### test.c ```c asm("fld f0, 0(sp)"); ``` ### Output ```console $ clang test.c -c -target riscv64-unknown-linux-gnu -flto=thin error: instruction requires the following: 'D' (Double-Precision Floating-Point) fld f0, 0(sp) ^ error: instruction requires the following: 'D' (Double-Precision Floating-Point) fld f0, 0(sp) ^ ``` The D feature is enabled by default on `riscv64-unknown-linux-gnu`, so this error should normally not be emitted. However compilation still completes successfully (exit code 0) and produces valid LLVM bitcode in `test.o`. Similar bug in rustc: https://github.com/rust-lang/rust/issues/80608