[SM6.9] Fix LV cast crash

lib/Transforms/Scalar/Float2Int.cpp

@@ -129,19 +129,25 @@ static Instruction::BinaryOps mapBinOpcode(unsigned Opcode) {
 
 // Find the roots - instructions that convert from the FP domain to
 // integer domain.
+// findRoots updated from upstream to skip vectors
 void Float2Int::findRoots(Function &F, SmallPtrSet<Instruction*,8> &Roots) {
-  for (auto &I : inst_range(F)) {
-    switch (I.getOpcode()) {
-    default: break;
-    case Instruction::FPToUI:
-    case Instruction::FPToSI:
-      Roots.insert(&I);
-      break;
-    case Instruction::FCmp:
-      if (mapFCmpPred(cast<CmpInst>(&I)->getPredicate()) != 
-          CmpInst::BAD_ICMP_PREDICATE)
+  for (BasicBlock &BB : F) {
+    for (Instruction &I : BB) {
+      if (isa<VectorType>(I.getType()))
+        continue;
+      switch (I.getOpcode()) {
+      default:
+        break;
+      case Instruction::FPToUI:
+      case Instruction::FPToSI:
         Roots.insert(&I);
-      break;
+        break;
+      case Instruction::FCmp:
+        if (mapFCmpPred(cast<CmpInst>(&I)->getPredicate()) !=
+            CmpInst::BAD_ICMP_PREDICATE)
+          Roots.insert(&I);
+        break;
+      }
     }
   }
 }
@@ -400,7 +406,7 @@ bool Float2Int::validateAndTransform() {
         R.isFullSet() || R.isSignWrappedSet())
       continue;
     assert(ConvertedToTy && "Must have set the convertedtoty by this point!");
-    
+
     // The number of bits required is the maximum of the upper and
     // lower limits, plus one so it can be signed.
     unsigned MinBW = std::max(R.getLower().getMinSignedBits(),

tools/clang/test/HLSLFileCheck/hlsl/types/cast/cast8.hlsl

@@ -0,0 +1,26 @@
+// RUN: %dxc -T cs_6_9 -DFTYPE=float2 %s | FileCheck %s
+// RUN: %dxc -T cs_6_9 -DFTYPE=half2 -enable-16bit-types %s | FileCheck %s
+
+
+// https://github.com/microsoft/DirectXShaderCompiler/issues/7915
+// Test long vector casting between uint2 and float2
+// which would crash as reported by a user.
+
+// CHECK: call %dx.types.Handle @dx.op.createHandleFromBinding
+// CHECK: fptoui
+// CHECK: uitofp
+// CHECK: fptoui
+// CHECK: uitofp
+// CHECK: call void @dx.op.rawBufferVectorStore
+
+RWStructuredBuffer<FTYPE> input;
+RWStructuredBuffer<FTYPE> output;
+
+FTYPE f1(uint2 p) { return p; }
+uint2 f2(FTYPE p) { return f1(p); }
+
+[numthreads(1,1,1)]
+void main()
+{
+  output[0] = f2(input[0]);
+}