Re: [PATCH nft 0/4] add operation cache for timestamp

[Pablo Neira Ayuso <pablo@xxxxxxxxxxxxx>]

On Fri, Aug 25, 2023 at 03:24:16PM +0200, Thomas Haller wrote:
> Add a cache for "time(NULL)" and tm_gmtoff from localtime_r(time(NULL), &tm).
> The point is to ensure that one parse/output operation fetches the current time
> and GMT offset at most once.
> 
> Follow up to ([1])
> 
>   Subject:  Re: [nft PATCH 2/2] meta: use reentrant localtime_r()/gmtime_r() functions
>   Date:     Tue, 22 Aug 2023 17:15:14 +0200
> 
> [1] https://marc.info/?l=netfilter-devel&m=169271724629901&w=4

To extend what I said yesterday. It would be great if you could
validate that we have sufficient tests for time support.

Probably you can use this ruleset that I am attaching as reference and
think of a ruleset to cover this? I am attaching an example ruleset
which is basically a "timetable" using nftables sets/maps.

table netdev filter {
	map ether_to_chain {
		typeof ether saddr : verdict
		elements = { 96:68:97:a7:e8:a7 comment "Device match" : jump fw_p0_dev0 }
	}

	map schedule_time {
		typeof meta time : verdict
		flags interval
		counter
		elements = { "2022-10-09 18:46:50" - "2022-10-09 19:16:50" comment "!Schedule OFFLINE override" : drop }
	}

	map schedule_day {
		typeof meta day . meta hour : verdict
		flags interval
		counter
		elements = { "Tuesday" . "06:00" - "07:00" : drop }
	}

	chain fw_p0_dev0 {
		meta time vmap @schedule_time
		meta day . meta hour vmap @schedule_day
	}

	chain my_devices_rules {
		ether saddr vmap @ether_to_chain
	}

	chain ingress {
		type filter hook ingress device eth0 priority filter; policy accept;
		jump my_devices_rules
	}
}