Re: [PATCH bpf v2 8/8] libbpf: ringbuf: Prevent missed wakeups

From: Tamir Duberstein

Date: Fri Nov 20 2026 - 15:56:56 EST


> so we do this one time retry

The retry is needed for correctness. The producer can commit after the
first BUSY read but observe the old consumer position and skip wakeup.
The consumer must reread the header after fencing before returning.

> [...]

> Wouldn't it be cleaner and easier to
> mentally follow to have `goto done`

Done for the bound exit in v3. Callback errors now publish, fence, and
return directly; cnt = err is gone.