[uClibc] build gcc with uClibc

Manuel Novoa III mjn3 at codepoet.org
Tue Aug 23 00:52:00 UTC 2005


On Mon, Aug 22, 2005 at 04:05:06PM -0700, Bahman Rafatjoo wrote:
> 
> I've been doing a fair bit of searching and reading on
> the list
> but there's still something I don't get about the gcc
> wrapper. I
> know this is a very old topic and I'm not looking to
> use it but
> I just want to understand so please don't flame me!
> 
> As I noted earlier my manual installation of binutils,
> gcc and uClibc seems to work OK. Note that I'm
> building statically linked non-linux
> code which I burn onto flash for a home made SBC. I
> have no
> filesystem, no loaders and no configure script. Just
> stand-alone.
> 
> I that context I don't get the comments about 'uClibc
  ^^^^^^^^^^^^^^   That's a pretty big qualifier.

> aware'
> toolchain? In my case I don't even see a need for a
> wrapper and I
> didn't do anything like patching gcc or ld.

The wrapper simply made things easier, so you didn't have to specify
dynamic linker to use, header paths, etc.  But it could still break,
depending on what you were doing.

> In the post
> (http://www.uclibc.org/lists/uclibc/2003-December/007853.html)
> Erik Andersen says that "configure scripts go poking
> about your build
> system and find the build system's headers, libraries,
> etc". I can see
> that for such packages how the buildroot environment
> can greatly simplify
> things. Is that the answer? Do the
> toolchain/glibc/wrapper issues only
> exist when building code using confgure or code that
> interacts with the
> kernel and not simple stand-alone monitor code?
> 
> I'd be grateful for some clarification.

As I've replied before, the wrapper works for some things.  But it
isn't worth the time to worry about when it does or doesn't when one
can simply build a uclibc-targeted toolchain.

Manuel



More information about the uClibc mailing list