bootstrapping that should make things build without
intervention now.
|
|
Adds a fix for broken configure due to a bad patch file upstream.
|
|
|
|
|
|
Contains one fix for a generated file.
|
|
Contains a fix for bootstrapping.
|
|
Don't know how it got a bad commit id.
|