Learn how debuggers work by building your own one - Sayani Bhattacharjee

Learn how to build a basic debugger from scratch by creating a target program, setting breakpoints, and controlling a child process using ptrace, and explore the internal implementation of step in and step out instructions.

Key takeaways

A debugger can be built from scratch by creating a target program, inserting a breakpoint, and controlling a child process using ptrace.
ptrace can be used to get the current program counter (PC), set breakpoints, and read registers.
To step through a program, insert breakpoints at specific lines and use ptrace to control the child process.
A step in instruction is different from a step in on the source code, and both are implemented using breakpoints.
A step over instruction can be implemented by inserting a breakpoint at the next line of the source code.
A step out instruction can be implemented by inserting a breakpoint at the return address of a function call.
The DWARF file can be used to parse function calls and get the program counter.
To read variables, the compiler’s symbol table and the DWARF file can be used.
The ptrace system call can be used to read and write memory.
A debugger can be built using a UI layer, a symbolic layer, and a target layer.
Delve, a debugger for Go, provides a UI for debugging and supports commands such as step in, step out, and step over.
The internal implementation of step in and step out instructions in Delve involves setting breakpoints at specific lines of code.

Learn how debuggers work by building your own one - Sayani Bhattacharjee

More talks