Re: More ARM binutils fuckage

["Michael K. Edwards" <medwards dot linux at gmail dot com>]

On 12/5/06, Khem Raj <kraj@mvista.com> wrote:
> hmmm libgcc should be providing __div0 and raise should come from libc.so.6
> I dont know how you configured the toolchain.

Right, libgcc now provides __div0() and other routines in libgcc use
it.  Apparently __div0() used to be in glibc instead, so U-Boot (at
least as of 1.1.4) also provides __div0().  The fix seems to be to
remove U-Boot's __div0() and instead provide a stub for raise(), which
libgcc's __div0() wants to call.

> its in the contribs
> http://kegel.com/crosstool/crosstool-0.42/contrib/2005b/
> not all arches supported nptl so well back then. It will probably make sense
> now to have a separate script for nptl toolchains. Havent had enough time
> lately to work on crosstool.

Martin reworked it not to break non-NPTL builds, and I tweaked it a
bit to fix up EABI and TLS handling when the host compiler, glibc,
kernel are older.  I'll be posting a new patch shortly that reconciles
some of the discrepancies associated with OABI softfloat.

Cheers,
- Michael

--
For unsubscribe information see http://sourceware.org/lists.html#faq