Control Daemon

1 Overview

>>>>TODO Overview text.

Currently, booting up the camera is as follows:

• A systemd service is activated which runs a shell script.
• This script then loads a bitstream into the FPGA and uses other scripts and C programs to train LVDS channels and set up the HDMI output

When the service is disabled, the user can run this script manually, which:

• Does not keep the user from running if it has already been activated by systemd.
• And on the other hand, simple scripts which query the registers (e.g. to get the temperature of the sensor) can be activated even if the FPGA bitstream didn't load.

All of those lead to a solid lockup of the Beta, because if you write to one of those memory addresses which are used for communicating with the PL, and there is no handler in the FPGA, the ARM cores lock up solid, no recovery possible they basically wait for an ACK/NACK forever.

Research is taking place in the Labs here. Code can be found on GitHub

2 Structure

The control daemon project currently consists of three different modules:

• Web UI - HTML5, sends requests to the backend, currently by using JQuery (still evaluating alternatives).
• WSServer - receives WebSocket requests, converts them to Flatbuffers packages and sends them to the daemon, through socket.
• Daemon - processes flatbuffer packages received over a UNIX domain socket and calls suitable handler.

The communication with the daemon is done via a UNIX domain socket. There are two different protocols used when communicating with the daemon. A handshake protocol which tells the user what parameters and modules are available, their minimum and maximum value, as well as a description.

Preliminary handshake protocol:

[
  {
    "module": "image_sensor",
    "parameter": "analog_gain",
    "possibleValues": [
      "1/3", "2/3", "1", "3/3", "4/3", "2", "3", "4"
    ],
    "currentValue": "",
    "defaultValue": "1",
    "range": [],
    "readOnly": false
  },
  {
    "module": "image_sensor",
    "parameter": "digital_gain",
    "possibleValues": [
      "1", "2", "3", "4", "6", "8", "10", "12", "14", "16"
    ],
    "currentValue": "",
    "defaultValue": "1",
    "range": [],
    "readOnly": false
  }
]

At the moment this parameters are residing in description.json, which is read by the daemon at the initialization phase. If following command is received, then the daemon fills in current values, places the data under value1 and sends it back to caller:

module: general
parameter: available_parameters

The second protocol is used for writing and reading the different parameters. It uses Flatbuffers. The schema for Flatbuffers can be found here.

Communication of the Web UI with the daemon it bridged from the UNIX domain socket and flatbuffers to websocket and JSON by the WSServer. The WSServer accepts following fields, which get translated to the Flatbuffers schema (and vice versa):

Field	Type	Example	Comment
sender	string	WSServer
module	string	"image_sensor"
command	string	"set" or "get"
parameter	string	"analog_gain"
value1	string	"4"
value2	string	"5.7"
status	string	"success" or "fail"	used for reply from Daemon
message	string		status message, to get more info when request fails
timestamp	string		date and time of camera, when the request was send back to client

TODO: Add JSON/REST package description from Lab (https://lab.apertus.org/T865)

3 Build

Required packages (names are varying between Linux distributions):

• cmake
• clang
• ninja
• git

Steps:

• Install required packages
• Clone the repo: git clone --recursive https://github.com/apertus-open-source-cinema/axiom-control-daemon
• cd into cloned repo
• mkdir build && cd build
• cmake -GNinja ..
• ninja

6 CLI

To set/get parameters from command line, DaemonCLI can be used:

Syntax: DaemonCLI <module> set/get <parameter> <value>

Examples:

Set the image sensor analog gain to value "2":

DaemonCLI image_sensor set analog_gain 2

Read the value from image sensor register address 1 which corresponds to the "Number_lines_tot[15:0]" according to the datasheet and means the number of lines to be read (default 3072):

DaemonCLI image_sensor get config_register 1

6.1 Available modules

general	General methods, like getting available parameters through get_available_methods
image_sensor	CMV12000 (currently)

6.2 Available parameters (per module)

image_sensor

digital_gain	Digital gain
analog_gain	Analog gain
config_register	read/write arbitrary sensor config register (further parameters are index and value)

7 Development notes

CMV12000Adapter will be used as example. Please look at the class for examples of implementation.

7.1 Add new parameters

Before registering new parameters, two methods should be added: a setter and a getter.

Declaration syntax:

   bool <setter_name>(std::string value1, std::string value2, std::string& message)

   bool <getter_name>(std::string& value, std::string& message)

Note the ampersand (&) after the type, it is very important for returning multiple values. As methods can return only one value, bool in this case, variables can be passed by reference (&) to be able to set their values inside methods. Values can be modified as usual and the value will be passed back to the caller (see IDaemonModule.h and MessageHandler.cpp for examples).

Setter receives two values, in case just one is required value2 will be 0 (zero).

After the setter and getter are implemented, it's time to attach them to a parameter name. Here they are registered to gain parameter:

 void CMV12000Adapter::RegisterAvailableMethods()

 {

   AddParameterHandler("analog_gain", GETTER_FUNC(&CMV12000Adapter::GetAnalogGain), SETTER_FUNC(&CMV12000Adapter::SetAnalogGain));

   AddParameterHandler("digital_gain", GETTER_FUNC(&CMV12000Adapter::GetDigitalGain), SETTER_FUNC(&CMV12000Adapter::SetDigitalGain));

 }

This code will attach methods like SetAnalogGain() and GetAnalogGain() to the corresponding parameter "analog_gain".

8 Unit tests

Unit tests have been added to the project to verify correct functionality. Catch2 framework is used because it's single-header only and utilizes the C++11 way of doing things.

Note: (for development on PC) - RAM access of the camera is different from x86/x64 CPUs, modified classes have to be used to bypass this, otherwise SEGFAULT would be the result.

In the CMake scripts a switch called ENABLE_MOCK was added so that users can disable any code which won't work on a regular PC (see CMV12000AdapterTests.cpp for an example). While running the build on camera cmake .. is sufficient, but for development one should use:

$ cmake -DENABLE_MOCK=ON ..