Re: compiling lshw with uClibc

On Fri Mar 21, 2003 at 04:17:18PM +0100, Roland Nagtegaal wrote:
> Hello,
> 
> as I'm new to uClibc, let me introduce myself:
> I am Roland Nagtegaal, and I work as a system admin at the university of
> Leiden, Netherlands (physics department).
> 
> I was working on a mini-linux distro, to fit on a floppy.
> 
> It must contain the kernel, busybox, scp, and lshw, and possibly enough
> of the uClibc libraries to make them run.
> 
> My problem is that I can't compile lshw under uClibc. For that, I need
> to compile libstd++ first. 
> 
> 1. So, my first question is: is there a simple way to compile libstd++ ?
>    I searched the mailinglist, but that did not help me much...


If you _really_ want to compile your own C++ toolchain, go here
    http://uclibc.org/cgi-bin/cvsweb/toolchain/
click on "Download tarball" and then you can compile your own.
Unless you really know what you are doing though, I do not
recommend you make your own.  I think you should grab
    http://www.uclibc.org/downloads/root_fs-i386.bz2
and compile the things you need using that, which will let you
easily see how big things are with uClibc.

> 2. How do I compile the linux kernel with uClibc ? I ran make bzImage,
>    but the executable was only 16KB. Too small, even with uClibc
>    (and it didn't run of course). make zImage also fails, and make      
>    vmlinux works, but gives a 2.5MB file.

The Linux kernel does not use any C library.  But you can use a
uClibc toolchain to compile the kernel.

> 3. uClibc compiled binaries are not always smaller, it seems. For 
>    instance: /bin/ls in the buildroot environment is 76KB, but on 
>    redhat 7.3 it is 46KB (I did check it is not actaully a link to
>    busybox, it isn't). That is not a question, but what is the
>    norm? How much smaller are uClibc binaries usually?

You probably did not strip your ls binary to remove debugging
symbols.  Generally, binaries compiled with uClibc are either
the same size or smaller than when compiled using glibc.

 -Erik

--
Erik B. Andersen             http://codepoet-consulting.com/
--This message was written using 73% post-consumer electrons--