New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
bpf, xdp: various optimizations for nodeport #11082
Conversation
If we do not have prefilter configured, then lets spare the cycles and also not have it needlessly compiled in. Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Allow GRO to operate naturally rather than potentially breaking a TCP pkt train in the middle of aggregation due to meta data mismatch. This does not matter at all in case we do not push up the stack, so do not perform unneeded work for ctx_act_{drop,tx}. Verified that before we had 24 bytes in skb meta data and afterwards 0 bytes. Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Make use of the meta data feature to signal to the skb layer that we've seen the xdp packet during nodeport handling and that it's a non-service packet where in tc ingress we can then bypass the entire nodeport processing (key extraction, CT lookup, rev NAT) since already done from XDP. Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
d785ccd
to
4059541
Compare
... which then also shortens cycles on GRO for comparing skbs when aggregating. Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
test-me-please |
test-gke |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
lgtm
(ci green) |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
LGTM
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
👍
See commit msg.