From 55c3ad279ef937320e2dad1c6ef737512206de0a Mon Sep 17 00:00:00 2001
From: Hank Chang <hank.chang@sifive.com>
Date: Mon, 11 Aug 2025 11:00:36 +0800
Subject: [PATCH 1/9] Change getTgtMemIntrinsic interface with returning a pair

* This patch change the interface of getTgtMemIntrinsic from return "bool" into return
"std::pair<std::optional<MemIntrinsicInfo>,std::optional<SmallVectorImpl<InterestingMemoryOperand>>>"
to allow Asan or EarlyCSE to choose the data structure they need with the same TTI hook.
---
 .../llvm/Analysis/TargetTransformInfo.h       |    6 +-
 .../llvm/Analysis/TargetTransformInfoImpl.h   |    7 +-
 llvm/lib/Analysis/TargetTransformInfo.cpp     |    8 +-
 .../AArch64/AArch64TargetTransformInfo.cpp    |   10 +-
 .../AArch64/AArch64TargetTransformInfo.h      |    5 +-
 .../AMDGPU/AMDGPUTargetTransformInfo.cpp      |   17 +-
 .../Target/AMDGPU/AMDGPUTargetTransformInfo.h |    5 +-
 .../Target/PowerPC/PPCTargetTransformInfo.cpp |   15 +-
 .../Target/PowerPC/PPCTargetTransformInfo.h   |    5 +-
 .../Target/RISCV/RISCVTargetTransformInfo.cpp |   80 +
 .../Target/RISCV/RISCVTargetTransformInfo.h   |    4 +
 .../Instrumentation/AddressSanitizer.cpp      |   26 +-
 llvm/lib/Transforms/Scalar/EarlyCSE.cpp       |    7 +-
 .../Transforms/Scalar/LoopStrengthReduce.cpp  |   18 +-
 .../RISCV/asan-rvv-intrinsics.ll              | 2304 +++++++++++++++++
 15 files changed, 2470 insertions(+), 47 deletions(-)
 create mode 100644 llvm/test/Instrumentation/AddressSanitizer/RISCV/asan-rvv-intrinsics.ll

diff --git a/llvm/include/llvm/Analysis/TargetTransformInfo.h b/llvm/include/llvm/Analysis/TargetTransformInfo.h
index 9186419715cc4..a5952b6d0d720 100644
--- a/llvm/include/llvm/Analysis/TargetTransformInfo.h
+++ b/llvm/include/llvm/Analysis/TargetTransformInfo.h
@@ -32,6 +32,7 @@
 #include "llvm/Support/BranchProbability.h"
 #include "llvm/Support/Compiler.h"
 #include "llvm/Support/InstructionCost.h"
+#include "llvm/Transforms/Instrumentation/AddressSanitizerCommon.h"
 #include <functional>
 #include <optional>
 #include <utility>
@@ -1696,8 +1697,9 @@ class TargetTransformInfo {
   /// will contain additional information - whether the intrinsic may write
   /// or read to memory, volatility and the pointer.  Info is undefined
   /// if false is returned.
-  LLVM_ABI bool getTgtMemIntrinsic(IntrinsicInst *Inst,
-                                   MemIntrinsicInfo &Info) const;
+  LLVM_ABI std::pair<std::optional<MemIntrinsicInfo>,
+                     std::optional<SmallVector<InterestingMemoryOperand, 1>>>
+  getTgtMemIntrinsic(IntrinsicInst *Inst) const;
 
   /// \returns The maximum element size, in bytes, for an element
   /// unordered-atomic memory intrinsic.
diff --git a/llvm/include/llvm/Analysis/TargetTransformInfoImpl.h b/llvm/include/llvm/Analysis/TargetTransformInfoImpl.h
index 200cbafbaa6e2..b6ed3097f7133 100644
--- a/llvm/include/llvm/Analysis/TargetTransformInfoImpl.h
+++ b/llvm/include/llvm/Analysis/TargetTransformInfoImpl.h
@@ -975,9 +975,10 @@ class TargetTransformInfoImplBase {
     return 0;
   }
 
-  virtual bool getTgtMemIntrinsic(IntrinsicInst *Inst,
-                                  MemIntrinsicInfo &Info) const {
-    return false;
+  virtual std::pair<std::optional<MemIntrinsicInfo>,
+                    std::optional<SmallVector<InterestingMemoryOperand, 1>>>
+  getTgtMemIntrinsic(IntrinsicInst *Inst) const {
+    return std::make_pair(std::nullopt, std::nullopt);
   }
 
   virtual unsigned getAtomicMemIntrinsicMaxElementSize() const {
diff --git a/llvm/lib/Analysis/TargetTransformInfo.cpp b/llvm/lib/Analysis/TargetTransformInfo.cpp
index 3141060a710ce..cb810601ee1f8 100644
--- a/llvm/lib/Analysis/TargetTransformInfo.cpp
+++ b/llvm/lib/Analysis/TargetTransformInfo.cpp
@@ -7,6 +7,7 @@
 //===----------------------------------------------------------------------===//
 
 #include "llvm/Analysis/TargetTransformInfo.h"
+#include "llvm/ADT/SmallVector.h"
 #include "llvm/Analysis/CFG.h"
 #include "llvm/Analysis/LoopIterator.h"
 #include "llvm/Analysis/TargetLibraryInfo.h"
@@ -1285,9 +1286,10 @@ TargetTransformInfo::getCostOfKeepingLiveOverCall(ArrayRef<Type *> Tys) const {
   return TTIImpl->getCostOfKeepingLiveOverCall(Tys);
 }
 
-bool TargetTransformInfo::getTgtMemIntrinsic(IntrinsicInst *Inst,
-                                             MemIntrinsicInfo &Info) const {
-  return TTIImpl->getTgtMemIntrinsic(Inst, Info);
+std::pair<std::optional<MemIntrinsicInfo>,
+          std::optional<SmallVector<InterestingMemoryOperand, 1>>>
+TargetTransformInfo::getTgtMemIntrinsic(IntrinsicInst *Inst) const {
+  return TTIImpl->getTgtMemIntrinsic(Inst);
 }
 
 unsigned TargetTransformInfo::getAtomicMemIntrinsicMaxElementSize() const {
diff --git a/llvm/lib/Target/AArch64/AArch64TargetTransformInfo.cpp b/llvm/lib/Target/AArch64/AArch64TargetTransformInfo.cpp
index 3042251cf754d..e9eba8311ebd5 100644
--- a/llvm/lib/Target/AArch64/AArch64TargetTransformInfo.cpp
+++ b/llvm/lib/Target/AArch64/AArch64TargetTransformInfo.cpp
@@ -5125,8 +5125,10 @@ Value *AArch64TTIImpl::getOrCreateResultFromMemIntrinsic(IntrinsicInst *Inst,
   }
 }
 
-bool AArch64TTIImpl::getTgtMemIntrinsic(IntrinsicInst *Inst,
-                                        MemIntrinsicInfo &Info) const {
+std::pair<std::optional<MemIntrinsicInfo>,
+          std::optional<SmallVector<InterestingMemoryOperand, 1>>>
+AArch64TTIImpl::getTgtMemIntrinsic(IntrinsicInst *Inst) const {
+  MemIntrinsicInfo Info;
   switch (Inst->getIntrinsicID()) {
   default:
     break;
@@ -5148,7 +5150,7 @@ bool AArch64TTIImpl::getTgtMemIntrinsic(IntrinsicInst *Inst,
 
   switch (Inst->getIntrinsicID()) {
   default:
-    return false;
+    return std::make_pair(std::nullopt, std::nullopt);
   case Intrinsic::aarch64_neon_ld2:
   case Intrinsic::aarch64_neon_st2:
     Info.MatchingId = VECTOR_LDST_TWO_ELEMENTS;
@@ -5162,7 +5164,7 @@ bool AArch64TTIImpl::getTgtMemIntrinsic(IntrinsicInst *Inst,
     Info.MatchingId = VECTOR_LDST_FOUR_ELEMENTS;
     break;
   }
-  return true;
+  return std::make_pair(Info, std::nullopt);
 }
 
