For speed it is just one single loop.

My point was that by duplicating that loop you can have the single byte case in one, and the multibyte case in the other and decide which loop to enter, thus neither case carries the burden of the repeated, single/multi bytes tests within the loop, and both cases benefit.

The inline functions (my preference) or multiline macros (yours?) discussion was simply a way to mitigate some or all of the copy&paste code duplication.