OpenBSD's PF packet filter has removed a 32-bit bandwidth limitation that capped traffic shaping at 4.29 Gbps. The fix expands HFSC scheduler bandwidth fields from 32-bit to 64-bit integers, enabling proper configuration for modern 10G, 25G, and 100G network interfaces. This change addresses a silent wraparound issue where 10G queue settings would produce unpredictable scheduling behavior.

Peter N. M. Hansteen reported the patch also corrects a display bug in pfptop(1) where bandwidth values above 4 Gbps showed incorrectly. The update allows queue bandwidth configuration up to 999G, with familiar syntax continuing to work for existing setups below 4G. Users can now specify values like queue rootq on em0 bandwidth 10G queue defq parent rootq bandwidth 8G default without encountering the previous 4 Gbps ceiling.

The development comes as OpenBSD advances SMP kernel support and adds drivers for high-speed network cards. The patch is ready for commit after brief discussion on the tech@ mailing list. This technical improvement ensures OpenBSD's PF queues remain viable for enterprise networking deployments where multi-gigabit traffic shaping is essential.