On Wed, Apr 20, 2016 at 03:17:40AM -0700, Sowmini Varadhan wrote: > This patchset follows up on the discussion in > https://www.mail-archive.com/netdev@vger.kernel.org/msg105090.html > > For RDS-TCP, we have to deal with the full gamut of > nonlinear sk_buffs, including all the frag_list variants. > Also, the parent skb has to remain unchanged, while the clone > is queued for Rx on the PF_RDS socket. > > Patch 1 of this patchset adds a pskb_extract() function that > does all this without the redundant memcpy's in pskb_expand_head() > and __pskb_pull_tail().
I applied this patchset and updated SCTP to also use it for data chunks. My tests results were very similar to what I had without it. Varying to better or worse, tending worse. Thing is, SCTP always works on linearized skbs as it can't crawl on fragments, so those clone/trim operations are just offset adjusts regarding the data, and it's shared. With pskb_extract, it implies in a new memory allocation and a copy, even in this best case, so for SCTP, for now, it's actually a drawback I'm afraid. Marcelo
