Hmm.  I didn't read your comment carefully enough before I replied.  I think you are saying that the bug fix is confined to the routine in question and doesn't change even its API, in which case the nature of the function doesn't come in to it at all.  The fix still needs to be evaluated for backport, though, since it is a behavior change.  (For example, in another issue Senthil fixed parsing of unknown schemes to be RFC compliant, but we may need to back that fix out of 2.6 since it changes behavior and will most likely break currently working 2.6 code).