An issue exists in the fair scheduler (`sched/fair`) regarding the tracking of `zero vruntime`. When two runnable tasks repeatedly perform a yield operation, they can enter a state where they leapfrog each other. Because `zero vruntime` is only updated during ticks or full enqueue/dequeue operations, and not during ` enqueue entity()` or ` dequeue entity()` calls from `put prev task()` or `set next task()`, the tracking can fall behind. If these tasks reach an overflow point within a single tick's duration, it can lead to system instability. This risk is further increased when multiple cgroups (control groups, a Linux kernel feature that limits, accounts for, and isolates the resource usage of a collection of processes) are involved, as there is no guarantee that every cgroup will receive a tick in a timely manner.