Re: [PATCH] kernel/range: Remove unused definition of ARRAY_SIZE()

[Yinghai Lu <yinghai.lu@oracle.com>]

On 06/07/2010 03:49 PM, Andrew Morton wrote:
> On Sat, 5 Jun 2010 21:32:15 +0200 (CEST)
> Geert Uytterhoeven <geert@linux-m68k.org> wrote:
> 
>> Remove duplicate definition of ARRAY_SIZE(), which was never used anyway.
>>
>> Signed-off-by: Geert Uytterhoeven <geert@linux-m68k.org>
>> ---
>>  kernel/range.c |    4 ----
>>  1 files changed, 0 insertions(+), 4 deletions(-)
>>
>> diff --git a/kernel/range.c b/kernel/range.c
>> index 74e2e61..471b66a 100644
>> --- a/kernel/range.c
>> +++ b/kernel/range.c
>> @@ -7,10 +7,6 @@
>>  
>>  #include <linux/range.h>
>>  
>> -#ifndef ARRAY_SIZE
>> -#define ARRAY_SIZE(x) (sizeof(x) / sizeof((x)[0]))
>> -#endif
>> -
>>  int add_range(struct range *range, int az, int nr_range, u64 start, u64 end)
>>  {
>>  	if (start >= end)
> 
> <discovers range.c>
> 
> That's not terribly great code, sorry.
> 
> - The names are all wrong.  Should be range_add(),
>   range_add_with_merge(), range_subtract(), etc.
> 
> - It's completely undocumented!
> 
> - It's linked into every vmlinux in the world, many of which won't use it
>   afacit.
> 
> - The return value from add_range() is a bit odd.  I guess callers must do
> 
> 	if (add_range(..., ..., nr_range, ..., ...) == nr_range)
> 		error()
> 
> - What does the identifier "az" mean?
> 
> - `az' and `nr_range' should be unsigned types.  That would make the
>   "Out of slots:" check non-buggy.
> 
> - The return value from add_range_with_merge() is unusable!  If it
>   did a merge into the final range it will return the caller's
>   nr_range.  If it failed to merge it will call add_range() and then
>   will return the caller's nr_range if it ran out of space.
> 
>   So the caller cannot determine from the return value whether or not
>   the range was added.
> 
>   Or something.  This is an advantage of actually documenting code -
>   it makes people think about such things.
> 
> - The main structure seems just wrong, or at least inappropriate.  Should be
> 
> 	struct range {
> 		/* Number of ranges presently at *ranges */
> 		unsigned nr_ranges;
> 		/* Maximum number of ranges storable at *ranges */
> 		unsigned max_ranges;
> 		struct {
> 			u64 start;
> 			u64 end;
> 		} *ranges;
> 	};
> 
>   Or similar.
> 
> - I can't be bothered working out what subtract_range() and
>   clean_sort_range() are supposed to be doing, so I didn't look at
>   them.
> 
> c'mon guys, we can do better than this.

will work on using lmb to replace range.c later after lmb for x86 is in tip and linux-next.

Thanks

Yinghai