[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <20220611201117.euqca7rgn5wydlwk@macbook-pro-3.dhcp.thefacebook.com>
Date: Sat, 11 Jun 2022 13:11:17 -0700
From: Alexei Starovoitov <alexei.starovoitov@...il.com>
To: Lorenzo Bianconi <lorenzo@...nel.org>
Cc: bpf@...r.kernel.org, netdev@...r.kernel.org, ast@...nel.org,
daniel@...earbox.net, andrii@...nel.org, davem@...emloft.net,
kuba@...nel.org, edumazet@...gle.com, pabeni@...hat.com,
pablo@...filter.org, fw@...len.de, netfilter-devel@...r.kernel.org,
lorenzo.bianconi@...hat.com, brouer@...hat.com, toke@...hat.com,
memxor@...il.com, yhs@...com
Subject: Re: [PATCH v4 bpf-next 00/14] net: netfilter: add kfunc helper to
update ct timeout
On Thu, May 26, 2022 at 11:34:48PM +0200, Lorenzo Bianconi wrote:
> Changes since v3:
> - split bpf_xdp_ct_add in bpf_xdp_ct_alloc/bpf_skb_ct_alloc and
> bpf_ct_insert_entry
> - add verifier code to properly populate/configure ct entry
> - improve selftests
Kumar, Lorenzo,
are you planning on sending v5 ?
The patches 1-5 look good.
Patch 6 is questionable as Florian pointed out.
What is the motivation to allow writes into ct->status?
The selftest doesn't do that anyway.
Patch 7 (acquire-release pairs) is too narrow.
The concept of a pair will not work well. There could be two acq funcs and one release.
Please think of some other mechanism. Maybe type based? BTF?
Or encode that info into type name? or some other way.
Powered by blists - more mailing lists