Optimisation for smp_num_cpus loop in hotplug

[James Bottomley]

The hotplug CPU patch introduces this to replace the loop over all active CPUs 
abstraction:

	for (i = 0; i < NR_CPUS; i++) {
		if (!cpu_online(i))
			continue;

Since the cpu online map is probably going to be quite sparse, could I suggest 
this alternative, which doesn't have to loop 32 times:

===== smp.h 1.11 vs edited =====
--- 1.11/include/asm-i386/smp.h Thu Jun 20 14:04:21 2002
+++ edited/smp.h        Thu Jun 20 12:39:33 2002
@@ -95,6 +95,11 @@
 
 #define cpu_online(cpu) (cpu_online_map & (1<<(cpu)))
 
+#define for_each_cpu(cpu, mask) \
+       for(mask = cpu_online_map; \
+           cpu = __ffs(mask), mask != 0; \
+           mask &= ~(1<<cpu))
+
 extern inline unsigned int num_online_cpus(void)
 {
        return hweight32(cpu_online_map);

I've implemented this for my voyager system (8 cpus, but still a sparse online 
bitmap), mainly because (for historical reasons), I have to do this loop in 
time critical IPI code.

James