i960 multilib support problem

i960 multilib support problem

[Neal Becker]

I build binutils-2.8.1.0.18 egcs-1.0.1 newlib-1.8.0.  After fiddling
with newlib a bit I got a successful build.  But it seems the multilib 
support is not complete.  I want to build for a CA architecture.  But
the multilib in egcs only produces float and soft-float.  This is
apparantly driven by egcs/gcc/config/i960/t-960bare, which says:

MULTILIB_OPTIONS=mnumerics/msoft-float
MULTILIB_DIRNAMES=float soft-float
MULTILIB_MATCHES=mnumerics=msb mnumerics=msc mnumerics=mkb mnumerics=mkc mnumerics=mmc mnumerics=mcb mnumerics=mcc mnumerics=mjf msoft-float=msa msoft-float=mka msoft-float=mca msoft-float=mcf

This isn't sufficient:

PATH=/usr/local/i960/bin:$PATH i960-intel-coff-gcc hello.c -v -mca -nostartfiles -lc -lmon960 -lc
Reading specs from /usr/local/i960/lib/gcc-lib/i960-intel-coff/egcs-2.90.23/specs
gcc version egcs-2.90.23 980102 (egcs-1.0.1 release)
 /usr/local/i960/lib/gcc-lib/i960-intel-coff/egcs-2.90.23/cpp -lang-c -v -undef -D__GNUC__=2 -D__GNUC_MINOR__=90 -Di960 -Di80960 -DI960 -DI80960 -D__i960__ -D__i80960__ -D__I960__ -D__I80960__ -D__i960 -D__i80960 -D__I960 -D__I80960 -Acpu(i960) -Amachine(i960) -D__CHAR_UNSIGNED__ -D__i960CA__ -D__i960_CA__ hello.c /tmp/cca26978.i
GNU CPP version egcs-2.90.23 980102 (egcs-1.0.1 release) (intel 80960)
#include "..." search starts here:
#include <...> search starts here:
 /usr/local/i960/lib/gcc-lib/i960-intel-coff/egcs-2.90.23/include
 /usr/local/i960/i960-intel-coff/include
End of search list.
 /usr/local/i960/lib/gcc-lib/i960-intel-coff/egcs-2.90.23/cc1 /tmp/cca26978.i -quiet -dumpbase hello.c -mca -version -o /tmp/cca26978.s
GNU C version egcs-2.90.23 980102 (egcs-1.0.1 release) (i960-intel-coff) compiled by GNU C version egcs-2.90.23 980102 (egcs-1.0.1 release).
 /usr/local/i960/i960-intel-coff/bin/as -ACA -o /tmp/cca269781.o /tmp/cca26978.s
 /usr/local/i960/lib/gcc-lib/i960-intel-coff/egcs-2.90.23/ld -ACA -L/usr/local/i960/lib/gcc-lib/i960-intel-coff/egcs-2.90.23/soft-float -L/usr/local/i960/lib/gcc-lib/i960-intel-coff/egcs-2.90.23 -L/usr/local/i960/i960-intel-coff/lib/soft-float -L/usr/local/i960/i960-intel-coff/lib /tmp/cca269781.o -lc -lmon960 -lc -lgcc -lgcc
/usr/local/i960/i960-intel-coff/bin/ld: warning: i960:kb_sb architecture of input file `/usr/local/i960/i960-intel-coff/lib/soft-float/libc.a(printf.o)' is incompatible with i960:ca output
/usr/local/i960/i960-intel-coff/bin/ld: warning: i960:kb_sb architecture of input file `/usr/local/i960/i960-intel-coff/lib/soft-float/libc.a(vfprintf.o)' is incompatible with i960:ca output
/usr/local/i960/i960-intel-coff/bin/ld: warning: i960:kb_sb architecture of input file `/usr/local/i960/i960-intel-coff/lib/soft-float/libc.a(impure.o)' is incompatible with i960:ca output
/usr/local/i960/i960-intel-coff/bin/ld: warning: i960:kb_sb architecture of input file `/usr/local/i960/i960-intel-coff/lib/soft-float/libc.a(fvwrite.o)' is incompatible with i960:ca output
/usr/local/i960/i960-intel-coff/bin/ld: warning: i960:kb_sb architecture of input file `/usr/local/i960/i960-intel-coff/lib/soft-float/libc.a(fflush.o)' is incompatible with i960:ca output
/usr/local/i960/i960-intel-coff/bin/ld: warning: i960:kb_sb architecture of input file `/usr/local/i960/i960-intel-coff/lib/soft-float/libc.a(locale.o)' is incompatible with i960:ca output
[...]
/usr/local/i960/i960-intel-coff/bin/ld: warning: i960:kb_sb architecture of input file `/usr/local/i960/lib/gcc-lib/i960-intel-coff/egcs-2.90.23/soft-float/libgcc.a(fp-bit.o)' is incompatible with i960:ca output
[...]

So both libc and libgcc were for kb_sb architecture, but I need ca.
Any ideas how to fix this?