linux-kernel - Re: regmap I3C 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] [thread-next>] [day] [month] [year] [list]

Message-ID: <88f34137-b215-4bee-b117-3ff00402ba6c@roeck-us.net>
Date: Wed, 13 Nov 2024 12:16:37 -0800
From: Guenter Roeck <linux@...ck-us.net>
To: Mark Brown <broonie@...nel.org>
Cc: "linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>,
 Frank Li <Frank.Li@....com>
Subject: Re: regmap I3C support

On 11/13/24 06:41, Mark Brown wrote:
> On Wed, Nov 13, 2024 at 06:15:38AM -0800, Guenter Roeck wrote:
>> On 11/13/24 06:01, Mark Brown wrote:
> 
>>> Don't these drivers end up with the same miserable problems with
>>> dependencies on variations of things being configured built in and
>>> modules anyway that mean we build separate SPI and I2C bus wrappers for
>>> the same case with devices that do both I2C and SPI?
> 
>> Not really. There is no equivalent to module_i3c_i2c_driver() to handle both
>> I2C and SPI variants of a chip. Also, SPI and I2C/I3C are not interdependent,
> 
> Sure, but lots of drivers were open coding an equivalent of that
> (possibly some still do).
> 
>> while I3C automatically selects I2C. That means it does make sense to handle
>> I2C and I3C in the same driver, but not I2C and SPI.
> 
> In terms of the devices they're very much the same and interdependent -
> it's generally one IP block and one set of pins that's doing both I2C
> and SPI, with nothing software visible different.  If I3C selects I2C
> then that does eliminate some of the problem space, I can't remember the
> speciifcs of how people (I think mainly randconfig people?) were
> breaking things.  You at least have I2C=y I3C=m which means that
> dependencies have to force the users modular.
> 
>> Sure, separate wrappers can be used, but that makes module_i3c_i2c_driver()
>> pointless.
> 
> That's kind of my question.  If we are going to have this sort of stuff
> we should also have it for I2C and SPI since it's such a common pattern.

I'll take that as "we are not going to provide a dummy regmap i3c registration
function". Thanks, that is all I needed to know.

Guenter