Re: [PATCH v2 4/4] coding-style: add explanation about pr_fmt macro

[Jonathan Corbet]

On Fri, 13 Sep 2019 19:03:00 -0300
Andrà Almeida <andrealmeid@xxxxxxxxxxxxx> wrote:

> The pr_fmt macro is useful to format log messages printed by pr_XXXX()
> functions. Add text to explain the purpose of it, how to use and an
> example.

So I've finally had a chance to take a real look at this...

> diff --git a/Documentation/process/coding-style.rst b/Documentation/process/coding-style.rst
> index f4a2198187f9..1a33a933fbd3 100644
> --- a/Documentation/process/coding-style.rst
> +++ b/Documentation/process/coding-style.rst
> @@ -819,7 +819,15 @@ which you should use to make sure messages are matched to the right device
>  and driver, and are tagged with the right level:  dev_err(), dev_warn(),
>  dev_info(), and so forth.  For messages that aren't associated with a
>  particular device, <linux/printk.h> defines pr_notice(), pr_info(),
> -pr_warn(), pr_err(), etc.
> +pr_warn(), pr_err(), etc. It's possible to format pr_XXX() messages using the
> +macro pr_fmt() to prevent rewriting the style of messages. It should be
> +defined before ``#include <linux/kernel.h>``, to avoid compiler warning about
> +redefinitions, or just use ``#undef pr_fmt``. This is particularly useful for
> +adding the name of the module at the beginning of the message, for instance:
> +
> +.. code-block:: c
> +
> +        #define pr_fmt(fmt) KBUILD_MODNAME ": " fmt

Honestly, I think that this is out of scope for a document on coding
style.  That document is already far too long for most people to read, I
don't think we should load it down with more stuff that isn't directly
style related.

That said, the information can be useful.  I wanted to say that it should
go with the documentation of the pr_* macros but ... well ... um ... we
don't seem to have a whole lot of that.  Figures.

I suspect this is more than you wanted to sign up for, but...IMO, the right
thing to do is to fill printk.h with a nice set of kerneldoc comments
describing how this stuff should be used, then to pull that information
into the core-api manual, somewhere near our extensive discussion of printk
formats.  It's amazing that we lack docs for something so basic.

Thanks,

jon