Re: [PATCH v2] docs: Explain the desired position of function attributes

On Thu, Sep 30, 2021 at 01:11:34PM -0700, Randy Dunlap wrote:

On 9/30/21 12:24 PM, Kees Cook wrote:

quoted

While discussing how to format the addition of various function
attributes, some "unwritten rules" of ordering surfaced[1]. Capture as
close as possible to Linus's preferences for future reference.

(Though I note the dissent voiced by Joe Perches, Alexey Dobriyan, and
others that would prefer all attributes live on a separate leading line.)

[1] https://lore.kernel.org/mm-commits/CAHk-=wiOCLRny5aifWNhr621kYrJwhfURsa0vFPeUEm8mF0ufg@mail.gmail.com/ (local)

Signed-off-by: Kees Cook <redacted>
---
  Documentation/process/coding-style.rst | 30 ++++++++++++++++++++++++++
  1 file changed, 30 insertions(+)

diff --git a/Documentation/process/coding-style.rst b/Documentation/process/coding-style.rst
index 42969ab37b34..6b4feb1c71e7 100644
--- a/Documentation/process/coding-style.rst
+++ b/Documentation/process/coding-style.rst

@@ -487,6 +487,36 @@ because it is a simple way to add valuable information for the reader.
  Do not use the ``extern`` keyword with function prototypes as this makes
  lines longer and isn't strictly necessary.
+When writing a function declarations, please keep the `order of elements regular
+<https://lore.kernel.org/mm-commits/CAHk-=wiOCLRny5aifWNhr621kYrJwhfURsa0vFPeUEm8mF0ufg@mail.gmail.com/>`_.
+For example::
+
+ extern __init void * __must_check void action(enum magic value, size_t size,

Drop that second "void" ?  or what does it mean?
Can __must_check and void be used together?

Gah, thanks. Fixed now in v3.

quoted

+ 	u8 count, char *fmt, ...) __printf(4, 5) __malloc;
+
+The preferred order of elements for a function prototype is:
+
+- storage class (here, ``extern``, and things like ``static __always_inline`` even though
+  ``__always_inline`` is technically an attribute, it is treated like ``inline``)
+- storage class attributes (here, ``__init`` -- i.e. section declarations, but also things like ``__cold``)
+- return type (here, ``void *``)
+- return type attributes (here, ``__must_check``)

I'm not trying to get you to change this, but I would prefer to see

extern __init __must_check void *action(...) <attributes>;

i.e., with the return type adjacent to the function name.

I have read and re-read Linus's emails, and did a frequency count in the
kernel, and it looks like the preference is [return type] [return type attrs]
but I personally agree with you. :)

# regex I built from __must_check hits...
$ re='((struct .*|void|char) \* ?|((unsigned )?(long|int)|bool|size_t)($| ))'

# type before __must_check
$ git grep -E "$re"'__must_check' | wc -l
746

# type after __must_check
$ git grep -E '\b(static (__always_)?inline )?__must_check($| '"$re"')' | wc -l
297

# type split(!) across __must_check or otherwise weird...
$ git grep -E '\b__must_check\b' | grep -Ev '\b(static (__always_)?inline )?__must_check($| '"$re"')' | grep -Ev "$re"'__must_check\b' | wc -l
44


-- 
Kees Cook