Chapter 9: Tracing the Kernel Flow

Tracing is the ability to collect relevant details as code executes. Typically, data collected will include function names (and perhaps parameters and return values) of function calls made along the code path being followed, the context that issued the call, when the call was made (a timestamp), the duration of the function call, and so on. Tracing allows you to study and understand the detailed flow of a system or a component within it. It's akin to the black box in an aircraft – it simply collects data, allowing you to interpret and analyze it later. (You can also consider tracing to be loosely analogous to logging.)

Profiling is different from tracing in that it typically works by taking samples (of various interesting events/counters) at periodic points in time. It won't capture everything; it (usually) captures just enough to help with runtime performance analysis. A profile of code execution, a report, can usually be generated...