[ovs-dev] [PATCH 01/10] lib/flow: Simplify miniflow accessors, add ipv6 support.

[Ethan Jackson]

In the commit message s/suppoort/support/

Out of curiosity why replace the miniflow_get inline functions with macros?

> +        /* Separate loops for better optimization. */

Why do we need separate loops?  You think it somehow improves the
branch predictor or something?  Does this actually help?  It smells
like premature optimization.

> +            hash = mhash_add(hash, saddr[0]);
> +            hash = mhash_add(hash, saddr[1]);
> +            hash = mhash_add(hash, saddr[2]);
> +            hash = mhash_add(hash, saddr[3]);
> +            hash = mhash_add(hash, daddr[0]);
> +            hash = mhash_add(hash, daddr[1]);
> +            hash = mhash_add(hash, daddr[2]);
> +            hash = mhash_add(hash, daddr[3]);

Maybe use a loop?

Acked-by: Ethan Jackson <ethan at nicira.com>