 /// See if \p I should be considered for address type promotion. We check if \p
diff --git a/llvm/lib/Target/AArch64/AArch64TargetTransformInfo.h b/llvm/lib/Target/AArch64/AArch64TargetTransformInfo.h
index 9c96fdd427814..6ade1b3d08405 100644
--- a/llvm/lib/Target/AArch64/AArch64TargetTransformInfo.h
+++ b/llvm/lib/Target/AArch64/AArch64TargetTransformInfo.h
@@ -273,8 +273,9 @@ class AArch64TTIImpl final : public BasicTTIImplBase<AArch64TTIImpl> {
   getOrCreateResultFromMemIntrinsic(IntrinsicInst *Inst, Type *ExpectedType,
                                     bool CanCreate = true) const override;
 
-  bool getTgtMemIntrinsic(IntrinsicInst *Inst,
-                          MemIntrinsicInfo &Info) const override;
+  std::pair<std::optional<MemIntrinsicInfo>,
+            std::optional<SmallVector<InterestingMemoryOperand, 1>>>
+  getTgtMemIntrinsic(IntrinsicInst *Inst) const override;
 
   bool isElementTypeLegalForScalableVector(Type *Ty) const override {
     if (Ty->isPointerTy())
diff --git a/llvm/lib/Target/AMDGPU/AMDGPUTargetTransformInfo.cpp b/llvm/lib/Target/AMDGPU/AMDGPUTargetTransformInfo.cpp
index 846a0b6280f19..e99466cf81dde 100644
--- a/llvm/lib/Target/AMDGPU/AMDGPUTargetTransformInfo.cpp
+++ b/llvm/lib/Target/AMDGPU/AMDGPUTargetTransformInfo.cpp
@@ -500,29 +500,32 @@ unsigned GCNTTIImpl::getMaxInterleaveFactor(ElementCount VF) const {
   return 8;
 }
 
-bool GCNTTIImpl::getTgtMemIntrinsic(IntrinsicInst *Inst,
-                                       MemIntrinsicInfo &Info) const {
+std::pair<std::optional<MemIntrinsicInfo>,
+          std::optional<SmallVector<InterestingMemoryOperand, 1>>>
+GCNTTIImpl::getTgtMemIntrinsic(IntrinsicInst *Inst) const {
   switch (Inst->getIntrinsicID()) {
   case Intrinsic::amdgcn_ds_ordered_add:
   case Intrinsic::amdgcn_ds_ordered_swap: {
     auto *Ordering = dyn_cast<ConstantInt>(Inst->getArgOperand(2));
     auto *Volatile = dyn_cast<ConstantInt>(Inst->getArgOperand(4));
     if (!Ordering || !Volatile)
-      return false; // Invalid.
+      return std::make_pair(std::nullopt, std::nullopt);
 
     unsigned OrderingVal = Ordering->getZExtValue();
-    if (OrderingVal > static_cast<unsigned>(AtomicOrdering::SequentiallyConsistent))
-      return false;
+    if (OrderingVal >
+        static_cast<unsigned>(AtomicOrdering::SequentiallyConsistent))
+      return std::make_pair(std::nullopt, std::nullopt);
 
+    MemIntrinsicInfo Info;
     Info.PtrVal = Inst->getArgOperand(0);
     Info.Ordering = static_cast<AtomicOrdering>(OrderingVal);
     Info.ReadMem = true;
     Info.WriteMem = true;
     Info.IsVolatile = !Volatile->isZero();
-    return true;
+    return std::make_pair(Info, std::nullopt);
   }
   default:
-    return false;
+    return std::make_pair(std::nullopt, std::nullopt);
   }
 }
 
diff --git a/llvm/lib/Target/AMDGPU/AMDGPUTargetTransformInfo.h b/llvm/lib/Target/AMDGPU/AMDGPUTargetTransformInfo.h
index 20da8344c9d37..9195915758dba 100644
--- a/llvm/lib/Target/AMDGPU/AMDGPUTargetTransformInfo.h
+++ b/llvm/lib/Target/AMDGPU/AMDGPUTargetTransformInfo.h
@@ -151,8 +151,9 @@ class GCNTTIImpl final : public BasicTTIImplBase<GCNTTIImpl> {
       std::optional<uint32_t> AtomicCpySize) const override;
   unsigned getMaxInterleaveFactor(ElementCount VF) const override;
 
-  bool getTgtMemIntrinsic(IntrinsicInst *Inst,
-                          MemIntrinsicInfo &Info) const override;
+  std::pair<std::optional<MemIntrinsicInfo>,
+            std::optional<SmallVector<InterestingMemoryOperand, 1>>>
+  getTgtMemIntrinsic(IntrinsicInst *Inst) const override;
 
   InstructionCost getArithmeticInstrCost(
       unsigned Opcode, Type *Ty, TTI::TargetCostKind CostKind,
diff --git a/llvm/lib/Target/PowerPC/PPCTargetTransformInfo.cpp b/llvm/lib/Target/PowerPC/PPCTargetTransformInfo.cpp
index 2fba090f2d501..28a8ae31d708a 100644
--- a/llvm/lib/Target/PowerPC/PPCTargetTransformInfo.cpp
+++ b/llvm/lib/Target/PowerPC/PPCTargetTransformInfo.cpp
@@ -975,8 +975,10 @@ bool PPCTTIImpl::shouldBuildRelLookupTables() const {
   return BaseT::shouldBuildRelLookupTables();
 }
 
-bool PPCTTIImpl::getTgtMemIntrinsic(IntrinsicInst *Inst,
-                                    MemIntrinsicInfo &Info) const {
+std::pair<std::optional<MemIntrinsicInfo>,
+          std::optional<SmallVector<InterestingMemoryOperand, 1>>>
+PPCTTIImpl::getTgtMemIntrinsic(IntrinsicInst *Inst) const {
+  MemIntrinsicInfo Info;
   switch (Inst->getIntrinsicID()) {
   case Intrinsic::ppc_altivec_lvx:
   case Intrinsic::ppc_altivec_lvxl:
@@ -993,7 +995,7 @@ bool PPCTTIImpl::getTgtMemIntrinsic(IntrinsicInst *Inst,
     Info.PtrVal = Inst->getArgOperand(0);
     Info.ReadMem = true;
     Info.WriteMem = false;
-    return true;
+    return std::make_pair(Info, std::nullopt);
   }
   case Intrinsic::ppc_altivec_stvx:
   case Intrinsic::ppc_altivec_stvxl:
@@ -1010,7 +1012,7 @@ bool PPCTTIImpl::getTgtMemIntrinsic(IntrinsicInst *Inst,
     Info.PtrVal = Inst->getArgOperand(1);
     Info.ReadMem = false;
     Info.WriteMem = true;
-    return true;
+    return std::make_pair(Info, std::nullopt);
   }
   case Intrinsic::ppc_stbcx:
   case Intrinsic::ppc_sthcx:
@@ -1019,13 +1021,12 @@ bool PPCTTIImpl::getTgtMemIntrinsic(IntrinsicInst *Inst,
     Info.PtrVal = Inst->getArgOperand(0);
     Info.ReadMem = false;
     Info.WriteMem = true;
-    return true;
+    return std::make_pair(Info, std::nullopt);
   }
   default:
     break;
   }
-
-  return false;
+  return std::make_pair(std::nullopt, std::nullopt);
 }
 
 bool PPCTTIImpl::supportsTailCallFor(const CallBase *CB) const {
diff --git a/llvm/lib/Target/PowerPC/PPCTargetTransformInfo.h b/llvm/lib/Target/PowerPC/PPCTargetTransformInfo.h
index 475472ac3720f..923931f14bc77 100644
--- a/llvm/lib/Target/PowerPC/PPCTargetTransformInfo.h
+++ b/llvm/lib/Target/PowerPC/PPCTargetTransformInfo.h
@@ -69,8 +69,9 @@ class PPCTTIImpl final : public BasicTTIImplBase<PPCTTIImpl> {
   bool canSaveCmp(Loop *L, BranchInst **BI, ScalarEvolution *SE, LoopInfo *LI,
                   DominatorTree *DT, AssumptionCache *AC,
                   TargetLibraryInfo *LibInfo) const override;
-  bool getTgtMemIntrinsic(IntrinsicInst *Inst,
-                          MemIntrinsicInfo &Info) const override;
+  std::pair<std::optional<MemIntrinsicInfo>,
+            std::optional<SmallVector<InterestingMemoryOperand, 1>>>
+  getTgtMemIntrinsic(IntrinsicInst *Inst) const override;
   void getUnrollingPreferences(Loop *L, ScalarEvolution &SE,
                                TTI::UnrollingPreferences &UP,
                                OptimizationRemarkEmitter *ORE) const override;
diff --git a/llvm/lib/Target/RISCV/RISCVTargetTransformInfo.cpp b/llvm/lib/Target/RISCV/RISCVTargetTransformInfo.cpp
index af78b3cc2c7ff..7ee153278208d 100644
--- a/llvm/lib/Target/RISCV/RISCVTargetTransformInfo.cpp
+++ b/llvm/lib/Target/RISCV/RISCVTargetTransformInfo.cpp
@@ -15,6 +15,7 @@
 #include "llvm/CodeGen/TargetLowering.h"
 #include "llvm/CodeGen/ValueTypes.h"
 #include "llvm/IR/Instructions.h"
+#include "llvm/IR/IntrinsicsRISCV.h"
 #include "llvm/IR/PatternMatch.h"
 #include <cmath>
 #include <optional>
@@ -2673,6 +2674,85 @@ void RISCVTTIImpl::getPeelingPreferences(Loop *L, ScalarEvolution &SE,
   BaseT::getPeelingPreferences(L, SE, PP);
 }
 
