Computer system and method for executing an automotive customer function

A method for executing a customer function comprising an automotive customer function, in a computer system, the method comprising:

• generating, by the customer function, customer function output data, based on which a machine comprising an automobile, is controlled,
• wherein the computer system comprises one or more processing cores,
• wherein the customer function comprises two or more applications, wherein each application comprises a plurality of tasks, wherein during the execution of an application of the two or more applications one or more tasks of the plurality of tasks of said application are executed,
• wherein said applications are executed in form of a computation chain one after the other in a defined sequence,
• wherein a computation chain receives customer function input data at its start and generates customer function output data which are provided at the end of the execution of the computation chain,
• and wherein during execution of the customer function said computation chain is executed once or several times,
• wherein the computer system provides containers, wherein the computer system is configured to activate and de-activate said containers, so that a container of the containers is active or inactive,
• wherein all tasks of the applications are assigned to containers, and wherein all tasks of each specific application are assigned to exactly one specific container,
• wherein in a timeframe, during which a container is active, one or more processing cores of the computer system are exclusively reserved for the execution of the tasks of the application of said container, such that no task of another container can run on said one or more exclusively-reserved processing cores during the timeframe, and
• wherein the computer system is configured such that when a container is inactive, the execution of all tasks of said inactive container is suspended and prevented on the computer system,
• wherein the computer system is configured to execute the containers according to the defined sequence of the applications, so that a container is activated before its immediately following container, and wherein activation timeframes of a container and its immediately following container of a computation chain are not allowed to overlap in time, and
• wherein for each container a task sequencer is provided, wherein said task sequencer is a first task started in the container when the container is activated, and wherein the task sequencer of a container decides (“task-sequencer-decision”) based on a directed acyclical graph defining task dependencies:
• which of the tasks of the application of the container have to be executed,
• a sequence of tasks to be executed, and
• for each task, to be executed, the one or more processing cores reserved for the container, on which the task is to be executed,
• and wherein the computer system is configured to execute, during the activation timeframe of each container, the tasks of each container according to said task-sequencer-decision of the task sequencer of each of the containers.

A computer-implemented method for executing real-time applications in a vehicular computing platform, the method comprising:

• controlling, by a platform scheduler executing on one or more processors, activation of a plurality of containers, each container encapsulating a respective set of application tasks;
• assigning, by the platform scheduler, exclusive access to at least one processor core to a container only during an activation interval of that container, such that no task of another container executes on the at least one processor core during the activation interval;
• executing, during the activation interval, only tasks belonging to the activated container on the at least one processor core;
• enforcing, by the platform scheduler, a non-overlapping temporal execution order among containers of a computation chain such that activation intervals of containers in the computation chain do not overlap in time;
• suspending execution of all tasks of a container outside the activation interval of that;
• and
• during the activation interval of a given container, managing, by a task sequencer, which is a task started within the given container and independent of the platform scheduler, an execution order of tasks belonging to the given container on the at least one processor core, and deciding, for each task to be executed, on which of the at least one processor core exclusively assigned for the given container the task is to be executed;
• wherein a task sequence auditor supervises one or more of the following correct execution properties of a container:
• non-execution of tasks,
• multiple execution of tasks, and
• incorrect sequence of task execution, and
• if any of these conditions is detected, the task sequence auditor reports an error.