I use GOLEM, a disassembler that allowed annotating the resulting source (adding lables and comments) and re-running the disassembler for ever-more-informative disassembly. It was also script-based so it could disassemble any machine code (if you were patient enough to write up all the patterns). It could also recognize complex sequences and produce pseudo-instructions e.g. Loop, TestAndBranch etc.

Anyway, not really related, this article is about detecting linux's signature and identifying the kernel, but it brings back old times.