[llvm] [msan] Support vst{2,3,4}_lane instructions (PR #101215)

[Vitaly Buka via llvm-commits]

================
@@ -3873,38 +3873,49 @@ struct MemorySanitizerVisitor : public InstVisitor<MemorySanitizerVisitor> {
     setOriginForNaryOp(I);
   }
 
-  /// Handle Arm NEON vector store intrinsics (vst{2,3,4} and vst1x_{2,3,4}).
+  /// Handle Arm NEON vector store intrinsics (vst{2,3,4}, vst1x_{2,3,4},
+  /// and vst{2,3,4}lane).
   ///
   /// Arm NEON vector store intrinsics have the output address (pointer) as the
-  /// last argument, with the initial arguments being the inputs. They return
-  /// void.
+  /// last argument, with the initial arguments being the inputs (and lane
+  /// number for vst{2,3,4}lane). They return void.
   ///
   /// - st4 interleaves the output e.g., st4 (inA, inB, inC, inD, outP) writes
   ///   abcdabcdabcdabcd... into *outP
   /// - st1_x4 is non-interleaved e.g., st1_x4 (inA, inB, inC, inD, outP)
   ///   writes aaaa...bbbb...cccc...dddd... into *outP
+  /// - st4lane has arguments of (inA, inB, inC, inD, lane, outP)
   /// These instructions can all be instrumented with essentially the same
   /// MSan logic, simply by applying the corresponding intrinsic to the shadow.
-  void handleNEONVectorStoreIntrinsic(IntrinsicInst &I) {
+  void handleNEONVectorStoreIntrinsic(IntrinsicInst &I, bool useLane) {
     IRBuilder<> IRB(&I);
 
     // Don't use getNumOperands() because it includes the callee
     int numArgOperands = I.arg_size();
-    assert(numArgOperands >= 1);
 
-    // The last arg operand is the output
+    // The last arg operand is the output (pointer)
+    assert(numArgOperands >= 1);
     Value *Addr = I.getArgOperand(numArgOperands - 1);
     assert(Addr->getType()->isPointerTy());
+    unsigned int skipTrailingOperands = 1;
----------------
vitalybuka wrote:

why it needs to be unsigned?
seems just `int` will simplify code

https://github.com/llvm/llvm-project/pull/101215