[uClibc]Busybox and Tinylogin vs uClibc: "segmentation fault"

[Stefano Costa]

On 12 Nov 2002 at 14:37, Erik Andersen wrote:

> On Tue Nov 12, 2002 at 09:54:34PM +0100, Stefano Costa wrote:
> > I recently built the complete ARM toolchain and Linux kernel for a
> > CDB89712 following the Roland Seuhs directions (with a few hints that
> > I'm planning to describe in our website soon). It seems to be
> > working, except for Busybox and Tinylogin: when called, both give the
> > message "segmentation fault".
> > 
> > To test this we first loaded the kernel (the one built here) and a
> > working root filesystem found at:
> > 
> > http://ttestasecca.free.fr/cdb89712/
> > 
> > and then added the "busybox" executable built here in a target
> > directory. Busybox was built starting both from the latest cvs
> > sources, the latest "stable" sources (0.60.5) and also from 0.60.2 to
> > be sure, using uClibc and "DOSTATIC". I also tried to build some
> > simple "hello.c" using the same uClibc and *it works* fine on
> > CDB89712.
> 
> Since you have checked several versions of busybox, it's a
> reasonable guess that busybox is working fine.  So lets operate
> on the assumption that it is either your kernel, your version of
> uClibc, or your toolchain that is broken.
Yes, this is probably the case, but the simple "hello.c" works, and 
the kernel as well.

> 
> What version of the arm-linux kernel are you using?
VERSION = 2
PATCHLEVEL = 4
SUBLEVEL = 6
EXTRAVERSION =-rmk1-rayl1
This is version 2.4.6 with patch "rmk1" (patches 618-621 I guess) as 
suggested by Roland Seuhs and others.

> 
> What version of gcc and binutils are you using to compile
> everything?  gcc-3.2?  binutils-2.12.1?
binutils: 2.11
gcc: 2.95.3
(I didn't try anything more recent, as all examples and suggestions 
are based on these versions). Do you recommend trying with gcc-3.2 
and binutils-2.12.1?

> 
> What version of uClibc?  0.9.16?  If so, do you see the same
> segfault when building vs uClibc 0.9.15?
uClibc is version 0.9.15, the latest from cvs a few days ago.

> 
> The first step is to narrow down the problem so you can try
> and find which compenent is at fault,
> 
>  -Erik

Thanks for your help!

bye- Stefano