; RUN: llc < %s -march=mipsel -mcpu=mips32 -O0 \ ; RUN: -fast-isel -mips-fast-isel -relocation-model=pic ; RUN: llc < %s -march=mipsel -mcpu=mips32r2 -O0 \ ; RUN: -fast-isel -mips-fast-isel -relocation-model=pic ; The test is just to make sure it is able to allocate ; registers for this example. There was an issue with allocating AC0 ; after a mul instruction. declare { i32, i1 } @llvm.smul.with.overflow.i32(i32, i32) define i32 @foo(i32 %a, i32 %b) { entry: %0 = mul i32 %a, %b %1 = call { i32, i1 } @llvm.smul.with.overflow.i32(i32 %0, i32 %b) %2 = extractvalue { i32, i1 } %1, 0 ret i32 %2 }