Ian Lance Taylor ian@airs.com
Tue Nov 8 14:52:00 GMT 2005

• Previous message (by thread): Link-Time Optimizations
• Next message (by thread): Link-Time Optimizations
• Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]

---

Javier Cabezas Rodríguez <javier.cabezas@gmail.com> writes:

> I haven't found a documented example of "Linker Relaxation", but I
> think that we aren't speaking about the same kind of optimizations.
> Diablo can do:
> 
> -  Dead Code and Data elimination, achieving up to 25% code size reduction.
> 
> - Whole program constant propagation.
> 
> - Branch elimination.
> 
> - Liveness analysis.
> 
> - Code factoring.
> 
> - Code obfuscation
> 
> - Code compression
> 
> - ... and much more.
> 
> At the moment it only allows this for statically linked programs, but
> support for dynamically linked programs is planned.

In the GNU toolchain the current general idea is to do these things in
the compiler.  See gcc's -combine option.  This does have some
drawbacks; in particular, the compiler in general won't see the whole
program at once.  And the compiler is never going to do code
obfuscation or code compression.

There are no plans that I know of to add any of this functionality to
the GNU linker.

Ian

---

• Previous message (by thread): Link-Time Optimizations
• Next message (by thread): Link-Time Optimizations
• Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]

---

More information about the Binutils mailing list