https://gcc.gnu.org/bugzilla/show_bug.cgi?id=85347
--- Comment #1 from kelvin at gcc dot gnu.org --- FWIW, the test runs ok on bns -m32 as written. Executing on host: /home/kelvin/gcc/build/gcc-trunk4fixvecldl.b4patch/gcc/xgcc \ -B/home/kelvin/gcc/build/gcc-trunk4fixvecldl.b4patch/gcc/ /home/kelvin/gcc/gcc-\ trunk4fixvecldl.b4patch/gcc/testsuite/gcc.target/powerpc/vec-ldl-1.c -fno-dia\ gnostics-show-caret -fdiagnostics-color=never -maltivec -O0 -Wall -lm -m32\ -o ./vec-ldl-1.exe (timeout = 300) spawn /home/kelvin/gcc/build/gcc-trunk4fixvecldl.b4patch/gcc/xgcc -B/home/kelvi\ n/gcc/build/gcc-trunk4fixvecldl.b4patch/gcc/ /home/kelvin/gcc/gcc-trunk4fixvecl\ dl.b4patch/gcc/testsuite/gcc.target/powerpc/vec-ldl-1.c -fno-diagnostics-show-c\ aret -fdiagnostics-color=never -maltivec -O0 -Wall -lm -m32 -o ./vec-ldl-1.exe^\ M PASS: gcc.target/powerpc/vec-ldl-1.c (test for excess errors) Would be nice to know where this fails so I can confirm that my "fix" works.