Jakub Kicinski <kuba@xxxxxxxxxx> writes: > On Wed, 29 May 2024 13:18:42 +0200 Petr Machata wrote: >> +fib_multipath_hash_seed - UNSIGNED INTEGER >> + The seed value used when calculating hash for multipath routes. Applies > > nits.. > > For RSS we call it key rather than seed, is calling it seed well > established for ECMP? > > Can we also call out that hashing implementation is not well defined? As others note, this seems to be industry nomenclature, so I'll keep it. >> + to both IPv4 and IPv6 datapath. Only valid for kernels built with > > s/valid/present/ ? Ack. >> + CONFIG_IP_ROUTE_MULTIPATH enabled. >> + >> + When set to 0, the seed value used for multipath routing defaults to an >> + internal random-generated one.
