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: <0779827f-c51c-4407-8544-c87ca6d76500@roeck-us.net>
Date: Tue, 19 Nov 2024 13:57:38 -0800
From: Guenter Roeck <linux@...ck-us.net>
To: Frank Li <Frank.li@....com>
Cc: Mark Brown <broonie@...nel.org>,
 "linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>
Subject: Re: regmap I3C support

On 11/19/24 11:30, Frank Li wrote:
> On Tue, Nov 19, 2024 at 10:41:05AM -0800, Guenter Roeck wrote:
>> On 11/19/24 09:46, Mark Brown wrote:
>>> On Fri, Nov 15, 2024 at 08:35:04PM -0800, Guenter Roeck wrote:
>>>> On 11/14/24 09:26, Mark Brown wrote:
>>>
>>>>> Right, so the fact that I3C depends on I2C deals with a lot of the
>>>>> problems that plague the I2C/SPI combination.  Ugh.  I guess the helper
>>>>> should be OK and there's not much doing for I2C/SPI.
>>>
>>>> It looks like we can use
>>>
>>>>          if (IS_ENABLED(CONFIG_REGMAP_I3C)) {
>>>>                  regmap = devm_regmap_init_i3c(i3cdev, &tmp108_regmap_config);
>>>>                  if (IS_ERR(regmap))
>>>>                          return dev_err_probe(dev, PTR_ERR(regmap),
>>>>                                               "Failed to register i3c regmap\n");
>>>> 		...
>>>> 	}
>>>
>>>> even if a stub function is not available as long as there is an external
>>>> declaration.
>>>
>>>> I don't really like it, but it turns out that this kind of code is already used
>>>> elsewhere in the kernel. It looks like dead code elimination can now assumed
>>>> to be available when building kernel code. We live and learn.
>>>
>>> Ah, that solves that problem then I guess?
>>
>>
>> Yes. It actually goes a step further - the IS_ENABLED(CONFIG_REGMAP_I3C)) in the
>> probe function isn't needed either because the entire i3c probe function is
>> optimized away if CONFIG_I3C=n.
>>
>> I'll send a patch dropping the #ifdef in the tmp108 driver after the commit
>> window closes.
> 
> Already tried this at v3
> https://lore.kernel.org/imx/7bdd2db8-41c8-43d8-ae73-84a221d2d004@roeck-us.net/
> 

Yes, I know. Sorry for that. We live and learn. I didn't think this works,
but it does.

> but I am not sure if it is good on rely on the compiler. Maybe some option
> like some debug option or -O0 cause problem.
> 

Yes, I thought so too, but it turns out that the kernel doesn't build anymore
with -O0 anyway, and other code already _does_ depend on dead code elimination.

Guenter