[oe-users] State of PowerPC

[Darcy Watkins]

Hi Philip,

Sorry about the long delay.

> WHat errors? Can you send in the end of the log?

During compile for glibc I saw 

   "warning: -msoft-float and -mlong-double-128 not supported"

The compile broke in fpu/s_isnan.c 45: error: impossible constraint is
'asm'

I think it is getting mixed up regarding presence / absence of HW FP
(the AMCC PPC405EP doesn't have a FPU).

---

During compile for uclibc I saw

CROSS COMPILE Badness: /usr/include in INCLUDEPATH
cc1: internal compiler error: in add_path, at c-incpath.c:362


----

I have also been checking out buildroot in which I am currently dealing
with the pciutils package not building.  (For buildroot, I am able to
build a toolchain and a basic root file system, but not a linux kernel
for the AMCC Taihu board - for that, I still have to use one I built
previously using the DENX ELDK toolchain).

For buildroot, the libs/configure file of pciutils package needs to be
patched to not depend on the host system's "uname" to determine
architecture related information.

The pciutil patch files for OE do not make any such changes, but instead
I found a "uname" executable:

build/tmp/work/i686-linux/gnu-config-native-0.1+cvs20050701-r5/config/un
ame

Is this hooked to be run instead of the build system's "uname" during
cross compile?

I tried running it, but saw no information regarding powerpc.  Could
this be contributing to the cross compile badness?

... /uname -s
iainfo2

... /uname -m
4400

... /uname -r
4.0

... or is this something totally unrelated?


Regards,

Darcy