Re: [PATCH v5] printk: Userspace format enumeration support

[Petr Mladek <pmladek@suse.com>]

On Tue 2021-03-16 14:28:12, Chris Down wrote:
> Rasmus Villemoes writes:
> > I think it's pointless renaming the symbol to _printk, with all the
> > churn and reduced readability that involves (especially when reading
> > assembly "why are we calling _printk and not printk here?"). There's
> > nothing wrong with providing a macro wrapper by the same name
> > 
> > #define printk(bla bla) ({ do_stuff; printk(bla bla); })
> > 
> > Only two places would need to be updated to surround the word printk in
> > parentheses to suppress macro expansion: The declaration and the
> > definition of printk. I.e.
> > 
> > int (printk)(const char *s, ...)
> 
> Hmm, I'm indifferent to either. Personally I don't like the ambiguity of
> having both a macro and function share the same name and having to think
> "what's the preprocessor context here?".

I would prefer to keep _printk. I agree that it creates some churn but
it is easier to see what is going on. Also cscope is able to
find the right thing.

Otherwise, Rasmus, thanks a lot for the review and great hints
about the macro storing the metadata into the elf section.
I am not familiar with these things.

Best Regards,
Petr