Synthetic Events and Tracing Latency Within the Kernel

Learn how to leverage synthetic events for kernel latency tracing in this conference talk that explores the Linux kernel's extensive monitoring capabilities and tracing infrastructure. Discover how synthetic events connect two separate kernel events to create a third composite event containing fields from both original events, plus the time delta between them to measure latency. Explore the kernel's ability to observe various activities including task wake-ups, CPU scheduling, task migration, block I/O operations, and memory faulting. Master the creation of synthetic events through practical examples used in production environments for debugging complex issues. Understand filtering techniques to extract relevant data while eliminating noise from trace outputs. Get introduced to the trace-cmd sqlhist tool that simplifies synthetic event creation through SQL queries, replacing the kernel's complex user interface with an intuitive database-like approach to kernel event analysis.