Not sure it was sent so trying again...
On 7/5/20 8:28 PM, Cong Wang wrote:
On Sun, Jul 5, 2020 at 2:50 PM Jamal Hadi Salim <j...@mojatatu.com> wrote:
BTW, nothing in skbedit is against computing what the new metadata
should be.
Yup.
IMO: A good arguement to not make it part of skbedit is if it adds
unnecessary complexity to skbedit or policy definitions.
TCA_HASH_ALG_L4 literally has 4 lines of code, has no way
to add any unnecessary complexity to skbedit. (The BPF algorithm
does not belong to skbedit, obviously.)
Thanks.
Moving TCA_HASH_ALG_L4 to skbedit is very simple, I agree.
However, supporting the bpf option via act_bpf is still problematic as
it is not offloadable.
We need some kind of indication that this is a hash computation program
and therefore
it requires specific identifier in the form of a new action.
Ariel
