linux-kernel - Re: [PATCH 04/11] lib/dlock-list: Make sibling CPUs share the same linked list

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: <20231207043129.cswpxpwow43b35lj@moria.home.lan>
Date:   Wed, 6 Dec 2023 23:31:29 -0500
From:   Kent Overstreet <kent.overstreet@...ux.dev>
To:     Dave Chinner <david@...morbit.com>
Cc:     linux-fsdevel@...r.kernel.org, linux-block@...r.kernel.org,
        linux-cachefs@...hat.com, dhowells@...hat.com,
        gfs2@...ts.linux.dev, dm-devel@...ts.linux.dev,
        linux-security-module@...r.kernel.org, selinux@...r.kernel.org,
        linux-kernel@...r.kernel.org
Subject: Re: [PATCH 04/11] lib/dlock-list: Make sibling CPUs share the same
 linked list

On Wed, Dec 06, 2023 at 05:05:33PM +1100, Dave Chinner wrote:
> From: Waiman Long <longman@...hat.com>
> 
> The dlock list needs one list for each of the CPUs available. However,
> for sibling CPUs, they are sharing the L2 and probably L1 caches
> too. As a result, there is not much to gain in term of avoiding
> cacheline contention while increasing the cacheline footprint of the
> L1/L2 caches as separate lists may need to be in the cache.
> 
> This patch makes all the sibling CPUs share the same list, thus
> reducing the number of lists that need to be maintained in each
> dlock list without having any noticeable impact on performance. It
> also improves dlock list iteration performance as fewer lists need
> to be iterated.
> 
> Signed-off-by: Waiman Long <longman@...hat.com>
> Reviewed-by: Jan Kara <jack@...e.cz>

We badly need this done in a more generic way. Besides shared caches,
I've done a bunch of percpu algorithms where "amount of x stranded on
percpu lists" is a major consideration and this would be preferable over
percpu lists (including in fs/aio.c).