[ASan][RISCV] Teach AddressSanitizer to support indexed load/store. (#160443)

This patch is based on #159713

This patch extends AddressSanitizer to support indexed/segment
instructions in RVV. It enables proper instrumentation for these memory
operations.

A new member, `MaybeOffset`, is added to `InterestingMemoryOperand` to
describe the offset between the base pointer and the actual memory
reference address.

Co-authored-by: Yeting Kuo <yeting.kuo@sifive.com>

Author: HankChang736

llvm/include/llvm/Analysis/InterestingMemoryOperand.h

@@ -32,14 +32,19 @@ class InterestingMemoryOperand {
   Value *MaybeEVL;
   // The Stride Value, if we're looking at a strided load/store.
   Value *MaybeStride;
+  // The Offset Value, if we're looking at a indexed load/store. The
+  // offset actually means byte-offset instead of array index.
+  Value *MaybeByteOffset;
 
   InterestingMemoryOperand(Instruction *I, unsigned OperandNo, bool IsWrite,
                            class Type *OpType, MaybeAlign Alignment,
                            Value *MaybeMask = nullptr,
                            Value *MaybeEVL = nullptr,
-                           Value *MaybeStride = nullptr)
+                           Value *MaybeStride = nullptr,
+                           Value *MaybeByteOffset = nullptr)
       : IsWrite(IsWrite), OpType(OpType), Alignment(Alignment),
-        MaybeMask(MaybeMask), MaybeEVL(MaybeEVL), MaybeStride(MaybeStride) {
+        MaybeMask(MaybeMask), MaybeEVL(MaybeEVL), MaybeStride(MaybeStride),
+        MaybeByteOffset(MaybeByteOffset) {
     const DataLayout &DL = I->getDataLayout();
     TypeStoreSize = DL.getTypeStoreSizeInBits(OpType);
     PtrUse = &I->getOperandUse(OperandNo);

llvm/lib/Target/RISCV/RISCVTargetTransformInfo.cpp

@@ -2785,6 +2785,44 @@ bool RISCVTTIImpl::getTgtMemIntrinsic(IntrinsicInst *Inst,
                                           Alignment, Mask, EVL, Stride);
     return true;
   }
+  case Intrinsic::riscv_vloxei_mask:
+  case Intrinsic::riscv_vluxei_mask:
+  case Intrinsic::riscv_vsoxei_mask:
+  case Intrinsic::riscv_vsuxei_mask:
+    HasMask = true;
+    [[fallthrough]];
+  case Intrinsic::riscv_vloxei:
+  case Intrinsic::riscv_vluxei:
+  case Intrinsic::riscv_vsoxei:
+  case Intrinsic::riscv_vsuxei: {
+    // Intrinsic interface (only listed ordered version):
+    // riscv_vloxei(merge, ptr, index, vl)
+    // riscv_vloxei_mask(merge, ptr, index, mask, vl, policy)
+    // riscv_vsoxei(val, ptr, index, vl)
+    // riscv_vsoxei_mask(val, ptr, index, mask, vl, policy)
+    bool IsWrite = Inst->getType()->isVoidTy();
+    Type *Ty = IsWrite ? Inst->getArgOperand(0)->getType() : Inst->getType();
+    const auto *RVVIInfo = RISCVVIntrinsicsTable::getRISCVVIntrinsicInfo(IID);
+    unsigned VLIndex = RVVIInfo->VLOperand;
+    unsigned PtrOperandNo = VLIndex - 2 - HasMask;
+    Value *Mask;
+    if (HasMask) {
+      Mask = Inst->getArgOperand(VLIndex - 1);
+    } else {
+      // Mask cannot be nullptr here: vector GEP produces <vscale x N x ptr>,
+      // and casting that to scalar i64 triggers a vector/scalar mismatch
+      // assertion in CreatePointerCast. Use an all-true mask so ASan lowers it
+      // via extractelement instead.
+      Type *MaskType = Ty->getWithNewType(Type::getInt1Ty(C));
+      Mask = ConstantInt::getTrue(MaskType);
+    }
+    Value *EVL = Inst->getArgOperand(VLIndex);
+    Value *OffsetOp = Inst->getArgOperand(PtrOperandNo + 1);
+    Info.InterestingOperands.emplace_back(Inst, PtrOperandNo, IsWrite, Ty,
+                                          Align(1), Mask, EVL,
+                                          /* Stride */ nullptr, OffsetOp);
+    return true;
+  }
   }
   return false;
 }

llvm/lib/Transforms/Instrumentation/AddressSanitizer.cpp

@@ -1786,6 +1786,25 @@ void AddressSanitizer::instrumentMop(ObjectSizeOffsetVisitor &ObjSizeVis,
   else
     NumInstrumentedReads++;
 
+  if (O.MaybeByteOffset) {
+    Type *Ty = Type::getInt8Ty(*C);
+    IRBuilder IB(O.getInsn());
+
+    Value *OffsetOp = O.MaybeByteOffset;
+    if (TargetTriple.isRISCV()) {
+      Type *OffsetTy = OffsetOp->getType();
+      // RVV indexed loads/stores zero-extend offset operands which are narrower
+      // than XLEN to XLEN.
+      if (OffsetTy->getScalarType()->getIntegerBitWidth() <
+          static_cast<unsigned>(LongSize)) {
+        VectorType *OrigType = cast<VectorType>(OffsetTy);
+        Type *ExtendTy = VectorType::get(IntptrTy, OrigType);
+        OffsetOp = IB.CreateZExt(OffsetOp, ExtendTy);
+      }
+    }
+    Addr = IB.CreateGEP(Ty, Addr, {OffsetOp});
+  }
+
   unsigned Granularity = 1 << Mapping.Scale;
   if (O.MaybeMask) {
     instrumentMaskedLoadOrStore(this, DL, IntptrTy, O.MaybeMask, O.MaybeEVL,