This talk aims to cover the new preemption model the linux kernel might be going towards. It is called PREEMPT_AUTO. It is a preemption model driven by the scheduler, instead of relying upon the existing framework of cond_resched. This talk would delve into what preemption is, what are the preemptions points, how they are triggered/set, what is the drawback of existing models and how preempt auto fixes some of the pain points of existing model. In this talk, further will discuss about the challenges that are currently present in the preempt auto model that was seen in PowerPC
v2:
https://lore.kernel.org/all/20240528003521.979836-1 ankur....@oracle.com/More info can be found through cover letter of v2.
Time preference: Full length talk.
Thanks
Shrikanth Hegde