Cross compiling glibc-2.2.3

Cross compiling glibc-2.2.3

[IGOR LURI]

Hi,
I am trying to crosscopile glibc-2.2.3 for the mpc8xx.

I use the folowing configure script, but when I do make it fails after a
while.

export AR=ppc_8xx-ar
export CC=ppc_8xx-gcc
export CFLAGS="-msoft-float -D_SOFT_FLOAT -mcpu=860"
export RANLIB=ppc_8xx-ranlib

./configure	--enable-kernel=2.4.4 --host=powerpc-linux
--build=i686-pc-linux-gnu \
		--with-headers=/usr/src/linux-2.4.4/include \
		--enable-add-ons=linuxthreads \
		--without-fp


Anybody has already crosscompiled glibc 2.2.3 for mpc8xx?

I know that for glibc-2.1.2 there is a patch in denx page, but for glibc
2.2.3 is there any patch?

** Sent via the linuxppc-embedded mail list. See http://lists.linuxppc.org/

Cross compiling glibc-2.2.3

[Sangmoon Kim]

Hi, IGOR LURI

> Anybody has already crosscompiled glibc 2.2.3 for mpc8xx?
I crosscompiled glibc-2.2.5.

> I know that for glibc-2.1.2 there is a patch in denx page, but for glibc
> 2.2.3 is there any patch?

I referenced the patch in denx page, and fixed the crosscompile problem by following procedure.

1. Move the sysdeps/powerpc/fclrexcpt.c to sysdeps/powerpc/fpu/
mv sysdeps/powerpc/fclrexcpt.c sysdeps/powerpc/fpu/

2. Delete the sysdeps/powerpc/memset.S
rm sysdeps/powerpc/memset.S

3. Edit the sysdeps/powerpc/Makefile
#ifeq ($(subdir),gmon)
#libm-support += fenv_const fe_nomask
#endif

4. Compile
CC=${TARGET}-gcc AR=${TARGET}-ar RANLIB=${TARGET}-ranlib \
        CFLAGS="-msoft-float -D_SOFT_FLOAT -mcpu=860 -O2" \
        ../${GLIBC}/configure --host=$TARGET --prefix=${PREFIX} \
        --enable-add-ons=linuxthreads --with-headers=${PREFIX}/include \
        --without-fp

Good luck.
- Sangmoon Kim -


** Sent via the linuxppc-embedded mail list. See http://lists.linuxppc.org/

Re: Cross compiling glibc-2.2.3

[Sangmoon Kim]

Hi, Dr. Craig Hollabaugh,

> >3. Edit the sysdeps/powerpc/Makefile
> >#ifeq ($(subdir),gmon)
> >#libm-support += fenv_const fe_nomask
> >#endif
>
> Kim,
>
> What does this step actually do?
>

I just referenced the denx patch. So I don't have the exact knowledge obout it. I guess the step is for omitting floating point because MPC860 doesn;t support it.I think supporting floating in kernel by floating-point interrupt,
and compile glibc without -msoft-float or -D_SOFT_FLOAT is better.But the guys in this list doesn't seam to recommand it.

> Also, should that be math above instead gmon?
>
> #ifeq ($(subdir),math)
> #libm-support += fenv_const fe_nomask
> #endif
>
That was my mistake. I meant...
#ifeq ($(subdir),math)
#libm-support += fenv_const fe_nomask
#endif


** Sent via the linuxppc-embedded mail list. See http://lists.linuxppc.org/

Re: Cross compiling glibc-2.2.3

[Dr. Craig Hollabaugh]

> I just referenced the denx patch. So I don't have the exact knowledge obout it. I guess the step is for omitting floating point because MPC860 doesn;t support it.I think supporting floating in kernel by floating-point interrupt,
> and compile glibc without -msoft-float or -D_SOFT_FLOAT is better.But the guys in this list doesn't seam to recommand it.

No, I asked about this last week and Wolfgang said turn off FP emulation in the kernel and compile
libs and apps with soft-float.

see thread, http://lists.linuxppc.org/linuxppc-embedded/200204/msg00174.html


>
> > Also, should that be math above instead gmon?
> >
> > #ifeq ($(subdir),math)
> > #libm-support += fenv_const fe_nomask
> > #endif
> >
> That was my mistake. I meant...
> #ifeq ($(subdir),math)
> #libm-support += fenv_const fe_nomask
> #endif


OK

** Sent via the linuxppc-embedded mail list. See http://lists.linuxppc.org/