Bug 121670

Description of problem:

I'm building i386 packages on x86_64 by creating a chroot environment
with a full i386 packageset.  All but two of the srpms from RHEL3 will
successfully rebuild in this way, gcc being one of the failures.

Version-Release number of selected component (if applicable):

gcc-3.2.3-20.src.rpm


How reproducible:

Always

Steps to Reproduce:
1.  Build up a chroot environment of RHEL3 while running the x86_64
kernel.
2.  Attempt rebuilding gcc-3.2.3-20.src.rpm.
3.  Watch it fail.
4.  Reboot into an i386 environment, plug back into the same chrooted
environment and repeat.  It will succeed.

I am doing all of the proper steps to create a full chrooted build
environment.  /proc, /dev/pts and /dev/shm are being mounted and a
setarch i386 is being done in front of the chroot command.  The script
used to create the chroot is available if needed.
  
Actual results:

Comparing captures of the build, things go wrong here:

checking whether we are using GNU C... (cached) yes
checking whether gcc accepts -g... (cached) yes
checking whether gcc and cc understand -c and -o together... yes
checking for gnatbind... gnatbind
checking for compiler driver that understands Ada... no

Expected results:

When building with an athlon kernel I get this correct output:

checking whether we are using GNU C... (cached) yes
checking whether gcc accepts -g... (cached) yes
checking whether gcc and cc understand -c and -o together... yes
checking for gnatbind... gnatbind
checking for compiler driver that understands Ada... gcc

I really tried to track down what is going wrong in more detail, but I
just can't make heads or tails out of the GCC spec file, to say
nothing of the actual configure script for that puppy.