A sound card, encapsulating playback and capture devices will be visible as single entity to the end user. There can be many playback and capture devices within a sound card and there can be multiple sound cards in a system.
The Machine driver creates a pipeline out of the ALSA drivers. This glue or DAI (Digital Audio Interface) link is made using registered device names or device nodes (using OF kernel framework). Each proper DAI link results as a device in a sound card. A sound card is thus a logical grouping of several such devices.
The Audio Formatter driver creates the platform device for the sound card. While creating the device, it passes the HDMI device tree node of either I2S/HDMI/SDI/SPDIF depending on the kind of sound card being created. When the sound card driver detects the kind of audio node (I2S/HDMI/SDI/SPDIF), the proper DAI link is selected from the available links.
HDMI Rx receives the data from the HDMI source and separates audio from the video content. The Xilinx Audio Formatter converts this AES data to PCM data and stores it in memory. HDMI TX gets the AES data from the Audio Formatter and embeds it into video.
The AES format contains PCM and channel status information. The Audio Formatter IP separates non-audio content such as channel status and stores it in registers. The Audio Formatter driver can parse the content of channel status to get audio parameters.
A dummy CPU DAI driver is used as there needs to be a CPU DAI to be registered with ASoC framework. Codec DAI will be part of HDMI Tx and Rx video drivers, as those provide and consume the digital audio data.
In this TRD design, a sound card is created with a record device for the HDMI-RX capture pipeline and a playback device for the HDMI-TX playback pipeline. The supported parameters are:
Sampling rate: 48 kHz
Sample width: 24 bits per sample
Sample encoding: Little endian
Number of channels: 2
Supported format: S24_32LE