+std::pair<std::optional<MemIntrinsicInfo>,
+          std::optional<SmallVector<InterestingMemoryOperand, 1>>>
+RISCVTTIImpl::getTgtMemIntrinsic(IntrinsicInst *Inst) const {
+  const DataLayout &DL = getDataLayout();
+  Intrinsic::ID IID = Inst->getIntrinsicID();
+  LLVMContext &C = Inst->getContext();
+  bool HasMask = false;
+  switch (IID) {
+  case Intrinsic::riscv_vle_mask:
+  case Intrinsic::riscv_vse_mask:
+    HasMask = true;
+    [[fallthrough]];
+  case Intrinsic::riscv_vle:
+  case Intrinsic::riscv_vse: {
+    // Intrinsic interface:
+    // riscv_vle(merge, ptr, vl)
+    // riscv_vle_mask(merge, ptr, mask, vl, policy)
+    // riscv_vse(val, ptr, vl)
+    // riscv_vse_mask(val, ptr, mask, vl, policy)
+    SmallVector<InterestingMemoryOperand, 1> Interesting;
+    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 - 1 - HasMask;
+    MaybeAlign Alignment =
+        Inst->getArgOperand(PtrOperandNo)->getPointerAlignment(DL);
+    Type *MaskType = Ty->getWithNewType(Type::getInt1Ty(C));
+    Value *Mask = ConstantInt::getTrue(MaskType);
+    if (HasMask)
+      Mask = Inst->getArgOperand(VLIndex - 1);
+    Value *EVL = Inst->getArgOperand(VLIndex);
+    Interesting.emplace_back(Inst, PtrOperandNo, IsWrite, Ty, Alignment, Mask,
+                             EVL);
+    return std::make_pair(std::nullopt, Interesting);
+  }
+  case Intrinsic::riscv_vlse_mask:
+  case Intrinsic::riscv_vsse_mask:
+    HasMask = true;
+    [[fallthrough]];
+  case Intrinsic::riscv_vlse:
+  case Intrinsic::riscv_vsse: {
+    // Intrinsic interface:
+    // riscv_vlse(merge, ptr, stride, vl)
+    // riscv_vlse_mask(merge, ptr, stride, mask, vl, policy)
+    // riscv_vsse(val, ptr, stride, vl)
+    // riscv_vsse_mask(val, ptr, stride, mask, vl, policy)
+    SmallVector<InterestingMemoryOperand, 1> Interesting;
+    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;
+    MaybeAlign Alignment =
+        Inst->getArgOperand(PtrOperandNo)->getPointerAlignment(DL);
+
+    Value *Stride = Inst->getArgOperand(PtrOperandNo + 1);
+    // Use the pointer alignment as the element alignment if the stride is a
+    // multiple of the pointer alignment. Otherwise, the element alignment
+    // should be the greatest common divisor of pointer alignment and stride.
+    // For simplicity, just consider unalignment for elements.
+    unsigned PointerAlign = Alignment.valueOrOne().value();
+    if (!isa<ConstantInt>(Stride) ||
+        cast<ConstantInt>(Stride)->getZExtValue() % PointerAlign != 0)
+      Alignment = Align(1);
+
+    Type *MaskType = Ty->getWithNewType(Type::getInt1Ty(C));
+    Value *Mask = ConstantInt::getTrue(MaskType);
+    if (HasMask)
+      Mask = Inst->getArgOperand(VLIndex - 1);
+    Value *EVL = Inst->getArgOperand(VLIndex);
+    Interesting.emplace_back(Inst, PtrOperandNo, IsWrite, Ty, Alignment, Mask,
+                             EVL, Stride);
+    return std::make_pair(std::nullopt, Interesting);
+  }
+  }
+  return std::make_pair(std::nullopt, std::nullopt);
+}
+
 unsigned RISCVTTIImpl::getRegUsageForType(Type *Ty) const {
   if (Ty->isVectorTy()) {
     // f16 with only zvfhmin and bf16 will be promoted to f32
diff --git a/llvm/lib/Target/RISCV/RISCVTargetTransformInfo.h b/llvm/lib/Target/RISCV/RISCVTargetTransformInfo.h
index 6a1f4b3e3bedf..d8a829ac21b64 100644
--- a/llvm/lib/Target/RISCV/RISCVTargetTransformInfo.h
+++ b/llvm/lib/Target/RISCV/RISCVTargetTransformInfo.h
@@ -158,6 +158,10 @@ class RISCVTTIImpl final : public BasicTTIImplBase<RISCVTTIImpl> {
   void getPeelingPreferences(Loop *L, ScalarEvolution &SE,
                              TTI::PeelingPreferences &PP) const override;
 
+  std::pair<std::optional<MemIntrinsicInfo>,
+            std::optional<SmallVector<InterestingMemoryOperand, 1>>>
+  getTgtMemIntrinsic(IntrinsicInst *Inst) const override;
+
   unsigned getMinVectorRegisterBitWidth() const override {
     return ST->useRVVForFixedLengthVectors() ? 16 : 0;
   }
diff --git a/llvm/lib/Transforms/Instrumentation/AddressSanitizer.cpp b/llvm/lib/Transforms/Instrumentation/AddressSanitizer.cpp
index 50258af5e26c3..f413c5eb7a2bb 100644
--- a/llvm/lib/Transforms/Instrumentation/AddressSanitizer.cpp
+++ b/llvm/lib/Transforms/Instrumentation/AddressSanitizer.cpp
@@ -29,6 +29,7 @@
 #include "llvm/Analysis/MemoryBuiltins.h"
 #include "llvm/Analysis/StackSafetyAnalysis.h"
 #include "llvm/Analysis/TargetLibraryInfo.h"
+#include "llvm/Analysis/TargetTransformInfo.h"
 #include "llvm/Analysis/ValueTracking.h"
 #include "llvm/BinaryFormat/MachO.h"
 #include "llvm/Demangle/Demangle.h"
@@ -803,7 +804,8 @@ struct AddressSanitizer {
 
   bool ignoreAccess(Instruction *Inst, Value *Ptr);
   void getInterestingMemoryOperands(
-      Instruction *I, SmallVectorImpl<InterestingMemoryOperand> &Interesting);
+      Instruction *I, SmallVectorImpl<InterestingMemoryOperand> &Interesting,
+      const TargetTransformInfo *TTI);
 
