Re: [PATCH] nodemask_t x86_64 changes [5/7]

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

On Mon, Mar 22, 2004 at 11:08:50PM -0800, Paul Jackson wrote:
> I'll be surprised if the following line works:
> 	nodemask_t node_offline_map = nodes_complement(node_online_map);
> 1) Doesn't nodes_complement return void, and work in place?
> 2) It might set bits above MAX_NUMNODES, if MAX_NUMNODES isn't a word size multiple.
> I am less sure of (2) - the exact details of handling the unused bits of
> a bitmask are still confusing me.  But this would be one of the very
> rare situations that I can find that would actually be sensitive to
> possible confusions here - most places don't set bits that aren't
> already set in some mask, or are careful to initialize a mask with just
> set bits in select positions from 0 to MAX_NUMNODES-1.

In general I attempted to model things after 3-address code.
bitmap_complement() is a glaring inconsistency I wouldn't mind seeing
shored up with the rest (though I guess it's only got 2 operands).


-- wli