Some platforms a single ethernet device for all ports with multiple rx rings
and NAPI threading enabled. In this case, the steering script was limiting
performance by keeping all NAPI threads assigned to the same CPU.
Fix this by assigning each rx queue and the corresponding NAPI task separately.
Additionally, if the number of rx queues is at least as big as the number of
CPUs, skip weight based assignment and distribute the load across all CPUs
directly.
Fixes: https://github.com/openwrt/openwrt/issues/17611
Signed-off-by: Felix Fietkau <nbd@nbd.name>
The new script uses a different strategy compared to the previous one.
Instead of trying to split flows by hash and spread them to all CPUs,
use RPS to redirect packets to a single core only.
Try to spread NAPI thread and RPS target CPUs across available CPUs
and try to ensure that the NAPI thread is on a different CPU than the
RPS target. This significantly reduces cycles wasted on the scheduler.
Signed-off-by: Felix Fietkau <nbd@nbd.name>
ifconfig is effectively deprecated for quite some time now. Let's
replace the remaining occurrences for packages by the
corresponding ip commands now.
Signed-off-by: Adrian Schmutzler <freifunk@adrianschmutzler.de>
