[PATCH] Remove hardcoded number of CFI chips

[andrzej.mialkowski at inetia.pl]

Cytowanie Deepak Saxena <dsaxena at plexity.net>:

> 
> All, 
> 
> The attached patch is an update of the patch I posted about a month
> ago [1] that removes the hardcoded CFI_MAX_CHIPS, which was originally
> written by Andrzej Mialkowski @ Intel.  This patch fixes two issues:
> 
> - Get rid of extra structure probe_info that was just carrying around 
>   a long* variable.
This 'stupid' single field structure was intentional to allow easier expansion 
if somebody fill need for additional fields.
> 
> - Fix bug in bitmap size on systems with 2-8 chips. The old patch had:
> 
>   +	max_chips = map->size >> cfi.chipshift;
>   +	probe.chip_map = kmalloc((max_chips + 7) / 8, GFP_KERNEL);
> 
>   This would cause systems with 2-8 chips to be allocated only one
>   byte for the bitmap instead of the 2 needed. Fixed code does:
> 
>   +  max_chips = map->size >> cfi.chipshift;
>   +  chip_map = kmalloc((max_chips / 8) + ((max_chip % 8) ? 1 : 0),
> GFP_KERNEL);
> 
Why I need two bytes to store up to 8 bits? By the way 2/8 + 2?1:0 = 1.
Initially I thought to put higher alignment here for instance ((max_chips + 
31) / 32)*4) but found no reason to do this. Maybe I was wrong...

Andrzej