matthew green wrote:
>    
>    The only one problem is the partition type, I guess I can rely on the
>    value of sep->se_dev (ie. if the swap is a valid block dev or not) to
>    determine if it's a partition or a file.
> 
> 
> se_dev will be NODEV for a file or a device for a partition.
> 
> 
> sizes are stored in DEV_BSIZE units, so you should adjust the value of
> se_nblks and se_inuse with "/ 1024 * DEV_BSIZE" or something.  i'm not
> really sure which order would be best for those two operation, in order
> to avoid integer underflow or overflow issues.
is it OK to do something like :

#define dbtokb(size)    ((size)>>(10-DEV_BSHIFT))

There can be issue with negative shift, is which case we can test if 
(10-DEV_BSHIFT) is <0, and make the opposite shift.

  - Arnaud