Taken SP25. Taught by Nils Napp.
A scheduling algorithm is a strategy used to pick a ready task for execution. A round robin scheduler has the ability to stop a process in the middle of execution, so it is a non-preemptive scheduler. One way we can judge a scheduler is by its average response time, in which we want to minimize the average tmie that a job waits in the ready queue before it gets scheduled....
Concurrency Recap In the past, we have been discussing concurrency, whether that be multiple programs on the same CPU or multiple processors accessing the same memory. We studied mutual exclusion, specifically using locks (which ensure that no two programs enter their critical sections at the same time). We studied how atomicity ensures only one programming is running at once, but that is a very strong requirement. Concurrency means only one critical section is running at any given time though we can still have other interleaving....
Dekker’s algorithm correctly implements mutual exclusion. However, this is a little bit convoluted. Additionally, each process must know about the other process, which violates modularity for programming. Therefore, we create a new abstraction: locks. Say we have a lock called l. Then, we can acquire the lock (lock(l)) or release the lock (unlock(l)). Then, a program consists of the NCS then a lock then CS then unlock. Note that using a single lock for different resources that need to be protected with a lock can be inefficient....
There are two ways of thinking about concurrency control: atomic actions vs. mutual exclusion. The latter is more fine-grained. Atomicity means that nothing else can happen at the same time. Imagine we had a one-lane bridge. If you were trying to cross the bridge, being atomic would mean stopping all cars in Ithaca so that you can safely cross. This is a very strong statement about the system and is overkill....
Imagine a bitstream controlling some LEDs. If you have an interrupt in the middle of a bitstream (not aligned with the 24 bits that encode the color), then things will go wrong. If we wanted blue light to travel down an LED strip faster than a red light, instead of using one loop, we should use two. But if we run two loops together, we can get weird bugs. If we want two pieces of code to interact in the same memory space, we call them threads (as opposed to processes)....
Context Switching If you open task manager, you will see hundreds of processes running all running at the same time. There is the illusion that there is a CPU dedicated to each program, but in reality, all of them have to time share on a single CPU. In order to do this, we need to stop programs and switch very quickly. We refer to running programs as processes. What does the CPU need to execute a program the same way at two different times?...
I2C Bus In CAN, we worried about clock synchronization. This is not an issue with I2C because I2C has separate data (SDA) and clock (SCL) lines. The data set set when SCL is low. Conventionally, when SCL is high, the data should not change. The master signals the start of a message by changing SDA from high to low while SCL is high. Conversely, a stop is indicated by changing SDA from low to high while SCL is high....
ISR Recap Recall that ISRs are like functions, and the processor is hardwired to execute ISR and return. This includes pushing the context onto the stack (everything needed to resume execution as if nothing happened). There are two types of interrupts: Non-maskable interrupts cannot be disabled, such as resets or HardFaults. Maskable interrupts are the most common. These are controlled by the user and can be selectively activated. Additionally, recall that in our case, we need to enable interrupts, which requires several things:...
Let’s say your embedded device has some buttons, which are ways the user can interract with the program. We want the user to be able to respond appropriately in all situations. For example, if a user presses two buttons at once, the program should not crash. One way to know if a button pressed is polling, where a loop continually reads the GPIO port. Another option is to have an interrupt, in which the main code causes the program counter to move to another chunk of code called an interrupt service routine (ISR) and go back to the main code....
So far, we have used the OpenSDA USB which connects to an ARM debugging chip which is serially connected to the KL46Z256. This can interface with the red LED. We can also interface with a segment LCD, magnetometer, and accelerometer. The chip has a lot of pins surrounding it, and it is isometric, so we can match the dot on the diagram with a dimple on the chip. Note that some pins are labeled similarly, e....