> From: Thomas Monjalon [mailto:[email protected]] > Sent: Thursday, 18 June 2026 18.20 > > 17/06/2026 13:48, Morten Brørup: > > > From: Huisong Li [mailto:[email protected]] > > > Sent: Wednesday, 17 June 2026 12.28 > > > > > > Add the RTE_LCORE_ prefix to the lcore role enum values in > > > rte_lcore_role_t > > > to follow DPDK naming conventions. > > > > > > - ROLE_RTE -> RTE_LCORE_ROLE_RTE > > > - ROLE_OFF -> RTE_LCORE_ROLE_OFF > > > - ROLE_SERVICE -> RTE_LCORE_ROLE_SERVICE > > > - ROLE_NON_EAL -> RTE_LCORE_ROLE_NON_EAL > > > > > > Old names are kept as macros aliasing to the new names to preserve > > > backward compatibility. > > > > > > > Series-Acked-by: Morten Brørup <[email protected]> > > Squashed and applied, thanks. > > I have a doubt about RTE_LCORE_ROLE_RTE: could we find a better name?
Yes, it made more sense when lcores were either used by DPDK or not. Note that it's used by both workers and the main lcore. > > Also we should probably add some comments to explain each role. +1 BTW, I thought the Non-EAL role was for registered control plane threads. But then I looked into Grout. Grout uses the Non-EAL role for its dataplane threads. And the RTE role for its main thread, which is handling the control plane (which I suppose is not unusual). Lcore roles and CPU core isolation is very flexible in DPDK. Maybe too flexible.
