Exceptions

Control Flow

Untitled

we currently use a state-based control flow that is completely dependent on the instructions that happens before
- we need to have some sort of reactivity, that could be accomplished either through polling (slow) or exceptional control flow

change the control flow in the system in response to a system event → throw exceptions
- exceptions —> process context switches (OS/Hardware) and send signals (OS)
Process context switch
- Implemented by OS software and hardware timer
Signals (messages from operating system)
- Implemented by OS software

an exception is a transfer of control to the OS kernel in response to some event
- for instance, lets say we run into a divide by 0 segfault, exception is thrown out and the OS will handle the instruction that is best fit

Untitled

exception tables will give us a handle over the different exceptions and will give us different approaches to solving the problems
- Basically think of a list of numbers that lead to different code for each of the issues that might arise

asynchronous exceptions are caused by external processes to the processor (handler returns to next instruction once done)
- consider a timer that is done and pauses other processes and takes over
- consider copy paste that takes data from the network and pastes it
synchronous exceptions → caused by executing an instruction
- traps are special intentional instructions that returns control to the next instruction
  - consider a system call (breakpoints)
    - registers contain the sycall number and other registers hold the arguments (return RAX) - negative return means fail
- faults are unintentional exceptions that can either redo the instruction or move on
  - think of deviding by 0
  - page fault → allocates memory if it has not been allocated already
- abort unintentional and unrecoverable which aborts the current program
  - illegal instructions