[llvm] DAG: Handle lowering unordered compare with inf (PR #100378)

[Qiu Chaofan via llvm-commits]

================
@@ -117,10 +117,12 @@ define i1 @isinf_float(float %x) nounwind {
 define i1 @isinf_ppc_fp128(ppc_fp128 %x) nounwind {
 ; CHECK-LABEL: isinf_ppc_fp128:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    xststdcdp 0, 1, 48
-; CHECK-NEXT:    li 3, 0
-; CHECK-NEXT:    li 4, 1
-; CHECK-NEXT:    iseleq 3, 4, 3
+; CHECK-NEXT:    addis 3, 2, .LCPI9_0 at toc@ha
----------------
ecnelises wrote:

Need special treatment for ppcf128: isfpclas is not legal for ppcf128 but legal for double, however, `isfpclass(ppcf128)` is actually lowered to `isfpclass(double)` (higher half).

https://github.com/llvm/llvm-project/pull/100378