[llvm] [llvm] Ensure that soft float targets don't use float/vector code for memops. (PR #107022)

[Ulrich Weigand via llvm-commits]

================
@@ -2023,6 +2023,11 @@ class TargetLoweringBase {
     return LLT();
   }
 
+  bool useIntScalarMemOps(const AttributeList &FuncAttributes) const {
----------------
uweigand wrote:

Just one additional comment: "if you want to also lower all float/vector computation to soft float" is not the actual concern for the primary use case of `-msoft-float` on our platform: the Linux kernel.   This code base already never uses any float/vector computation in the first place.  The primary concern is to prevent the compiler from introducing uses of float/vector registers even for code that does have any float/vector operations at the source level.

https://github.com/llvm/llvm-project/pull/107022