The parameters in __setup_param() should be four rather than three when
MODULE isn't definited.

Fixes: 1da177e4c3f4 ("Linux-2.6.12-rc2")
Signed-off-by: Yajun Deng <yajun.deng@linux.dev>
---
 include/linux/init.h | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/include/linux/init.h b/include/linux/init.h
index d82b4b2e1d25..62a77850f10e 100644
--- a/include/linux/init.h
+++ b/include/linux/init.h
@@ -355,7 +355,7 @@ void __init parse_early_options(char *cmdline);
 
 #else /* MODULE */
 
-#define __setup_param(str, unique_id, fn)	/* nothing */
+#define __setup_param(str, unique_id, fn, early)/* nothing */
 #define __setup(str, func) 			/* nothing */
 #endif
 
-- 
2.32.0

On Mon, 20 Dec 2021 at 04:55, Yajun Deng <yajun.deng@linux.dev> wrote:
>
> The parameters in __setup_param() should be four rather than three when
> MODULE isn't definited.

This is actually "when MODULE is defined". __setup_param() becomes a
nop when compiling as a module.

But that begs the question: why hasn't this been caught before?
Probably because nobody should be using __setup_param() if something
can also be compiled as a module, in which case module_param() and
friends should be used. But perhaps there are valid usecases where i
t's meant to become a nop if MODULE.

I don't object this fix, since the !MODULE __setup_param() seems like
it was meant to be defined.

Just curious: did you actually encounter a problem with some new code
using __setup_param()?

> Fixes: 1da177e4c3f4 ("Linux-2.6.12-rc2")
> Signed-off-by: Yajun Deng <yajun.deng@linux.dev>
> ---
>  include/linux/init.h | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/include/linux/init.h b/include/linux/init.h
> index d82b4b2e1d25..62a77850f10e 100644
> --- a/include/linux/init.h
> +++ b/include/linux/init.h
> @@ -355,7 +355,7 @@ void __init parse_early_options(char *cmdline);
>
>  #else /* MODULE */
>
> -#define __setup_param(str, unique_id, fn)      /* nothing */
> +#define __setup_param(str, unique_id, fn, early)/* nothing */
>  #define __setup(str, func)                     /* nothing */
>  #endif
>
> --
> 2.32.0
>

December 20, 2021 7:20 PM, "Marco Elver" <elver@google.com> wrote:

> On Mon, 20 Dec 2021 at 04:55, Yajun Deng <yajun.deng@linux.dev> wrote:
> 
>> The parameters in __setup_param() should be four rather than three when
>> MODULE isn't definited.
> 
> This is actually "when MODULE is defined". __setup_param() becomes a
> nop when compiling as a module.
> 
> But that begs the question: why hasn't this been caught before?
> Probably because nobody should be using __setup_param() if something
> can also be compiled as a module, in which case module_param() and
> friends should be used. But perhaps there are valid usecases where i
> t's meant to become a nop if MODULE.
> 
> I don't object this fix, since the !MODULE __setup_param() seems like
> it was meant to be defined.
> 
> Just curious: did you actually encounter a problem with some new code
> using __setup_param()?

NO, it is just code inspection. But for the current code, it's really a bug.
vim drivers/clk/imx/clk.c  +161

#ifndef MODULE

...
__setup_param("earlycon", imx_keep_uart_earlycon,
              imx_keep_uart_clocks_param, 0); 
__setup_param("earlyprintk", imx_keep_uart_earlyprintk,
              imx_keep_uart_clocks_param, 0); 


> 
>> Fixes: 1da177e4c3f4 ("Linux-2.6.12-rc2")
>> Signed-off-by: Yajun Deng <yajun.deng@linux.dev>
>> ---
>> include/linux/init.h | 2 +-
>> 1 file changed, 1 insertion(+), 1 deletion(-)
>> 
>> diff --git a/include/linux/init.h b/include/linux/init.h
>> index d82b4b2e1d25..62a77850f10e 100644
>> --- a/include/linux/init.h
>> +++ b/include/linux/init.h
>> @@ -355,7 +355,7 @@ void __init parse_early_options(char *cmdline);
>> 
>> #else /* MODULE */
>> 
>> -#define __setup_param(str, unique_id, fn) /* nothing */
>> +#define __setup_param(str, unique_id, fn, early)/* nothing */
>> #define __setup(str, func) /* nothing */
>> #endif
>> 
>> --
>> 2.32.0