Received-SPF: pass (google.com: domain of linux-kernel+bounces-78431-linux.lists.archive=gmail.com@vger.kernel.org designates 2604:1380:4601:e00::3 as permitted sender) client-ip=2604:1380:4601:e00::3;
Date: Fri, 23 Feb 2024 14:55:01 +0100
From: "Jason A. Donenfeld" <Jason@zx2c4.com>
To: "Matthew Wilcox (Oracle)" <willy@infradead.org>
Cc: linux-kernel@vger.kernel.org, Thomas Graf <tgraf@suug.ch>,
	Herbert Xu <herbert@gondor.apana.org.au>, netdev@vger.kernel.org,
	linux-fsdevel@vger.kernel.org, maple-tree@lists.infradead.org,
	rcu@vger.kernel.org
Subject: Re: [PATCH 0/1] Rosebush, a new hash table
Message-ID: <ZdijtQNFuziHnqH2@zx2c4.com>
References: <20240222203726.1101861-1-willy@infradead.org>
Precedence: bulk
MIME-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Disposition: inline
In-Reply-To: <20240222203726.1101861-1-willy@infradead.org>

Hi Matthew,

On Thu, Feb 22, 2024 at 08:37:23PM +0000, Matthew Wilcox (Oracle) wrote:
> Rosebush is a resizing, scalable, cache-aware, RCU optimised hash table.
> I've written a load of documentation about how it works, mostly in
> Documentation/core-api/rosebush.rst but some is dotted through the
> rosebush.c file too.

If you're interested, WireGuard has some pretty primitive hashtables,
for which maybe Rosebush would be an interesting replacement:

https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/drivers/net/wireguard/peerlookup.c
https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/drivers/net/wireguard/peerlookup.h#n17
https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/drivers/net/wireguard/ratelimiter.c#n167

In peerlookup.c, note the "At the moment, we limit" comment for an idea
of some of the hairy issues involved in replacing these. But I wouldn't
be entirely opposed to it, if you see some interesting potential for
Rosebush here. That's a very tentative interest -- maybe it won't work
out in the end -- but nonetheless, seeing this piqued my curiosity. If
you're looking to see how this behaves in a place beyond dcache, this
might be something to play with.

Jason