   void instrumentMop(ObjectSizeOffsetVisitor &ObjSizeVis,
                      InterestingMemoryOperand &O, bool UseCalls,
@@ -843,7 +845,7 @@ struct AddressSanitizer {
   void instrumentMemIntrinsic(MemIntrinsic *MI, RuntimeCallInserter &RTCI);
   Value *memToShadow(Value *Shadow, IRBuilder<> &IRB);
   bool suppressInstrumentationSiteForDebug(int &Instrumented);
-  bool instrumentFunction(Function &F, const TargetLibraryInfo *TLI);
+  bool instrumentFunction(Function &F, const TargetLibraryInfo *TLI, const TargetTransformInfo *TTI);
   bool maybeInsertAsanInitAtFunctionEntry(Function &F);
   bool maybeInsertDynamicShadowAtFunctionEntry(Function &F);
   void markEscapedLocalAllocas(Function &F);
@@ -1312,7 +1314,8 @@ PreservedAnalyses AddressSanitizerPass::run(Module &M,
         Options.MaxInlinePoisoningSize, Options.CompileKernel, Options.Recover,
         Options.UseAfterScope, Options.UseAfterReturn);
     const TargetLibraryInfo &TLI = FAM.getResult<TargetLibraryAnalysis>(F);
-    Modified |= FunctionSanitizer.instrumentFunction(F, &TLI);
+    const TargetTransformInfo &TTI = FAM.getResult<TargetIRAnalysis>(F);
+    Modified |= FunctionSanitizer.instrumentFunction(F, &TLI, &TTI);
   }
   Modified |= ModuleSanitizer.instrumentModule();
   if (!Modified)
@@ -1450,7 +1453,8 @@ bool AddressSanitizer::ignoreAccess(Instruction *Inst, Value *Ptr) {
 }
 
 void AddressSanitizer::getInterestingMemoryOperands(
-    Instruction *I, SmallVectorImpl<InterestingMemoryOperand> &Interesting) {
+    Instruction *I, SmallVectorImpl<InterestingMemoryOperand> &Interesting,
+    const TargetTransformInfo *TTI) {
   // Do not instrument the load fetching the dynamic shadow address.
   if (LocalDynamicShadow == I)
     return;
@@ -1568,6 +1572,15 @@ void AddressSanitizer::getInterestingMemoryOperands(
       break;
     }
     default:
+      if (auto *II = dyn_cast<IntrinsicInst>(I)) {
+        std::pair<std::optional<MemIntrinsicInfo>,
+                  std::optional<SmallVector<InterestingMemoryOperand, 1>>>
+            MemInfo;
+        MemInfo = TTI->getTgtMemIntrinsic(II);
+        if (MemInfo.second != std::nullopt)
+          Interesting = *MemInfo.second;
+        return;
+      }
       for (unsigned ArgNo = 0; ArgNo < CI->arg_size(); ArgNo++) {
         if (!ClInstrumentByval || !CI->isByValArgument(ArgNo) ||
             ignoreAccess(I, CI->getArgOperand(ArgNo)))
@@ -2983,7 +2996,8 @@ bool AddressSanitizer::suppressInstrumentationSiteForDebug(int &Instrumented) {
 }
 
 bool AddressSanitizer::instrumentFunction(Function &F,
-                                          const TargetLibraryInfo *TLI) {
+                                          const TargetLibraryInfo *TLI,
+					  const TargetTransformInfo *TTI) {
   bool FunctionModified = false;
 
   // Do not apply any instrumentation for naked functions.
@@ -3036,7 +3050,7 @@ bool AddressSanitizer::instrumentFunction(Function &F,
       if (Inst.hasMetadata(LLVMContext::MD_nosanitize))
         continue;
       SmallVector<InterestingMemoryOperand, 1> InterestingOperands;
-      getInterestingMemoryOperands(&Inst, InterestingOperands);
+      getInterestingMemoryOperands(&Inst, InterestingOperands, TTI);
 
       if (!InterestingOperands.empty()) {
         for (auto &Operand : InterestingOperands) {
diff --git a/llvm/lib/Transforms/Scalar/EarlyCSE.cpp b/llvm/lib/Transforms/Scalar/EarlyCSE.cpp
index 0f8cc6ca6ed21..5fb07026672bc 100644
--- a/llvm/lib/Transforms/Scalar/EarlyCSE.cpp
+++ b/llvm/lib/Transforms/Scalar/EarlyCSE.cpp
@@ -808,8 +808,6 @@ class EarlyCSE {
       : Inst(Inst) {
       if (IntrinsicInst *II = dyn_cast<IntrinsicInst>(Inst)) {
         IntrID = II->getIntrinsicID();
-        if (TTI.getTgtMemIntrinsic(II, Info))
-          return;
         if (isHandledNonTargetIntrinsic(IntrID)) {
           switch (IntrID) {
           case Intrinsic::masked_load:
@@ -834,6 +832,11 @@ class EarlyCSE {
             break;
           }
         }
+        std::pair<std::optional<MemIntrinsicInfo>,
+                  std::optional<SmallVector<InterestingMemoryOperand, 1>>>
+            MemInfo = TTI.getTgtMemIntrinsic(II);
+        if (MemInfo.first != std::nullopt)
+          Info = *MemInfo.first;
       }
     }
 
diff --git a/llvm/lib/Transforms/Scalar/LoopStrengthReduce.cpp b/llvm/lib/Transforms/Scalar/LoopStrengthReduce.cpp
index e3ef9d8680b53..049087a59abe6 100644
--- a/llvm/lib/Transforms/Scalar/LoopStrengthReduce.cpp
+++ b/llvm/lib/Transforms/Scalar/LoopStrengthReduce.cpp
@@ -1018,9 +1018,11 @@ static bool isAddressUse(const TargetTransformInfo &TTI,
         isAddress = true;
       break;
     default: {
-      MemIntrinsicInfo IntrInfo;
-      if (TTI.getTgtMemIntrinsic(II, IntrInfo)) {
-        if (IntrInfo.PtrVal == OperandVal)
+      std::pair<std::optional<MemIntrinsicInfo>,
+                std::optional<SmallVector<InterestingMemoryOperand, 1>>>
+          MemInfo = TTI.getTgtMemIntrinsic(II);
+      if (MemInfo.first != std::nullopt) {
+        if (MemInfo.first->PtrVal == OperandVal)
           isAddress = true;
       }
     }
@@ -1074,10 +1076,12 @@ static MemAccessTy getAccessType(const TargetTransformInfo &TTI,
           II->getArgOperand(1)->getType()->getPointerAddressSpace();
       break;
     default: {
-      MemIntrinsicInfo IntrInfo;
-      if (TTI.getTgtMemIntrinsic(II, IntrInfo) && IntrInfo.PtrVal) {
-        AccessTy.AddrSpace
-          = IntrInfo.PtrVal->getType()->getPointerAddressSpace();
+      std::pair<std::optional<MemIntrinsicInfo>,
+                std::optional<SmallVector<InterestingMemoryOperand, 1>>>
+          MemInfo = TTI.getTgtMemIntrinsic(II);
+      if (MemInfo.first != std::nullopt && MemInfo.first->PtrVal) {
+        AccessTy.AddrSpace =
+            MemInfo.first->PtrVal->getType()->getPointerAddressSpace();
       }
 
