Re: mklibgcc fallout

[David Edelsohn <dje at watson dot ibm dot com>]

>>>>> Zack Weinberg writes:

Zack> It's not that simple.

Zack> A powerpc64-linux build tree from October contains the following files
Zack> named libgcc_s*so* which are of principal relevance

Zack> libgcc_s.so.1
Zack> 32/libgcc_s.so.1
Zack> 32/nof/libgcc_s_nof.so.1
Zack> libgcc_s.so -> libgcc_s.so.1
Zack> libgcc_s_32.so -> 32/libgcc_s.so.1
Zack> libgcc_s_32_nof.so -> 32/nof/libgcc_s_nof.so.1

	A build of powerpc-ibm-aix5.1.0.0 from 20040819 creates (ignoring
that AIX shared libraries have a ".a" suffix):

gcc/libgcc_s.a
gcc/libgcc_s_power.a
gcc/libgcc_s_powerpc.a
gcc/libgcc_s_ppc64.a
gcc/libgcc_s_pthread.a
gcc/libgcc_s_pthread_power.a
gcc/libgcc_s_pthread_powerpc.a
gcc/libgcc_s_pthread_ppc64.a

gcc/stage2/libgcc_s.a
gcc/stage2/libgcc_s_power.a
gcc/stage2/libgcc_s_powerpc.a
gcc/stage2/libgcc_s_ppc64.a
gcc/stage2/libgcc_s_pthread.a
gcc/stage2/libgcc_s_pthread_power.a
gcc/stage2/libgcc_s_pthread_powerpc.a
gcc/stage2/libgcc_s_pthread_ppc64.a

	It appears that powerpc64-linux has been buggy all along.

David