About multiple core bpf execution

[hazelnutsgz]

I am a newbie to bpf and TC, I was wondering if

1. When multiple cores are running the same ebpf program on ingress, let's say, int cls_entry(struct __sk_buff *skb) , would this introduce any troubles for updating of the bpf_elf_map?

2. Would updated_state.r1 = tnow - B or tmin = tnow - B; make any sense? (given the assumption that the timestamp of new packets arriving after this bpf execution would be always larger than tnow)

Thanks.