       break;
diff --git a/llvm/test/Instrumentation/AddressSanitizer/RISCV/asan-rvv-intrinsics.ll b/llvm/test/Instrumentation/AddressSanitizer/RISCV/asan-rvv-intrinsics.ll
new file mode 100644
index 0000000000000..34b88daca4025
--- /dev/null
+++ b/llvm/test/Instrumentation/AddressSanitizer/RISCV/asan-rvv-intrinsics.ll
@@ -0,0 +1,2304 @@
+; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
+; RUN: opt < %s -mtriple=riscv64 -mattr=+v -passes=asan \
+; RUN: -asan-instrumentation-with-call-threshold=0 -S | FileCheck %s
+
+declare <vscale x 1 x i32> @llvm.riscv.vle.nxv1i32(
+  <vscale x 1 x i32>,
+  <vscale x 1 x i32>*,
+  i64)
+define <vscale x 1 x i32> @intrinsic_vle_v_nxv1i32_nxv1i32(<vscale x 1 x i32>* align 4 %0, i64 %1) sanitize_address {
+; CHECK-LABEL: @intrinsic_vle_v_nxv1i32_nxv1i32(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP2:%.*]] = load i64, ptr @__asan_shadow_memory_dynamic_address, align 8
+; CHECK-NEXT:    [[TMP3:%.*]] = icmp ne i64 [[TMP1:%.*]], 0
+; CHECK-NEXT:    br i1 [[TMP3]], label [[TMP4:%.*]], label [[TMP12:%.*]]
+; CHECK:       4:
+; CHECK-NEXT:    [[TMP5:%.*]] = call i64 @llvm.vscale.i64()
+; CHECK-NEXT:    [[TMP6:%.*]] = call i64 @llvm.umin.i64(i64 [[TMP1]], i64 [[TMP5]])
+; CHECK-NEXT:    br label [[DOTSPLIT:%.*]]
+; CHECK:       .split:
+; CHECK-NEXT:    [[IV:%.*]] = phi i64 [ 0, [[TMP4]] ], [ [[IV_NEXT:%.*]], [[TMP11:%.*]] ]
+; CHECK-NEXT:    [[TMP7:%.*]] = extractelement <vscale x 1 x i1> splat (i1 true), i64 [[IV]]
+; CHECK-NEXT:    br i1 [[TMP7]], label [[TMP8:%.*]], label [[TMP11]]
+; CHECK:       8:
+; CHECK-NEXT:    [[TMP9:%.*]] = getelementptr <vscale x 1 x i32>, ptr [[TMP0:%.*]], i64 0, i64 [[IV]]
+; CHECK-NEXT:    [[TMP10:%.*]] = ptrtoint ptr [[TMP9]] to i64
+; CHECK-NEXT:    call void @__asan_load4(i64 [[TMP10]])
+; CHECK-NEXT:    br label [[TMP11]]
+; CHECK:       11:
+; CHECK-NEXT:    [[IV_NEXT]] = add nuw nsw i64 [[IV]], 1
+; CHECK-NEXT:    [[IV_CHECK:%.*]] = icmp eq i64 [[IV_NEXT]], [[TMP6]]
+; CHECK-NEXT:    br i1 [[IV_CHECK]], label [[DOTSPLIT_SPLIT:%.*]], label [[DOTSPLIT]]
+; CHECK:       .split.split:
+; CHECK-NEXT:    br label [[TMP12]]
+; CHECK:       12:
+; CHECK-NEXT:    [[A:%.*]] = call <vscale x 1 x i32> @llvm.riscv.vle.nxv1i32.p0.i64(<vscale x 1 x i32> undef, ptr [[TMP0]], i64 [[TMP1]])
+; CHECK-NEXT:    ret <vscale x 1 x i32> [[A]]
+;
+entry:
+  %a = call <vscale x 1 x i32> @llvm.riscv.vle.nxv1i32(
+  <vscale x 1 x i32> undef,
+  <vscale x 1 x i32>* %0,
+  i64 %1)
+  ret <vscale x 1 x i32> %a
+}
+
+declare <vscale x 1 x i32> @llvm.riscv.vle.mask.nxv1i32(
+  <vscale x 1 x i32>,
+  <vscale x 1 x i32>*,
+  <vscale x 1 x i1>,
+  i64,
+  i64)
+define <vscale x 1 x i32> @intrinsic_vle_mask_v_nxv1i32_nxv1i32(<vscale x 1 x i32> %0, <vscale x 1 x i32>* align 4 %1, <vscale x 1 x i1> %2, i64 %3) sanitize_address {
+; CHECK-LABEL: @intrinsic_vle_mask_v_nxv1i32_nxv1i32(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP4:%.*]] = load i64, ptr @__asan_shadow_memory_dynamic_address, align 8
+; CHECK-NEXT:    [[TMP5:%.*]] = icmp ne i64 [[TMP3:%.*]], 0
+; CHECK-NEXT:    br i1 [[TMP5]], label [[TMP6:%.*]], label [[TMP14:%.*]]
+; CHECK:       6:
+; CHECK-NEXT:    [[TMP7:%.*]] = call i64 @llvm.vscale.i64()
+; CHECK-NEXT:    [[TMP8:%.*]] = call i64 @llvm.umin.i64(i64 [[TMP3]], i64 [[TMP7]])
+; CHECK-NEXT:    br label [[DOTSPLIT:%.*]]
+; CHECK:       .split:
+; CHECK-NEXT:    [[IV:%.*]] = phi i64 [ 0, [[TMP6]] ], [ [[IV_NEXT:%.*]], [[TMP13:%.*]] ]
+; CHECK-NEXT:    [[TMP9:%.*]] = extractelement <vscale x 1 x i1> [[TMP2:%.*]], i64 [[IV]]
+; CHECK-NEXT:    br i1 [[TMP9]], label [[TMP10:%.*]], label [[TMP13]]
+; CHECK:       10:
+; CHECK-NEXT:    [[TMP11:%.*]] = getelementptr <vscale x 1 x i32>, ptr [[TMP1:%.*]], i64 0, i64 [[IV]]
+; CHECK-NEXT:    [[TMP12:%.*]] = ptrtoint ptr [[TMP11]] to i64
+; CHECK-NEXT:    call void @__asan_load4(i64 [[TMP12]])
+; CHECK-NEXT:    br label [[TMP13]]
+; CHECK:       13:
+; CHECK-NEXT:    [[IV_NEXT]] = add nuw nsw i64 [[IV]], 1
+; CHECK-NEXT:    [[IV_CHECK:%.*]] = icmp eq i64 [[IV_NEXT]], [[TMP8]]
+; CHECK-NEXT:    br i1 [[IV_CHECK]], label [[DOTSPLIT_SPLIT:%.*]], label [[DOTSPLIT]]
+; CHECK:       .split.split:
+; CHECK-NEXT:    br label [[TMP14]]
+; CHECK:       14:
+; CHECK-NEXT:    [[A:%.*]] = call <vscale x 1 x i32> @llvm.riscv.vle.mask.nxv1i32.p0.i64(<vscale x 1 x i32> [[TMP0:%.*]], ptr [[TMP1]], <vscale x 1 x i1> [[TMP2]], i64 [[TMP3]], i64 1)
+; CHECK-NEXT:    ret <vscale x 1 x i32> [[A]]
+;
+entry:
+  %a = call <vscale x 1 x i32> @llvm.riscv.vle.mask.nxv1i32(
+  <vscale x 1 x i32> %0,
+  <vscale x 1 x i32>* %1,
+  <vscale x 1 x i1> %2,
+  i64 %3, i64 1)
+  ret <vscale x 1 x i32> %a
+}
+
+declare void @llvm.riscv.vse.nxv1i32(
+  <vscale x 1 x i32>,
+  <vscale x 1 x i32>*,
+  i64)
+define void @intrinsic_vse_v_nxv1i32_nxv1i32(<vscale x 1 x i32> %0, <vscale x 1 x i32>* align 4 %1, i64 %2) sanitize_address {
+; CHECK-LABEL: @intrinsic_vse_v_nxv1i32_nxv1i32(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP3:%.*]] = load i64, ptr @__asan_shadow_memory_dynamic_address, align 8
+; CHECK-NEXT:    [[TMP4:%.*]] = icmp ne i64 [[TMP2:%.*]], 0
+; CHECK-NEXT:    br i1 [[TMP4]], label [[TMP5:%.*]], label [[TMP13:%.*]]
+; CHECK:       5:
+; CHECK-NEXT:    [[TMP6:%.*]] = call i64 @llvm.vscale.i64()
+; CHECK-NEXT:    [[TMP7:%.*]] = call i64 @llvm.umin.i64(i64 [[TMP2]], i64 [[TMP6]])
+; CHECK-NEXT:    br label [[DOTSPLIT:%.*]]
+; CHECK:       .split:
+; CHECK-NEXT:    [[IV:%.*]] = phi i64 [ 0, [[TMP5]] ], [ [[IV_NEXT:%.*]], [[TMP12:%.*]] ]
+; CHECK-NEXT:    [[TMP8:%.*]] = extractelement <vscale x 1 x i1> splat (i1 true), i64 [[IV]]
