On Mon, 31 Aug 2020 09:35:18 +0300 Andrew Rybchenko <arybche...@solarflare.com> wrote:
> >>>>> multisegment packets. > >>>> > >>>> I hope it will be mentioned in the feature documentation in the future, > >>>> but > >>>> I'm not 100% sure that it is required. See below. > >>> I suppose there is the hierarchy: > >>> - applications configures DEV_RX_OFFLOAD_SCATTER on the port and tells in > >>> this way: > >>> "Hey, driver, I'm ready to handle multi-segment packets". Readiness in > >>> general. > >>> - application configures BUFFER_SPLIT and tells PMD _HOW_ it wants to > >>> split, in particular way: > >>> "Hey, driver, please, drop ten bytes here, here and here, and the rest - > >>> over there" > >> > >> My idea is to keep SCATTER and BUFFER_SPLIT independent. > >> SCATTER is a possibility to make multi-segment packets getting > >> mbufs from main rxq mempool as many as required. > >> BUFFER_SPLIT is support of many mempools and splitting > >> received packets as specified. > > > > No. > > Once again, drivers should take anything from application and rely on using > > logic to choose best path. Modern CPU's have good branch predictors, and > > making > > the developer do that work is counter productive. > > Please, add a bit more details. I simply can see relationship. > So, right now for me it looks like just misunderstanding. > > Thanks, > Andrew. Ok, documenting the existing behaviour is good. I was just concerned that this was going to lead to more per-queue flags.
