julia

ab77314d - codegen: Configure `--expand-fp-convert-bits 64` on targets without `__int128_t` (#61439)

codegen: Configure `--expand-fp-convert-bits 64` on targets without `__int128_t` (#61439) This prevents LLVM from emitting Libcalls to `__floatuntidf`, `__floattidf`, `__fixdfti`, `__fixunsdfti` which are not included in compiler-rt / libgcc on these platforms. Resolves https://github.com/JuliaLang/julia/issues/61436. It would be good to make the JIT error for similar bugs in the future instead of hanging.