Re: [PATCH][2.6] first/next_cpu returns values > NR_CPUS

[William Lee Irwin III <wli@holomorphy.com>]

On Sun, Aug 01, 2004 at 03:22:56AM -0400, Zwane Mwaikambo wrote:
> NR_CPUS was 3, the test case may as well be passing first_cpu or next_cpu
> a value of 0 for the map. The "bug" in the i386 find_next_bit really
> looks like a feature if you look at the code.

Hmm. I'm actually somewhat puzzled by this also. Shouldn't things only
check for inequalities between the results of these and NR_CPUS? i.e.
things like:
	if (any_online_cpu(cpus) >= NR_CPUS)
and
	for (cpu = first_cpu(cpus); cpu < NR_CPUS; cpu = next_cpu(cpus))
etc.?

Maybe the few callers that are sensitive to the precise return value
should use min_t(int, NR_CPUS, ...) instead of all callers taking the
branch on behalf of those few.


-- wli