[llvm] [ValueTracking] Compute knownbits from known fp classes (PR #86409)

[Yingwei Zheng via llvm-commits]

================
@@ -1121,6 +1121,40 @@ static void computeKnownBitsFromOperator(const Operator *I,
       break;
     }
 
+    Value *V;
+    // Handle bitcast from floating point to integer.
+    if (match(const_cast<Operator *>(I), m_ElementWiseBitCast(m_Value(V))) &&
+        V->getType()->isFPOrFPVectorTy()) {
+      KnownFPClass Result = computeKnownFPClass(V, fcAllFlags, Depth + 1, Q);
+      if (Result.SignBit) {
+        if (*Result.SignBit)
+          Known.makeNegative();
+        else
+          Known.makeNonNegative();
+      }
+
+      Type *FPType = V->getType()->getScalarType();
+      if (FPType->isIEEELikeFPTy()) {
----------------
dtcxzyw wrote:

Should we handle `x86_fp80`? `getFPMantissaWidth` returns 64 for `x86_fp80`. But fp80 doesn't follow the IEEE floating-point format :( 
https://en.wikipedia.org/wiki/Extended_precision#x86_extended_precision_format



https://github.com/llvm/llvm-project/pull/86409