Taking Interruptions in Stride: Debugging an xv6 Lost Wakeup

Fri, 03 Apr 2026 19:39:43 -0400

A few months ago, I decided to add a stride scheduler to the xv6 operating system, to get a better understanding of OS schedulers. I wrote a blog post about the experience, including my code and how it performed. Toward the end of the post, I mentioned some odd behavior when running my scheduler on a single-core VM. In particular, at bootup the OS would hang when starting the initial process. I was able to determine that the process had gone to sleep, but wakeups were failing because when they fired, init was still considered an active process. I worked around the issue by removing one of the wakeup guard conditions and decided to “punt” a deeper investigation.

Gdb on Nick Chandler

Taking Interruptions in Stride: Debugging an xv6 Lost Wakeup