My vague recollection is that the kernel in Unix v2 (still called UNIX in those days) was written in assembly language, but that many of the commands were written in B. Because B was interpretive on the PDP-11 (it suffered from the same problem as BCPL, in that it expected a word-addressed machine), it would have been impractical to use it in the kernel. If I recall correctly, the first Unix to reach Berkeley was some kind of v2.
The C compiler appeared in v3, and the kernel rewrite happened for Version 4, which is the version described in the paper.
Using .S for assembly language source is a historical accident, by the time the majority of code was written in C, .a was defined as the conventional filetype for archive (ar) files.
The C compiler appeared in v3, and the kernel rewrite happened for Version 4, which is the version described in the paper.
Using .S for assembly language source is a historical accident, by the time the majority of code was written in C, .a was defined as the conventional filetype for archive (ar) files.