From: Barry Song <v-songbaohua@oppo.com>

Recent commit 77292bb8ca69c80 ("crypto: scomp - remove memcpy if
sg_nents is 1 and pages are lowmem") leads to warnings on xtensa
and loongarch,
   In file included from crypto/scompress.c:12:
   include/crypto/scatterwalk.h: In function 'scatterwalk_pagedone':
   include/crypto/scatterwalk.h:76:30: warning: variable 'page' set but not used [-Wunused-but-set-variable]
      76 |                 struct page *page;
         |                              ^~~~
   crypto/scompress.c: In function 'scomp_acomp_comp_decomp':
>> crypto/scompress.c:174:38: warning: unused variable 'dst_page' [-Wunused-variable]
     174 |                         struct page *dst_page = sg_page(req->dst);
         |

The reason is that flush_dcache_page() is implemented as a noop
macro on these platforms as below,

 #define flush_dcache_page(page) do { } while (0)

The driver code, for itself, seems be quite innocent and placing
maybe_unused seems pointless,

 struct page *dst_page = sg_page(req->dst);

 for (i = 0; i < nr_pages; i++)
 	flush_dcache_page(dst_page + i);

And it should be independent of architectural implementation
differences.

Let's provide guidance on coding style for requesting parameter
evaluation or proposing the migration to a static inline
function.

Signed-off-by: Barry Song <v-songbaohua@oppo.com>
Suggested-by: Max Filippov <jcmvbkbc@gmail.com>
Reviewed-by: Mark Brown <broonie@kernel.org>
Cc: Chris Zankel <chris@zankel.net>
Cc: Huacai Chen <chenhuacai@loongson.cn>
Cc: Herbert Xu <herbert@gondor.apana.org.au>
Cc: Guenter Roeck <linux@roeck-us.net>
Cc: Stephen Rothwell <sfr@canb.auug.org.au>
Cc: Andy Whitcroft <apw@canonical.com>
Cc: Dwaipayan Ray <dwaipayanray1@gmail.com>
Cc: Joe Perches <joe@perches.com>
Cc: Jonathan Corbet <corbet@lwn.net>
Cc: Lukas Bulwahn <lukas.bulwahn@gmail.com>
Cc: Xining Xu <mac.xxn@outlook.com>
---
 Documentation/process/coding-style.rst | 16 ++++++++++++++++
 1 file changed, 16 insertions(+)

diff --git a/Documentation/process/coding-style.rst b/Documentation/process/coding-style.rst
index 9c7cf7347394..791d333a57fd 100644
--- a/Documentation/process/coding-style.rst
+++ b/Documentation/process/coding-style.rst
@@ -827,6 +827,22 @@ Macros with multiple statements should be enclosed in a do - while block:
 				do_this(b, c);		\
 		} while (0)
 
+Function-like macros with unused parameters should be replaced by static
+inline functions to avoid the issue of unused variables:
+
+.. code-block:: c
+
+	static inline void fun(struct foo *foo)
+	{
+	}
+
+For historical reasons, many files still use the cast to (void) to evaluate
+parameters, but this method is not recommended:
+
+.. code-block:: c
+
+	#define macrofun(foo) do { (void) (foo); } while (0)
+
 Things to avoid when using macros:
 
 1) macros that affect control flow:
-- 
2.34.1

So I'm not sure what your desired path for getting this upstream is.  I
can take it, but I'm generally quite leery of taking coding-style
changes without some serious acks on them - nobody elected me as the
arbiter of proper coding style.

A nit below

Barry Song <21cnbao@gmail.com> writes:

