Wavious LPDDR Software project. This project contains necessary device driver and logic for controlling Wavious LPDDR Hardware. See wav-lpddr-hw for more details about the hardware platform.
This project is licensed as Apache-2.0. Previous releases used GPL-3.0 and LGPL-3.0.
At a high-level the project is organized into 5 main directories:
RTOS contains all of the source code for FreeRTOS and any additons that have been added by Wavious for running FreeRTOS on Wavious developed RISC-V hardware.
Drivers are the layer that interacts directly with the Wavious LPDDR hardware.
These functions are identified by the reg_if suffix. All interactions with the
hardware consolidated to this layer in order to make the code easier to
maintain and debug. The driver layer sits on top of the RTOS kernel and relies
on reg_read and reg_write functions for manipulating CSRs.
Devices are higher abstractions of device drivers. Devices can wrap underlying drivers in order to provide better abstractions or even combine multiple drivers and device to create a higher-level device. Devices sit at the layer above drivers.
PHY Firmware is the management entity that controls the PHY, as well as performs state management. It responds to events from DFI interface (via IRQs) or events requested through the PHY Firmware API. The PHY Firmware API encapsulates all PHY features that an application can control, such as frequency switch prep.
Applications are complete programs that are meant to run on LPDDR hardware.
Applications should interact with the PHY via the PHY Firmware API. They are
not able to control or update the PHY directly. Typically, applications can be
"fully baked" mission type applications (wddr_boot) or test applications.
For this repository, only mission-mode applications have been included.
git clone https://github.com/waviousllc/wav-lpddr-sw.gitcd wav-lpddr-swgit submodule update --init --recursive
It is recommended that Docker is used for the build environment. Build the Docker
Image using the Dockerfile located under the docker directory of this
project. Once built, use ./run.sh script to create a shell within the Docker
environment. All code is linked into the container via a Docker volume. Thus,
changes persist between runs and edits are shared outside and within the
Docker container.
docker build -t <IMAGE_NAME>:latest docker/
./docker/run.sh <IMAGE_NAME>
---- In Docker Container ----
./configure (only required once)
cd build
make
| Varaible Name | Default Value | Description |
|---|---|---|
| CONFIG_CALIBRATE_PLL | true | Enables PLL calibration at boot |
| CONFIG_CALIBRATE_ZQCAL | true | Enables ZQCAL calibration at boot |
| CONFIG_CALIBRATE_SA | true | Enables Sense Amp calibration at boot |
| CONFIG_DRAM_TRAIN | false | Enables DRAM Training at boot |
| DCONFIG_CAL_PERIODIC | false | Enables PHY Periodic Calibration |
It is recommended that all binaries are built with the default configuration. However, in some instances, it might be convenient to disable certain features while debuging issues.
After running the commands given in the Building section of this document, the configuration can be updated as follows:
cd build
cmake .. -DCONFIG_CALIBRATE_PLL=<true|false> -DCONFIG_CALIBRATE_ZQCAL=<true|false> -DCONFIG_CALIBRATE_SA=<true|false> -DCONFIG_DRAM_TRAIN=<true|false> -DCONFIG_CAL_PERIODIC=<true|false>
make
The generic command: cmake .. -D<CMAKE_VARIABLE_NAME>=<VAL>
In order to allow for extended capabilites not required for PHY functionality (such as training), an interface library was added to the WDDR device, named wddr_ext. In order to utilize this library, you can add additional source and include files to this library to be included when building the WDDR device library. This will enable external functions to be added to the build to override weak functions within the WDDR device (such as wddr_train).
Currently, the WDDR device includes the necessary hooks to perform DRAM training during boot of the PHY, but the algorithms are not included in this release. These algorithms can be added using the wddr_ext interface library. If training is a requirement of your application, contact Wavious for support.