dual entry asm routine

I have this case of two virtually identical routines that must run at the ultimate speed.

One approach would be:

           RSEG ?DT?FormatBitSegment
FBSfcol:   DS    1

           RSEG  ?PR?FormatBitSegment?AVFMTBS
           USING  0

FormatScrollSegment:
           ......
           mov   a,FBSfcol
           ......
           sjmp  common

FormatBitSegment:
           ......
           mov   a,FBSfcol
           .......
common:    .......
           ......
           mov   a,FBSfcol
           ......
           ret