> From: Barry Song <v-songbaohua@oppo.com>
>
> Recent commit 77292bb8ca69c80 ("crypto: scomp - remove memcpy if
> sg_nents is 1 and pages are lowmem") leads to warnings on xtensa
> and loongarch,
>    In file included from crypto/scompress.c:12:
>    include/crypto/scatterwalk.h: In function 'scatterwalk_pagedone':
>    include/crypto/scatterwalk.h:76:30: warning: variable 'page' set but not used [-Wunused-but-set-variable]
>       76 |                 struct page *page;
>          |                              ^~~~
>    crypto/scompress.c: In function 'scomp_acomp_comp_decomp':
>>> crypto/scompress.c:174:38: warning: unused variable 'dst_page' [-Wunused-variable]
>      174 |                         struct page *dst_page = sg_page(req->dst);
>          |
>
> The reason is that flush_dcache_page() is implemented as a noop
> macro on these platforms as below,
>
>  #define flush_dcache_page(page) do { } while (0)
>
> The driver code, for itself, seems be quite innocent and placing
> maybe_unused seems pointless,
>
>  struct page *dst_page = sg_page(req->dst);
>
>  for (i = 0; i < nr_pages; i++)
>  	flush_dcache_page(dst_page + i);
>
> And it should be independent of architectural implementation
> differences.
>
> Let's provide guidance on coding style for requesting parameter
> evaluation or proposing the migration to a static inline
> function.
>
> Signed-off-by: Barry Song <v-songbaohua@oppo.com>
> Suggested-by: Max Filippov <jcmvbkbc@gmail.com>
> Reviewed-by: Mark Brown <broonie@kernel.org>
> Cc: Chris Zankel <chris@zankel.net>
> Cc: Huacai Chen <chenhuacai@loongson.cn>
> Cc: Herbert Xu <herbert@gondor.apana.org.au>
> Cc: Guenter Roeck <linux@roeck-us.net>
> Cc: Stephen Rothwell <sfr@canb.auug.org.au>
> Cc: Andy Whitcroft <apw@canonical.com>
> Cc: Dwaipayan Ray <dwaipayanray1@gmail.com>
> Cc: Joe Perches <joe@perches.com>
> Cc: Jonathan Corbet <corbet@lwn.net>
> Cc: Lukas Bulwahn <lukas.bulwahn@gmail.com>
> Cc: Xining Xu <mac.xxn@outlook.com>
> ---
>  Documentation/process/coding-style.rst | 16 ++++++++++++++++
>  1 file changed, 16 insertions(+)
>
> diff --git a/Documentation/process/coding-style.rst b/Documentation/process/coding-style.rst
> index 9c7cf7347394..791d333a57fd 100644
> --- a/Documentation/process/coding-style.rst
> +++ b/Documentation/process/coding-style.rst
> @@ -827,6 +827,22 @@ Macros with multiple statements should be enclosed in a do - while block:
>  				do_this(b, c);		\
>  		} while (0)
>  
> +Function-like macros with unused parameters should be replaced by static
> +inline functions to avoid the issue of unused variables:
> +
> +.. code-block:: c

I would just use the "::" notation here; the ..code-block:: just adds
noise IMO.

> +	static inline void fun(struct foo *foo)
> +	{
> +	}
> +
> +For historical reasons, many files still use the cast to (void) to evaluate
> +parameters, but this method is not recommended:
> +
> +.. code-block:: c
> +
> +	#define macrofun(foo) do { (void) (foo); } while (0)
> +

1) If you're putting in examples of something *not* to do, it's probably
better to also put in something like:

   /* don't do this */

people don't always read closely.

2) Can we say *why* it's not recommended?

Thanks,

jon

On Wed, Apr 3, 2024 at 5:13 AM Jonathan Corbet <corbet@lwn.net> wrote:
>
> So I'm not sure what your desired path for getting this upstream is.  I
> can take it, but I'm generally quite leery of taking coding-style
> changes without some serious acks on them - nobody elected me as the
> arbiter of proper coding style.

Hi Jonathan,
Thanks!
Andrew previously integrated it into mm-nomm and tagged it as [TO-BE-UPDATED]
before removing it a few days back. Here's the link:
https://lore.kernel.org/mm-commits/20240330025857.CD609C433F1@smtp.kernel.org/
So if feasible, I'd prefer to stick with Andrew's channel.