+; CHECK-NEXT:    br i1 [[TMP8]], label [[TMP9:%.*]], label [[TMP12]]
+; CHECK:       9:
+; CHECK-NEXT:    [[TMP10:%.*]] = getelementptr <vscale x 1 x i32>, ptr [[TMP1:%.*]], i64 0, i64 [[IV]]
+; CHECK-NEXT:    [[TMP11:%.*]] = ptrtoint ptr [[TMP10]] to i64
+; CHECK-NEXT:    call void @__asan_store4(i64 [[TMP11]])
+; CHECK-NEXT:    br label [[TMP12]]
+; CHECK:       12:
+; CHECK-NEXT:    [[IV_NEXT]] = add nuw nsw i64 [[IV]], 1
+; CHECK-NEXT:    [[IV_CHECK:%.*]] = icmp eq i64 [[IV_NEXT]], [[TMP7]]
+; CHECK-NEXT:    br i1 [[IV_CHECK]], label [[DOTSPLIT_SPLIT:%.*]], label [[DOTSPLIT]]
+; CHECK:       .split.split:
+; CHECK-NEXT:    br label [[TMP13]]
+; CHECK:       13:
+; CHECK-NEXT:    call void @llvm.riscv.vse.nxv1i32.p0.i64(<vscale x 1 x i32> [[TMP0:%.*]], ptr [[TMP1]], i64 [[TMP2]])
+; CHECK-NEXT:    ret void
+;
+entry:
+  call void @llvm.riscv.vse.nxv1i32(
+  <vscale x 1 x i32> %0,
+  <vscale x 1 x i32>* %1,
+  i64 %2)
+  ret void
+}
+
+declare void @llvm.riscv.vse.mask.nxv1i32(
+  <vscale x 1 x i32>,
+  <vscale x 1 x i32>*,
+  <vscale x 1 x i1>,
+  i64)
+define void @intrinsic_vse_mask_v_nxv1i32_nxv1i32(<vscale x 1 x i32> %0, <vscale x 1 x i32>* align 4 %1, <vscale x 1 x i1> %2, i64 %3) sanitize_address {
+; CHECK-LABEL: @intrinsic_vse_mask_v_nxv1i32_nxv1i32(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP4:%.*]] = load i64, ptr @__asan_shadow_memory_dynamic_address, align 8
+; CHECK-NEXT:    [[TMP5:%.*]] = icmp ne i64 [[TMP3:%.*]], 0
+; CHECK-NEXT:    br i1 [[TMP5]], label [[TMP6:%.*]], label [[TMP14:%.*]]
+; CHECK:       6:
+; CHECK-NEXT:    [[TMP7:%.*]] = call i64 @llvm.vscale.i64()
+; CHECK-NEXT:    [[TMP8:%.*]] = call i64 @llvm.umin.i64(i64 [[TMP3]], i64 [[TMP7]])
+; CHECK-NEXT:    br label [[DOTSPLIT:%.*]]
+; CHECK:       .split:
+; CHECK-NEXT:    [[IV:%.*]] = phi i64 [ 0, [[TMP6]] ], [ [[IV_NEXT:%.*]], [[TMP13:%.*]] ]
+; CHECK-NEXT:    [[TMP9:%.*]] = extractelement <vscale x 1 x i1> [[TMP2:%.*]], i64 [[IV]]
+; CHECK-NEXT:    br i1 [[TMP9]], label [[TMP10:%.*]], label [[TMP13]]
+; CHECK:       10:
+; CHECK-NEXT:    [[TMP11:%.*]] = getelementptr <vscale x 1 x i32>, ptr [[TMP1:%.*]], i64 0, i64 [[IV]]
+; CHECK-NEXT:    [[TMP12:%.*]] = ptrtoint ptr [[TMP11]] to i64
+; CHECK-NEXT:    call void @__asan_store4(i64 [[TMP12]])
+; CHECK-NEXT:    br label [[TMP13]]
+; CHECK:       13:
+; CHECK-NEXT:    [[IV_NEXT]] = add nuw nsw i64 [[IV]], 1
+; CHECK-NEXT:    [[IV_CHECK:%.*]] = icmp eq i64 [[IV_NEXT]], [[TMP8]]
+; CHECK-NEXT:    br i1 [[IV_CHECK]], label [[DOTSPLIT_SPLIT:%.*]], label [[DOTSPLIT]]
+; CHECK:       .split.split:
+; CHECK-NEXT:    br label [[TMP14]]
+; CHECK:       14:
+; CHECK-NEXT:    call void @llvm.riscv.vse.mask.nxv1i32.p0.i64(<vscale x 1 x i32> [[TMP0:%.*]], ptr [[TMP1]], <vscale x 1 x i1> [[TMP2]], i64 [[TMP3]])
+; CHECK-NEXT:    ret void
+;
+entry:
+  call void @llvm.riscv.vse.mask.nxv1i32(
+  <vscale x 1 x i32> %0,
+  <vscale x 1 x i32>* %1,
+  <vscale x 1 x i1> %2,
+  i64 %3)
+  ret void
+}
+
+
+declare target("riscv.vector.tuple", <vscale x 4 x i8>, 2) @llvm.riscv.vlseg2.triscv.vector.tuple_nxv4i8_2t(target("riscv.vector.tuple", <vscale x 4 x i8>, 2), ptr, i64, i64)
+declare target("riscv.vector.tuple", <vscale x 4 x i8>, 2) @llvm.riscv.vlseg2.mask.triscv.vector.tuple_nxv4i8_2t.nxv1i1(target("riscv.vector.tuple", <vscale x 4 x i8>, 2), ptr, <vscale x 1 x i1>, i64, i64, i64)
+
+define <vscale x 1 x i32> @test_vlseg2_nxv1i32(ptr %base, i64 %vl) sanitize_address {
+; CHECK-LABEL: @test_vlseg2_nxv1i32(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i64, ptr @__asan_shadow_memory_dynamic_address, align 8
+; CHECK-NEXT:    [[TMP24:%.*]] = tail call target("riscv.vector.tuple", <vscale x 4 x i8>, 2) @llvm.riscv.vlseg2.triscv.vector.tuple_nxv4i8_2t.p0.i64(target("riscv.vector.tuple", <vscale x 4 x i8>, 2) undef, ptr [[BASE:%.*]], i64 [[VL:%.*]], i64 5)
+; CHECK-NEXT:    [[TMP25:%.*]] = call <vscale x 1 x i32> @llvm.riscv.tuple.extract.nxv1i32.triscv.vector.tuple_nxv4i8_2t(target("riscv.vector.tuple", <vscale x 4 x i8>, 2) [[TMP24]], i32 1)
+; CHECK-NEXT:    ret <vscale x 1 x i32> [[TMP25]]
+;
+entry:
+  %0 = tail call target("riscv.vector.tuple", <vscale x 4 x i8>, 2) @llvm.riscv.vlseg2.triscv.vector.tuple_nxv4i8_2t(target("riscv.vector.tuple", <vscale x 4 x i8>, 2) undef, ptr %base, i64 %vl, i64 5)
+  %1 = call <vscale x 1 x i32> @llvm.riscv.tuple.extract.nxv1i32.triscv.vector.tuple_nxv4i8_2t(target("riscv.vector.tuple", <vscale x 4 x i8>, 2) %0, i32 1)
+  ret <vscale x 1 x i32> %1
+}
+
+define <vscale x 1 x i32> @test_vlseg2_mask_nxv1i32(ptr %base, i64 %vl, <vscale x 1 x i1> %mask) sanitize_address {
+; CHECK-LABEL: @test_vlseg2_mask_nxv1i32(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i64, ptr @__asan_shadow_memory_dynamic_address, align 8
+; CHECK-NEXT:    [[TMP24:%.*]] = tail call target("riscv.vector.tuple", <vscale x 4 x i8>, 2) @llvm.riscv.vlseg2.mask.triscv.vector.tuple_nxv4i8_2t.p0.nxv1i1.i64(target("riscv.vector.tuple", <vscale x 4 x i8>, 2) undef, ptr [[BASE:%.*]], <vscale x 1 x i1> [[MASK:%.*]], i64 [[VL:%.*]], i64 1, i64 5)
+; CHECK-NEXT:    [[TMP25:%.*]] = call <vscale x 1 x i32> @llvm.riscv.tuple.extract.nxv1i32.triscv.vector.tuple_nxv4i8_2t(target("riscv.vector.tuple", <vscale x 4 x i8>, 2) [[TMP24]], i32 1)
+; CHECK-NEXT:    ret <vscale x 1 x i32> [[TMP25]]
+;
+entry:
+  %0 = tail call target("riscv.vector.tuple", <vscale x 4 x i8>, 2) @llvm.riscv.vlseg2.mask.triscv.vector.tuple_nxv4i8_2t.nxv1i1(target("riscv.vector.tuple", <vscale x 4 x i8>, 2) undef, ptr %base, <vscale x 1 x i1> %mask, i64 %vl, i64 1, i64 5)
