Re: [PATCH RFC] modpost: Support I2C Aliases from OF tables

["Enrico Weigelt, metux IT consult" <lkml@metux.net>]

On 08.08.19 15:24, Andy Shevchenko wrote:
> On Thu, Aug 08, 2019 at 03:12:47PM +0200, Enrico Weigelt, metux IT consult wrote:
>> On 06.08.19 19:12, Javier Martinez Canillas wrote:
>>
>>> Right, we could add a macro for that. Although it should probably be called
>>> I2C_OF_MODULE_DEVICE_TABLE() or something like that since is specific to OF.
>>
>> At that point it should be completely noop when OF is disabled, so we
>> also can get rid of many ifdef's.
> 
> Why?

For cases where drivers work w/ or w/o oftree. Not sure whether it
applies to i2c specifically, but there're other places where we still
need nasty ifdef's (eg. gpio-keyboard).

>> I've got some patch somewhere for introducing a MODULE_OF_TABLE() macro
>> as replacement for many MODULE_DEVICE_TABLE(of, ...) cases, which noops
>> when CONFIG_OF is disabled. (and similar ones for other table types).
> 
> It's simple wrong to have #ifdef CONFIG_OF without counterpart of_match_ptr().

Of course, but that's just a part of the story. (actually I'd prefer
using it everywhere, even if the driver only supports oftree).

> And taking into consideration that ID table itself doesn't depend to OF at all,
> why not simple drop that #ifdef and of_match_ptr() all together?

Consumes less space. Yes, it isn't much, but in some scenarios one needs
to heavily reduce the kernel size. And I wouldn't like to use
of_match_ptr() inside a MODULE_DEVICE_TABLE() call :o


--mtx

-- 
Enrico Weigelt, metux IT consult
Free software and Linux embedded engineering
info@metux.net -- +49-151-27565287