Received-SPF: pass (google.com: domain of linux-kernel+bounces-54667-linux.lists.archive=gmail.com@vger.kernel.org designates 2604:1380:40f1:3f00::1 as permitted sender) client-ip=2604:1380:40f1:3f00::1;
Precedence: bulk
MIME-Version: 1.0
References: <20240206071314.8721-1-liubo03@inspur.com> <CAMuHMdU7fYCsNT9ditqJ-saUsRm9J2zLh=-q-zmExhBRJeE8NQ@mail.gmail.com>
 <ZcIE/RMTq34TgpQt@finisterre.sirena.org.uk>
In-Reply-To: <ZcIE/RMTq34TgpQt@finisterre.sirena.org.uk>
From: Geert Uytterhoeven <geert@linux-m68k.org>
Date: Tue, 6 Feb 2024 11:14:11 +0100
Message-ID: <CAMuHMdVj1bS9s69ASrd5xULc8oELoBbnb8HEX9MEmA43853_EQ@mail.gmail.com>
Subject: Re: [PATCH 00/18] mfd: convert to use maple tree register cache
To: Mark Brown <broonie@kernel.org>
Cc: Bo Liu <liubo03@inspur.com>, lee@kernel.org, wens@csie.org, 
	marek.vasut+renesas@gmail.com, support.opensource@diasemi.com, 
	neil.armstrong@linaro.org, ckeepax@opensource.cirrus.com, 
	rf@opensource.cirrus.com, mazziesaccount@gmail.com, mcoquelin.stm32@gmail.com, 
	alexandre.torgue@foss.st.com, linux-kernel@vger.kernel.org, 
	linux-renesas-soc@vger.kernel.org, linux-amlogic@lists.infradead.org, 
	patches@opensource.cirrus.com, linux-stm32@st-md-mailman.stormreply.com, 
	linux-arm-kernel@lists.infradead.org
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable

Hi Mark,

On Tue, Feb 6, 2024 at 11:09=E2=80=AFAM Mark Brown <broonie@kernel.org> wro=
te:
> On Tue, Feb 06, 2024 at 10:33:22AM +0100, Geert Uytterhoeven wrote:
> > If all of this is true, is there any reason to keep REGCACHE_RBTREE
> > around?  If not, perhaps REGCACHE_RBTREE should be treated as
> > REGCACHE_MAPLE in the regmap core code first, followed by a single
> > tree-wide patch to replace REGCACHE_RBTREE?
>
> There is a very small niche for devices where cache syncs are a
> particularly important part of the workload where rbtree's choices might
> give better performance, especially on systems with low end CPUs.

The REGCACHE_* value is specified by the device, not by the CPU?
While some of these MFD devices are on-SoC, and thus there is some
relation between device and CPU, several others (e.g. PMICs) are
external, and thus might be present on systems with a variety of CPU
performance.

Perhaps the value should depend on some CPU heuristic instead?

Gr{oetje,eeting}s,

                        Geert

--=20
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@linux-m68k=
org

In personal conversations with technical people, I call myself a hacker. Bu=
t
when I'm talking to journalists I just say "programmer" or something like t=
hat.
                                -- Linus Torvalds