Is the co-pilot itself performance-sensitive?

I don't use this program. Because of this, I can't test any changes to it and because of that, I won't make any as code which hasn't been tested likely doesn't work, anyway. :-)

In theory, ie, it will likely be impossible to hit this in practice, the

&pdubuf[vindex]

could also trigger a pointer wraparound at the end of the address space and the >= test could thus fail because the code is really equivalent to

(char *)pdubuf + vindex * sizeof(*pdubuf)

The multiplication which opens up the path to a possible overflow isn't visible when using index-notation but the actual address calculation will have to employ it nevertheless.