+  %1 = call <vscale x 1 x i32> @llvm.riscv.tuple.extract.nxv1i32.triscv.vector.tuple_nxv4i8_2t(target("riscv.vector.tuple", <vscale x 4 x i8>, 2) %0, i32 1)
+  ret <vscale x 1 x i32> %1
+}
+
+
+declare target("riscv.vector.tuple", <vscale x 4 x i8>, 3) @llvm.riscv.vlseg3.triscv.vector.tuple_nxv4i8_3t(target("riscv.vector.tuple", <vscale x 4 x i8>, 3), ptr, i64, i64)
+declare target("riscv.vector.tuple", <vscale x 4 x i8>, 3) @llvm.riscv.vlseg3.mask.triscv.vector.tuple_nxv4i8_3t.nxv1i1(target("riscv.vector.tuple", <vscale x 4 x i8>, 3), ptr, <vscale x 1 x i1>, i64, i64, i64)
+
+define <vscale x 1 x i32> @test_vlseg3_nxv1i32(ptr %base, i64 %vl) sanitize_address {
+; CHECK-LABEL: @test_vlseg3_nxv1i32(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i64, ptr @__asan_shadow_memory_dynamic_address, align 8
+; CHECK-NEXT:    [[TMP36:%.*]] = tail call target("riscv.vector.tuple", <vscale x 4 x i8>, 3) @llvm.riscv.vlseg3.triscv.vector.tuple_nxv4i8_3t.p0.i64(target("riscv.vector.tuple", <vscale x 4 x i8>, 3) undef, ptr [[BASE:%.*]], i64 [[VL:%.*]], i64 5)
+; CHECK-NEXT:    [[TMP37:%.*]] = call <vscale x 1 x i32> @llvm.riscv.tuple.extract.nxv1i32.triscv.vector.tuple_nxv4i8_3t(target("riscv.vector.tuple", <vscale x 4 x i8>, 3) [[TMP36]], i32 1)
+; CHECK-NEXT:    ret <vscale x 1 x i32> [[TMP37]]
+;
+entry:
+  %0 = tail call target("riscv.vector.tuple", <vscale x 4 x i8>, 3) @llvm.riscv.vlseg3.triscv.vector.tuple_nxv4i8_3t(target("riscv.vector.tuple", <vscale x 4 x i8>, 3) undef, ptr %base, i64 %vl, i64 5)
+  %1 = call <vscale x 1 x i32> @llvm.riscv.tuple.extract.nxv1i32.triscv.vector.tuple_nxv4i8_3t(target("riscv.vector.tuple", <vscale x 4 x i8>, 3) %0, i32 1)
+  ret <vscale x 1 x i32> %1
+}
+
+define <vscale x 1 x i32> @test_vlseg3_mask_nxv1i32(ptr %base, i64 %vl, <vscale x 1 x i1> %mask) sanitize_address {
+; CHECK-LABEL: @test_vlseg3_mask_nxv1i32(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i64, ptr @__asan_shadow_memory_dynamic_address, align 8
+; CHECK-NEXT:    [[TMP36:%.*]] = tail call target("riscv.vector.tuple", <vscale x 4 x i8>, 3) @llvm.riscv.vlseg3.mask.triscv.vector.tuple_nxv4i8_3t.p0.nxv1i1.i64(target("riscv.vector.tuple", <vscale x 4 x i8>, 3) undef, ptr [[BASE:%.*]], <vscale x 1 x i1> [[MASK:%.*]], i64 [[VL:%.*]], i64 1, i64 5)
+; CHECK-NEXT:    [[TMP37:%.*]] = call <vscale x 1 x i32> @llvm.riscv.tuple.extract.nxv1i32.triscv.vector.tuple_nxv4i8_3t(target("riscv.vector.tuple", <vscale x 4 x i8>, 3) [[TMP36]], i32 1)
+; CHECK-NEXT:    ret <vscale x 1 x i32> [[TMP37]]
+;
+entry:
+  %0 = tail call target("riscv.vector.tuple", <vscale x 4 x i8>, 3) @llvm.riscv.vlseg3.mask.triscv.vector.tuple_nxv4i8_3t.nxv1i1(target("riscv.vector.tuple", <vscale x 4 x i8>, 3) undef, ptr %base, <vscale x 1 x i1> %mask, i64 %vl, i64 1, i64 5)
+  %1 = call <vscale x 1 x i32> @llvm.riscv.tuple.extract.nxv1i32.triscv.vector.tuple_nxv4i8_3t(target("riscv.vector.tuple", <vscale x 4 x i8>, 3) %0, i32 1)
+  ret <vscale x 1 x i32> %1
+}
+
+
+declare target("riscv.vector.tuple", <vscale x 4 x i8>, 4) @llvm.riscv.vlseg4.triscv.vector.tuple_nxv4i8_4t(target("riscv.vector.tuple", <vscale x 4 x i8>, 4), ptr, i64, i64)
+declare target("riscv.vector.tuple", <vscale x 4 x i8>, 4) @llvm.riscv.vlseg4.mask.triscv.vector.tuple_nxv4i8_4t.nxv1i1(target("riscv.vector.tuple", <vscale x 4 x i8>, 4), ptr, <vscale x 1 x i1>, i64, i64, i64)
+
+define <vscale x 1 x i32> @test_vlseg4_nxv1i32(ptr %base, i64 %vl) sanitize_address {
+; CHECK-LABEL: @test_vlseg4_nxv1i32(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i64, ptr @__asan_shadow_memory_dynamic_address, align 8
+; CHECK-NEXT:    [[TMP48:%.*]] = tail call target("riscv.vector.tuple", <vscale x 4 x i8>, 4) @llvm.riscv.vlseg4.triscv.vector.tuple_nxv4i8_4t.p0.i64(target("riscv.vector.tuple", <vscale x 4 x i8>, 4) undef, ptr [[BASE:%.*]], i64 [[VL:%.*]], i64 5)
+; CHECK-NEXT:    [[TMP49:%.*]] = call <vscale x 1 x i32> @llvm.riscv.tuple.extract.nxv1i32.triscv.vector.tuple_nxv4i8_4t(target("riscv.vector.tuple", <vscale x 4 x i8>, 4) [[TMP48]], i32 1)
+; CHECK-NEXT:    ret <vscale x 1 x i32> [[TMP49]]
+;
+entry:
+  %0 = tail call target("riscv.vector.tuple", <vscale x 4 x i8>, 4) @llvm.riscv.vlseg4.triscv.vector.tuple_nxv4i8_4t(target("riscv.vector.tuple", <vscale x 4 x i8>, 4) undef, ptr %base, i64 %vl, i64 5)
+  %1 = call <vscale x 1 x i32> @llvm.riscv.tuple.extract.nxv1i32.triscv.vector.tuple_nxv4i8_4t(target("riscv.vector.tuple", <vscale x 4 x i8>, 4) %0, i32 1)
+  ret <vscale x 1 x i32> %1
+}
+
+define <vscale x 1 x i32> @test_vlseg4_mask_nxv1i32(ptr %base, i64 %vl, <vscale x 1 x i1> %mask) sanitize_address {
+; CHECK-LABEL: @test_vlseg4_mask_nxv1i32(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i64, ptr @__asan_shadow_memory_dynamic_address, align 8
+; CHECK-NEXT:    [[TMP48:%.*]] = tail call target("riscv.vector.tuple", <vscale x 4 x i8>, 4) @llvm.riscv.vlseg4.mask.triscv.vector.tuple_nxv4i8_4t.p0.nxv1i1.i64(target("riscv.vector.tuple", <vscale x 4 x i8>, 4) undef, ptr [[BASE:%.*]], <vscale x 1 x i1> [[MASK:%.*]], i64 [[VL:%.*]], i64 1, i64 5)
+; CHECK-NEXT:    [[TMP49:%.*]] = call <vscale x 1 x i32> @llvm.riscv.tuple.extract.nxv1i32.triscv.vector.tuple_nxv4i8_4t(target("riscv.vector.tuple", <vscale x 4 x i8>, 4) [[TMP48]], i32 1)
+; CHECK-NEXT:    ret <vscale x 1 x i32> [[TMP49]]
+;
+entry:
+  %0 = tail call target("riscv.vector.tuple", <vscale x 4 x i8>, 4) @llvm.riscv.vlseg4.mask.triscv.vector.tuple_nxv4i8_4t.nxv1i1(target("riscv.vector.tuple", <vscale x 4 x i8>, 4) undef, ptr %base, <vscale x 1 x i1> %mask, i64 %vl, i64 1, i64 5)
