Re: cpu-2.5.64-1

[Keith Owens]

On Sun, 16 Mar 2003 02:10:55 -0800, 
William Lee Irwin III <[e-mail]> wrote:
>On Sun, Mar 16, 2003 at 08:19:31PM +1100, Keith Owens wrote:
>> Good, it lets us optimize for 1/32/64/lots of cpus.  NR_CPUS > 8 *
>> sizeof(unsigned long) is the interesting case, it needs arrays.
>
>Hmm. It shouldn't make a difference with respect to optimizing them.
>My API passes transparently by reference:
>#define cpu_isset(cpu, map)		test_bit(cpu, (map).mask)

Some of the 64 bit archs implement test_bit() as taking int * instead
of long *.  That generates unoptimized code for the case of NR_CPUS <
64.

#if NR_CPUS < 64
#define cpu_isset(cpu, map)		(map.mask & (1UL << (cpu)))
...

generates better code on those architectures.  Yet another special case :(