[X86][CodeGen] Emit float128 libcalls for math functions

[pranavk]

Make LLVM emit libcalls to proper float128 variants for float128 types.

[llvmbot]

@llvm/pr-subscribers-backend-x86
@llvm/pr-subscribers-backend-aarch64
@llvm/pr-subscribers-llvm-globalisel

@llvm/pr-subscribers-llvm-ir

Author: Pranav Kant (pranavk)

Changes

Make LLVM emit libcalls to proper float128 variants for float128 types.

There are many other variants that may require this change. I am just putting this out for now to enable it for simple functions that are blocking me.

---

Full diff: https://github.com/llvm/llvm-project/pull/79611.diff

1 Files Affected:

• (modified) llvm/include/llvm/IR/RuntimeLibcalls.def (+3-3)

diff --git a/llvm/include/llvm/IR/RuntimeLibcalls.def b/llvm/include/llvm/IR/RuntimeLibcalls.def
index 19dea60bebf9be5..01cab92bf8414d8 100644
--- a/llvm/include/llvm/IR/RuntimeLibcalls.def
+++ b/llvm/include/llvm/IR/RuntimeLibcalls.def
@@ -252,17 +252,17 @@ HANDLE_LIBCALL(FLOOR_PPCF128, "floorl")
 HANDLE_LIBCALL(COPYSIGN_F32, "copysignf")
 HANDLE_LIBCALL(COPYSIGN_F64, "copysign")
 HANDLE_LIBCALL(COPYSIGN_F80, "copysignl")
-HANDLE_LIBCALL(COPYSIGN_F128, "copysignl")
+HANDLE_LIBCALL(COPYSIGN_F128, "copysignf128")
 HANDLE_LIBCALL(COPYSIGN_PPCF128, "copysignl")
 HANDLE_LIBCALL(FMIN_F32, "fminf")
 HANDLE_LIBCALL(FMIN_F64, "fmin")
 HANDLE_LIBCALL(FMIN_F80, "fminl")
-HANDLE_LIBCALL(FMIN_F128, "fminl")
+HANDLE_LIBCALL(FMIN_F128, "fminf128")
 HANDLE_LIBCALL(FMIN_PPCF128, "fminl")
 HANDLE_LIBCALL(FMAX_F32, "fmaxf")
 HANDLE_LIBCALL(FMAX_F64, "fmax")
 HANDLE_LIBCALL(FMAX_F80, "fmaxl")
-HANDLE_LIBCALL(FMAX_F128, "fmaxl")
+HANDLE_LIBCALL(FMAX_F128, "fmaxf128")
 HANDLE_LIBCALL(FMAX_PPCF128, "fmaxl")
 HANDLE_LIBCALL(LROUND_F32, "lroundf")
 HANDLE_LIBCALL(LROUND_F64, "lround")

[MaskRay]

This LGTM, but we'd need __builtin_*f128 clang/test/CodeGen tests.

My digging into the log says that 4bf47bc (2013) (@TNorthover) introduced these l libcall names. aarch64 long double is quad precision and its l suffix (for long double) does indicate quad precision.

glibc only introduced *f128 functions (aliases to *l on ldbl128 systems?) in 2017.

[github-actions]

✅ With the latest revision this PR passed the C/C++ code formatter.

[pranavk]

@MaskRay clang codegens correct LLVM IR. So I don't think we need any more tests in clang/test/CodeGen/. All the fp128 __builtin tests we have there already check for generation of appropriate fp128 LLVM IR calls (https://github.com/llvm/llvm-project/blob/3d71e4166de81bc3b86d127d9ac6607bda2b2755/clang/test/CodeGen/X86/math-builtins.c)

The problem is in lowering of LLVM IR into machine code which is what this patch is trying to fix for X86_64. I have changed the tests to check for that lowering.

[pranavk]

Also worth noting that x86 codegen for fp128 in clang right now seems completely broken because LLVM lowers to fFOOl variants without proper casting.

https://godbolt.org/z/drP5Wrr47

[andykaylor]

This looks good to me.