Re: Makefile best practices for embedded development

[Ken Rose <kenrose at tfb dot com>]

Bryce Schober wrote:
> We're working on an arm7 project with the gcc toolchain here, and 
> currently our Makefiles are a mess.  I've yet to get really going 
> full-bore on this project, and I'd like to do some cleanup before I dive 
> in.  I've done a bit of research around the web including 
> http://make.paulandlesley.org/autodep.html, and it looks promising, but 
> there's a catch.  What do you do about dependencies for assembly files? 
> Doing gcc -M on my toolchain doesn't catch any includes.  Ideally, we 
> won't be using them much, but I'm planning on needing to do some 
> optimized (assembly) fixed-point routines in assembly for our 
> application.  What is a good clean makefile solution that minimizes 
> makefile maintenance and is embedded-friendly?

I don't have any specific advice there.  When the project I was working 
on encountered a similar problem, we modified our assembler to generate 
dependencies.  We weren't using gas, so that doesn't help you.  This 
does seem like a good thing for gas (or gasp?) to do, though.

More generally, be sure to read "Recursive Make Considered Harmful", 
which you can get from http://aegis.sourceforge.net/auug97.pdf

Altenatively, for a much better tool that's harder to learn, and is 
unfortunately limited in host platforms, take a look at Vesta 
(http://www.vestasys.org).  It captures dependencies on environment 
variables, library versions, and just about everything else that can 
affect the output bits, while also giving you a distributed build and 
detecting and reusing identical pieces.

 - ken



------
Want more information?  See the CrossGCC FAQ, http://www.objsw.com/CrossGCC/
Want to unsubscribe? Send a note to crossgcc-unsubscribe@sources.redhat.com