Re: [PATCH NAND 0/5] Replace printk statements with pr_*macros

[Boris Brezillon <boris.brezillon@bootlin.com>]

Hi Richard,

On Fri, 16 Feb 2018 18:19:40 +0100
Richard Weinberger <richard@nod.at> wrote:

> Am Freitag, 16. Februar 2018, 17:50:09 CET schrieb Shreeya Patel:
> > This patchset removes all the log levels i.e. KERN_WARN,
> > KERN_NOTICE, KERN_ERR, KERN_INFO, KERN_DEBUG used in the printk
> > statements and replaces the printk statements with appropriate
> > pr_*macros.
> > According to the kernel coding style, pr_*macro is the preferred
> > way to print the message.  
> 
> Beside of that, how does it improve the code?
> Don't get me wrong, pr_* is the way to go for new code, but I don't think it 
> is worth "fixing" in existing code and make working with git blame more 
> painful.

That's true about git blame, though I doubt we'll find a bug in the
printk message, and IIRC, there's an option to find the original
offender even when cosmetic changes have happened in between.

Now, I'd like to explain a bit why I sometimes accept cosmetic/coding
style patches and why I sometimes don't. When this is a first 
contribution, I consider this as a way for new comers to learn the
patch submission/review process. But as soon as I start receiving a lot
of cosmetic changes from the same contributor and I see the person does
not move to more useful contributions I stop accepting her/his
patches. I also tend to take patches from people who write coccinelle
scripts, because they try to make things consistent across the whole
source tree and most importantly, they know what they're doing :-).

Maybe I'm wrong in my reasoning, but I thought it was worth explaining
the decision process ;-).

Regards,

Boris

-- 
Boris Brezillon, Bootlin (formerly Free Electrons)
Embedded Linux and Kernel engineering
http://bootlin.com