Introduction

Sound Open Firmware is mainly written in C with a small assembler for DSP initialization and some DSP intrinsic values for media processing. The intended audience is software developers familiar with C programming and media processing.

Developers wishing to participate with upstream should also be familiar with git and GitHub.

Knowledge of hardware debugging (such as JTAG) and use of emulators is also desirable if bringing up new hardware.

SOF Core Concepts

Following are core concepts and terms used by SOF developers.

Component An audio or signal processing component that processes input data into output data. Components can have one or more input source buffers and one or more output sink buffers. Components can also send and receive runtime configuration data that can be used to monitor or alter the data processing.

Buffer A memory region that can be used to share audio processing data between components. Buffers can have certain attributes depending on their usage, such as a DMA’able buffer.

Pipeline A collection of audio processing components and buffers that are scheduled for processing together such as a schedA pipeline that can have multiple source and sink endpoints. The endpoints may be other pipelines or components.

DAI Digital Audio Interface. A hardware audio serial interface used to send audio data between hardware devices. Examples are I2S, Soundwire, PDM, HDA, and HDMI.

Topology A high-level description of the network of all pipelines and components enumerated on the DSP. This is initially described in text format before being compiled into a binary that firmware can process.

Module Another name for component. Implies that a component is linked at runtime rather than at build time.

Driver A device driver used by firmware to control hardware devices (such as DMA or I2S) or a Sound Open Firmware host OS device driver.

AAL Architecture Abstraction Layer. A firmware abstraction layer used to abstract architecture-specific code.

SRC Sample Rate Converter. An audio component used to convert the sample rate of a synchronous input stream to a synchronous output stream.