Re: deadcode optimization

Norman Black wrote:
> "Smart linking" is trivial for a compiler do handle and have ALL
> linkers in existence support the unused code "removal".
>
> All that need be done is have the compiler output library/archive
> files directly with one procedure per object in the library. Our
> compilers have done this since 1987 and any linker appropriately does
> not link in unused code.


Using a smarter linker can do more than just eliminate dead code.


For instance, the Plan 9/Inferno C compilers emit "high level" machine
code which is translated as appropriate by the linker (e.g. the
emitted code can reference the "frame pointer" when none actually
exists - the linker statically keeps track of the stack pointer and
inserts the appropriate offsets)


The linker can then do all sorts of interesting things: automatically
inline small procedures, remove unreferenced global data and functions
(the former can be a big win on space), peephole and interprocedural
optimisation, etc.


Another example of the flexibility of this approach: we've got a
linker for the ARM that can generate ARM or Thumb code - the compiler
produces the same intermediate files for both.


    cheers,
        rog.
[I've always been surprised we don't see more link-time optimization,
since the linker is usually the only phase of the compilation process
that sees the entire program. -John]

Return to the comp.compilers page.
Search the comp.compilers archives again.