On Fri, Jan 24, 2014 at 16:27, Christian Weisgerber wrote: > Henning Brauer <lists-openbsdt...@bsws.de> wrote: > >> i need this tested on an sk(4). >> I don't have that hardware at all. > [Summary: Henning wants to confine in_cksum_phdr() to ip_output.c and > remove its only other user sk_rxcsum().] > > This is an instance of a very rudimentary RX checksumming engine > that only runs over the payload and we have to supply the pseudo-header > checksum. > > Apart from sk(4), there are other drivers in the tree that have > more or less the same code: gem(4) and the two instances of hme(4). > Instead of calling in_cksum_phdr() they have their own inline > implementation, though.
Some questions: How much does this rudimentary half checksumming help? How much penalty is there be to simply not using hw csum on these cards? Are people still using sk, gem, or hme (!) in pps performance critical situations? Can we simplify the code without causing great harm?
