Commits
Alexey Kuznetsov authored and Konstantin Khorenko committed 137e8807d5b
net: zerocopy over unix sockets Observation is that af_unix sockets today became slower and eat a lot of more cpu than 100G ethernet. So, implement MSG_ZEROCOPY over af_unix sockets to be able to talk to local services without collapse of performance. Unexpectedly, this makes sense! F.e. zerocopy cannot be done in TCP over loopback, because skbs when passing over loopback change ownership. But unix sockets traditionally implemented in the way that skbs queued on destination sockets still preserve ownership of source socket, so that completion notifications can be delivered back to sender without problems. Code looks neat and clean and might be submitted to mainstream one day. A little bit of dirt is that it shares IP_RECVERR control message, but rather this dirt is not mine, but by original author, who used protocol level message for generic socket level notifications without any good reasons, creating a mess for users. https://pmc.acronis.work/browse/VSTOR-79527 Signed-off-by: Alexey Kuznetsov <kuznet@acronis.com> Feature: net: zerocopy over unix sockets