Ok.. after a few unrelated issues with the sparc build, this tests out correctly (using data generated from s10 sparc) and also includes a few minor cstyle cleanups in the existing code: http://cr.illumos.org/view/x2ewa65v/