linux-kernel - Re: [PATCH v1 0/3] cgroup: Add lock guard support

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] [day] [month] [year] [list]

Message-ID: <20250623140321.GA3372@cmpxchg.org>
Date: Mon, 23 Jun 2025 16:03:21 +0200
From: Johannes Weiner <hannes@...xchg.org>
To: Tejun Heo <tj@...nel.org>
Cc: Jemmy Wong <jemmywong512@...il.com>,
	Michal Koutný <mkoutny@...e.com>,
	Waiman Long <longman@...hat.com>, cgroups@...r.kernel.org,
	linux-kernel@...r.kernel.org
Subject: Re: [PATCH v1 0/3] cgroup: Add lock guard support

On Fri, Jun 20, 2025 at 04:52:03PM -1000, Tejun Heo wrote:
> On Fri, Jun 20, 2025 at 06:45:54PM +0800, Jemmy Wong wrote:
> ...
> > > Tejun:
> > >> There are no practical benefits to converting the code base at this point.
> > > 
> > > I'd expect future backports (into such code) to be more robust wrt
> > > pairing errors.
> > > At the same time this is also my biggest concern about this change, the
> > > wide-spread diff would make current backporting more difficult.  (But
> > > I'd counter argue that one should think forward here.)
> 
> Well, I'm not necessarily against it but I generally dislike wholesale
> cleanups which create big patch application boundaries. If there are enough
> practical benefits, sure, we should do it, but when it's things like this -
> maybe possibly it's a bit better in the long term - the calculus isn't clear
> cut. People can argue these things to high heavens on abstract grounds, but
> if you break it down to practical gains vs. costs, it's not a huge
> difference.
> 
> But, again, I'm not against it. Johannes, any second thoughts?

Yeah, letting the primitives get used organically in new code and
patches sounds better to me than retrofitting it into an existing
function graph that wasn't designed with these in mind.