AudioFlinger: fix RecordThread initial device
A regression was introduced when the audio device enums where
modified for a 32 bit representation: the device passed when
constructing a RecordThread was still the concatenation of input
device and output device bit fields on one 32 bit value which
is not possible anymore.
The fix consists in modifying the RecordThread constructor to accept
separate values for input and output devices.
Change-Id: I81fb5f4718428b54251e65d74b86e198ce15193e
diff --git a/services/audioflinger/Threads.cpp b/services/audioflinger/Threads.cpp
index 744a7df..af0dccc 100644
--- a/services/audioflinger/Threads.cpp
+++ b/services/audioflinger/Threads.cpp
@@ -3545,9 +3545,10 @@
uint32_t sampleRate,
audio_channel_mask_t channelMask,
audio_io_handle_t id,
- audio_devices_t device,
+ audio_devices_t outDevice,
+ audio_devices_t inDevice,
const sp<NBAIO_Sink>& teeSink) :
- ThreadBase(audioFlinger, id, AUDIO_DEVICE_NONE, device, RECORD),
+ ThreadBase(audioFlinger, id, outDevice, inDevice, RECORD),
mInput(input), mResampler(NULL), mRsmpOutBuffer(NULL), mRsmpInBuffer(NULL),
// mRsmpInIndex and mInputBytes set by readInputParameters()
mReqChannelCount(popcount(channelMask)),