>
> A nit below
>
> Barry Song <21cnbao@gmail.com> writes:
>
> > From: Barry Song <v-songbaohua@oppo.com>
> >
> > Recent commit 77292bb8ca69c80 ("crypto: scomp - remove memcpy if
> > sg_nents is 1 and pages are lowmem") leads to warnings on xtensa
> > and loongarch,
> >    In file included from crypto/scompress.c:12:
> >    include/crypto/scatterwalk.h: In function 'scatterwalk_pagedone':
> >    include/crypto/scatterwalk.h:76:30: warning: variable 'page' set but not used [-Wunused-but-set-variable]
> >       76 |                 struct page *page;
> >          |                              ^~~~
> >    crypto/scompress.c: In function 'scomp_acomp_comp_decomp':
> >>> crypto/scompress.c:174:38: warning: unused variable 'dst_page' [-Wunused-variable]
> >      174 |                         struct page *dst_page = sg_page(req->dst);
> >          |
> >
> > The reason is that flush_dcache_page() is implemented as a noop
> > macro on these platforms as below,
> >
> >  #define flush_dcache_page(page) do { } while (0)
> >
> > The driver code, for itself, seems be quite innocent and placing
> > maybe_unused seems pointless,
> >
> >  struct page *dst_page = sg_page(req->dst);
> >
> >  for (i = 0; i < nr_pages; i++)
> >       flush_dcache_page(dst_page + i);
> >
> > And it should be independent of architectural implementation
> > differences.
> >
> > Let's provide guidance on coding style for requesting parameter
> > evaluation or proposing the migration to a static inline
> > function.
> >
> > Signed-off-by: Barry Song <v-songbaohua@oppo.com>
> > Suggested-by: Max Filippov <jcmvbkbc@gmail.com>
> > Reviewed-by: Mark Brown <broonie@kernel.org>
> > Cc: Chris Zankel <chris@zankel.net>
> > Cc: Huacai Chen <chenhuacai@loongson.cn>
> > Cc: Herbert Xu <herbert@gondor.apana.org.au>
> > Cc: Guenter Roeck <linux@roeck-us.net>
> > Cc: Stephen Rothwell <sfr@canb.auug.org.au>
> > Cc: Andy Whitcroft <apw@canonical.com>
> > Cc: Dwaipayan Ray <dwaipayanray1@gmail.com>
> > Cc: Joe Perches <joe@perches.com>
> > Cc: Jonathan Corbet <corbet@lwn.net>
> > Cc: Lukas Bulwahn <lukas.bulwahn@gmail.com>
> > Cc: Xining Xu <mac.xxn@outlook.com>
> > ---
> >  Documentation/process/coding-style.rst | 16 ++++++++++++++++
> >  1 file changed, 16 insertions(+)
> >
> > diff --git a/Documentation/process/coding-style.rst b/Documentation/process/coding-style.rst
> > index 9c7cf7347394..791d333a57fd 100644
> > --- a/Documentation/process/coding-style.rst
> > +++ b/Documentation/process/coding-style.rst
> > @@ -827,6 +827,22 @@ Macros with multiple statements should be enclosed in a do - while block:
> >                               do_this(b, c);          \
> >               } while (0)
> >
> > +Function-like macros with unused parameters should be replaced by static
> > +inline functions to avoid the issue of unused variables:
> > +
> > +.. code-block:: c
>
> I would just use the "::" notation here; the ..code-block:: just adds
> noise IMO.

I am not quite sure we want this. as reading the whole coding-style.rst,
.. code-block:: c is everywhere :-)   Should I do something different or
just follow the tradition?

>
> > +     static inline void fun(struct foo *foo)
> > +     {
> > +     }
> > +
> > +For historical reasons, many files still use the cast to (void) to evaluate
> > +parameters, but this method is not recommended:
> > +
> > +.. code-block:: c
> > +
> > +     #define macrofun(foo) do { (void) (foo); } while (0)
> > +
>
> 1) If you're putting in examples of something *not* to do, it's probably
> better to also put in something like:
>
>    /* don't do this */
>
> people don't always read closely.

ok.

>
> 2) Can we say *why* it's not recommended?
>

Andrew makes a valid point.

https://lore.kernel.org/all/20240321104427.730b859087afecf5973d1c58@linux-foundation.org/

"I think so.  My overall view is that we should write things in C.  Only
use macros if the thing we're trying to do simply cannot be done in a C
function.

- inline functions don't have the "expression with side effects
  evaluated more than once" problem.

- inline functions avoid the unused-variable issue which started this thread

- inline functions look better

- for some reason, people are more inclined to document inline
  functions than macros."

Andrew's point seems too lengthy for inclusion in the coding-style.rst document?
I'll attempt to condense it.

> Thanks,
>
> jon

Thanks
Barry

On Wed, 2024-04-03 at 10:21 +1300, Barry Song wrote:
> On Wed, Apr 3, 2024 at 5:13 AM Jonathan Corbet <corbet@lwn.net> wrote:
> > 
> > So I'm not sure what your desired path for getting this upstream is.  I
> > can take it, but I'm generally quite leery of taking coding-style
> > changes without some serious acks on them - nobody elected me as the
> > arbiter of proper coding style.

I believe it is generally appropriate for macros that take
arguments to use static inlines where feasible, so:

Acked-by: Joe Perches <joe@perches.com>

And yes, mm is the usual path for upstreaming at least this
sort of checkpatch change.