Re: [RFC PATCH 1/2] regmap: add configurable lock class key for lockdep

From: Mark Brown <broonie@kernel.org>
To: Lars-Peter Clausen <lars@metafoo.de>
Cc: Nicolas Boichat <drinkcat@chromium.org>,
	Arjan van de Ven <arjan@linux.intel.com>,
	Mauro Carvalho Chehab <mchehab@osg.samsung.com>,
	Antti Palosaari <crope@iki.fi>, Ingo Molnar <mingo@redhat.com>,
	Greg Kroah-Hartman <gregkh@linuxfoundation.org>,
	linux-kernel@vger.kernel.org, Bard Liao <bardliao@realtek.com>,
	Oder Chiou <oder_chiou@realtek.com>,
	Liam Girdwood <lgirdwood@gmail.com>,
	Jaroslav Kysela <perex@perex.cz>, Takashi Iwai <tiwai@suse.de>,
	alsa-devel@alsa-project.org,
	Anatol Pomozov <anatol.pomozov@gmail.com>
Subject: Re: [RFC PATCH 1/2] regmap: add configurable lock class key for lockdep
Date: Mon, 29 Jun 2015 15:19:40 +0100	[thread overview]
Message-ID: <20150629141940.GD11162@sirena.org.uk> (raw)
In-Reply-To: <5591414D.6080802@metafoo.de>

[-- Attachment #1: Type: text/plain, Size: 924 bytes --]

On Mon, Jun 29, 2015 at 02:59:57PM +0200, Lars-Peter Clausen wrote:

> Maybe we are just over-thinking this and should just add one key to each
> regmap instance. That solves the issue without requiring the any user
> interaction. The only downside is that it might impact the performance of
> lockdep and uses quite a few lock classes. Its probably manageable right now
> but could grow into a problem as regmap adoption further progresses. But
> maybe we can leave the hard work of figuring a better solution to our future
> selves.

I thought the issue with that was that all lockdep classes need to be
allocated statically?  If we can just do something that scales with
regmap instances I'm not sure there's a big problem, even if adoption
increases we're talking about something that scales with the number of
devices in the system rather than the number of drivers using the API
which should be a lot more manageable.

[-- Attachment #2: Digital signature --]
[-- Type: application/pgp-signature, Size: 473 bytes --]