netdev - Re: [RESEND PATCH v2 bpf-next 02/12] bpf: tcp: Make sure iter->batch always contains a full bucket snapshot

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <aFnIxcsROSNowexy@mini-arch>
Date: Mon, 23 Jun 2025 14:36:05 -0700
From: Stanislav Fomichev <stfomichev@...il.com>
To: Jordan Rife <jordan@...fe.io>
Cc: netdev@...r.kernel.org, bpf@...r.kernel.org,
	Daniel Borkmann <daniel@...earbox.net>,
	Martin KaFai Lau <martin.lau@...ux.dev>,
	Willem de Bruijn <willemdebruijn.kernel@...il.com>,
	Alexei Starovoitov <alexei.starovoitov@...il.com>
Subject: Re: [RESEND PATCH v2 bpf-next 02/12] bpf: tcp: Make sure iter->batch
 always contains a full bucket snapshot

On 06/23, Jordan Rife wrote:
> > Untested code to illustrate the idea below. Any reason it won't work?
> 
> In theory, I like the idea of unrolling the code a bit here to make
> the flow more clear (and to make it clear what's happening to the
> locks!). IIRC there was some reason this was hard, but I will think
> about it a bit again.
> 
> I also want to make sure things stay relatively consistent between the
> UDP and TCP socket iterator code structure. The UDP socket iterators
> already do the `goto fill_batch` and `goto again` thing, which is
> where I borrowed this from. If we end up diverging here, I'd want to
> go back and update the UDP code as well.
> 
> Thanks for the suggestion. I'll take a closer look a bit later and see
> if I can work this in. In the meantime, hopefully Martin can chime in
> as well. We went back and forth on the code structure quite a bit in
> the patch series for UDP socket iterators, so he might have some
> opinions here.

Martin is OOO so you'll have to wait a bit for his feedback.
UDP iterator seems to be more low level to me (with explicit locking),
so maybe all this non-unrolled retry logic there is justified, but
I haven't looked too deep.