The XDP CPU destination array/set, configured via xdp-cpu-redirect,
will always be fairly small. My different benchmarking showed that
the current modulo hashing into the CPU array can easily result in bad
distribution, expecially if the number of CPU is an even number.
This patch uses a proper hashing function on the input key. The key
used for hashing is inspired by the ippair hashing code in
src/tmqh-flow.c, and is based on the IP src + dst.
An important property is that the hashing is flow symmetric, meaning
that if the source and destintation gets swapped then the selected CPU
will remain the same. This is important for Suricate.
That hashing INITVAL (15485863 the 10^6th prime number) was fairly
arbitrary choosen, but experiments with kernel tree pktgen scripts
(pktgen_sample04_many_flows.sh +pktgen_sample05_flow_per_thread.sh)
showed this improved the distribution.
Signed-off-by: Jesper Dangaard Brouer <netoptimizer@brouer.com>
Adjusted function call API to take an initval. This allow the API
user to set the initial value, as a seed. This could also be used for
inputting the previous hash.
Signed-off-by: Jesper Dangaard Brouer <brouer@redhat.com>
This patch adds a boolean option "xdp-cpu-redirect" to af-packet
interface configuration. If set, then the XDP filter will load
balance the skb creation on specified CPUs instead of doing the
creation on the CPU handling the packet. In the case of a card
with asymetric hashing this will allow to avoid saturating the
single CPU handling the trafic.
The XDP filter must contains a set of map allowing load balancing.
This is the case of xdp_filter.bpf.
Fixed-by: Jesper Dangaard Brouer <netoptimizer@brouer.com>
eBPF has a data type which is a per CPU array. By adding one element
to the array it is in fact added to all per CPU arrays in the kernel.
This allows to have a lockless structure in the kernel even when doing
counter update.
In userspace, we need to update the flow bypass code to fetch all
elements of the per CPU arrays.
This patch implements bypass capability for af-packet.
The filter only bypass TCP and UDP in IPv4 and IPv6. It don't
don't bypass IPv6 with extended headers.
This patch also introduces a bypassed flow manager that takes
care of timeouting the bypassed flows. It uses a 60 sec
timeout on flow. As they are supposed to be active we can
try that. If they are not active then we don't care to get them
back in Suricata.
This patch introduces the ebpf cluster mode. This mode is using
an extended BPF function that is loaded into the kernel and
provide the load balancing.
An example of cluster function is provided in the ebpf
subdirectory and provide ippair load balancing function.
This is a function which uses the same method as
the one used in autofp ippair to provide a symetrical
load balancing based on IP addresses.
A simple filter example allowing to drop IPv6 is added to the
source.
This patch also prepares the infrastructure to be able to load
and use map inside eBPF files. This will be used later for flow
bypass.
Eric Leblond
Jesper Dangaard Brouer
Jesper Dangaard Brouer