On Mon, Jul 9, 2012 at 7:43 AM, Markus Armbruster <arm...@redhat.com> wrote:
> blauwir...@gmail.com writes:
>
>> From: Blue Swirl <blauwir...@gmail.com>
>>
>> Use 'bool' type for return value of bit test functions.
>
> Matter of taste. 'bool' makes sense if you think of these functions as
> predicates (ugly ones, with side effects). 'int' makes sense if you
> think of them as returning the bit value (a bit is not a truth value in
> my book).
The documentation actually talks about returning the old value.
But I think the kernel version (with the same documentation) however
returns a truth value:
static inline int test_and_clear_bit(int nr, volatile unsigned long *addr)
{
int oldbit;
asm volatile(LOCK_PREFIX "btr %2,%1\n\t"
"sbb %0,%0"
: "=r" (oldbit), ADDR : "Ir" (nr) : "memory");
return oldbit;
}
SBB oldbit, oldbit with oldbit = 0 initially and carry set will produce -1, no?
>
> In my opinion, this is just unnecessary churn, and unnecessary deviation
> from established kernel practice.
We use for example NetBSD queue macros with heavy modification. If we
have a need to modify code derived from Linux kernel, we should do it
as well. In this case the kernel practice seems to be inconsistent and
worth fixing.