+  %1 = call <vscale x 1 x i32> @llvm.riscv.tuple.extract.nxv1i32.triscv.vector.tuple_nxv4i8_4t(target("riscv.vector.tuple", <vscale x 4 x i8>, 4) %0, i32 1)
+  ret <vscale x 1 x i32> %1
+}
+
+
+declare target("riscv.vector.tuple", <vscale x 4 x i8>, 5) @llvm.riscv.vlseg5.triscv.vector.tuple_nxv4i8_5t(target("riscv.vector.tuple", <vscale x 4 x i8>, 5), ptr, i64, i64)
+declare target("riscv.vector.tuple", <vscale x 4 x i8>, 5) @llvm.riscv.vlseg5.mask.triscv.vector.tuple_nxv4i8_5t.nxv1i1(target("riscv.vector.tuple", <vscale x 4 x i8>, 5), ptr, <vscale x 1 x i1>, i64, i64, i64)
+
+define <vscale x 1 x i32> @test_vlseg5_nxv1i32(ptr %base, i64 %vl) sanitize_address {
+; CHECK-LABEL: @test_vlseg5_nxv1i32(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i64, ptr @__asan_shadow_memory_dynamic_address, align 8
+; CHECK-NEXT:    [[TMP60:%.*]] = tail call target("riscv.vector.tuple", <vscale x 4 x i8>, 5) @llvm.riscv.vlseg5.triscv.vector.tuple_nxv4i8_5t.p0.i64(target("riscv.vector.tuple", <vscale x 4 x i8>, 5) undef, ptr [[BASE:%.*]], i64 [[VL:%.*]], i64 5)
+; CHECK-NEXT:    [[TMP61:%.*]] = call <vscale x 1 x i32> @llvm.riscv.tuple.extract.nxv1i32.triscv.vector.tuple_nxv4i8_5t(target("riscv.vector.tuple", <vscale x 4 x i8>, 5) [[TMP60]], i32 1)
+; CHECK-NEXT:    ret <vscale x 1 x i32> [[TMP61]]
+;
+entry:
+  %0 = tail call target("riscv.vector.tuple", <vscale x 4 x i8>, 5) @llvm.riscv.vlseg5.triscv.vector.tuple_nxv4i8_5t(target("riscv.vector.tuple", <vscale x 4 x i8>, 5) undef, ptr %base, i64 %vl, i64 5)
+  %1 = call <vscale x 1 x i32> @llvm.riscv.tuple.extract.nxv1i32.triscv.vector.tuple_nxv4i8_5t(target("riscv.vector.tuple", <vscale x 4 x i8>, 5) %0, i32 1)
+  ret <vscale x 1 x i32> %1
+}
+
+define <vscale x 1 x i32> @test_vlseg5_mask_nxv1i32(ptr %base, i64 %vl, <vscale x 1 x i1> %mask) sanitize_address {
+; CHECK-LABEL: @test_vlseg5_mask_nxv1i32(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i64, ptr @__asan_shadow_memory_dynamic_address, align 8
+; CHECK-NEXT:    [[TMP60:%.*]] = tail call target("riscv.vector.tuple", <vscale x 4 x i8>, 5) @llvm.riscv.vlseg5.mask.triscv.vector.tuple_nxv4i8_5t.p0.nxv1i1.i64(target("riscv.vector.tuple", <vscale x 4 x i8>, 5) undef, ptr [[BASE:%.*]], <vscale x 1 x i1> [[MASK:%.*]], i64 [[VL:%.*]], i64 1, i64 5)
+; CHECK-NEXT:    [[TMP61:%.*]] = call <vscale x 1 x i32> @llvm.riscv.tuple.extract.nxv1i32.triscv.vector.tuple_nxv4i8_5t(target("riscv.vector.tuple", <vscale x 4 x i8>, 5) [[TMP60]], i32 1)
+; CHECK-NEXT:    ret <vscale x 1 x i32> [[TMP61]]
+;
+entry:
+  %0 = tail call target("riscv.vector.tuple", <vscale x 4 x i8>, 5) @llvm.riscv.vlseg5.mask.triscv.vector.tuple_nxv4i8_5t.nxv1i1(target("riscv.vector.tuple", <vscale x 4 x i8>, 5) undef, ptr %base, <vscale x 1 x i1> %mask, i64 %vl, i64 1, i64 5)
+  %1 = call <vscale x 1 x i32> @llvm.riscv.tuple.extract.nxv1i32.triscv.vector.tuple_nxv4i8_5t(target("riscv.vector.tuple", <vscale x 4 x i8>, 5) %0, i32 1)
+  ret <vscale x 1 x i32> %1
+}
+
+
+declare target("riscv.vector.tuple", <vscale x 4 x i8>, 6) @llvm.riscv.vlseg6.triscv.vector.tuple_nxv4i8_6t(target("riscv.vector.tuple", <vscale x 4 x i8>, 6), ptr, i64, i64)
+declare target("riscv.vector.tuple", <vscale x 4 x i8>, 6) @llvm.riscv.vlseg6.mask.triscv.vector.tuple_nxv4i8_6t.nxv1i1(target("riscv.vector.tuple", <vscale x 4 x i8>, 6), ptr, <vscale x 1 x i1>, i64, i64, i64)
+
+define <vscale x 1 x i32> @test_vlseg6_nxv1i32(ptr %base, i64 %vl) sanitize_address {
+; CHECK-LABEL: @test_vlseg6_nxv1i32(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i64, ptr @__asan_shadow_memory_dynamic_address, align 8
+; CHECK-NEXT:    [[TMP72:%.*]] = tail call target("riscv.vector.tuple", <vscale x 4 x i8>, 6) @llvm.riscv.vlseg6.triscv.vector.tuple_nxv4i8_6t.p0.i64(target("riscv.vector.tuple", <vscale x 4 x i8>, 6) undef, ptr [[BASE:%.*]], i64 [[VL:%.*]], i64 5)
+; CHECK-NEXT:    [[TMP73:%.*]] = call <vscale x 1 x i32> @llvm.riscv.tuple.extract.nxv1i32.triscv.vector.tuple_nxv4i8_6t(target("riscv.vector.tuple", <vscale x 4 x i8>, 6) [[TMP72]], i32 1)
+; CHECK-NEXT:    ret <vscale x 1 x i32> [[TMP73]]
+;
+entry:
+  %0 = tail call target("riscv.vector.tuple", <vscale x 4 x i8>, 6) @llvm.riscv.vlseg6.triscv.vector.tuple_nxv4i8_6t(target("riscv.vector.tuple", <vscale x 4 x i8>, 6) undef, ptr %base, i64 %vl, i64 5)
+  %1 = call <vscale x 1 x i32> @llvm.riscv.tuple.extract.nxv1i32.triscv.vector.tuple_nxv4i8_6t(target("riscv.vector.tuple", <vscale x 4 x i8>, 6) %0, i32 1)
+  ret <vscale x 1 x i32> %1
+}
+
+define <vscale x 1 x i32> @test_vlseg6_mask_nxv1i32(ptr %base, i64 %vl, <vscale x 1 x i1> %mask) sanitize_address {
+; CHECK-LABEL: @test_vlseg6_mask_nxv1i32(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i64, ptr @__asan_shadow_memory_dynamic_address, align 8
+; CHECK-NEXT:    [[TMP72:%.*]] = tail call target("riscv.vector.tuple", <vscale x 4 x i8>, 6) @llvm.riscv.vlseg6.mask.triscv.vector.tuple_nxv4i8_6t.p0.nxv1i1.i64(target("riscv.vector.tuple", <vscale x 4 x i8>, 6) undef, ptr [[BASE:%.*]], <vscale x 1 x i1> [[MASK:%.*]], i64 [[VL:%.*]], i64 1, i64 5)
+; CHECK-NEXT:    [[TMP73:%.*]] = call <vscale x 1 x i32> @llvm.riscv.tuple.extract.nxv1i32.triscv.vector.tuple_nxv4i8_6t(target("riscv.vector.tuple", <vscale x 4 x i8>, 6) [[TMP72]], i32 1)
+; CHECK-NEXT:    ret <vscale x 1 x i32> [[TMP73]]
+;
+entry:
+  %0 = tail call target("riscv.vector.tuple", <vscale x 4 x i8>, 6) @llvm.riscv.vlseg6.mask.triscv.vector.tuple_nxv4i8_6t.nxv1i1(target("riscv.vector.tuple", <vscale x 4 x i8>, 6) undef, ptr %base, <vscale x 1 x i1> %mask, i64 %vl, i64 1, i64 5)
+  %1 = call <vscale x 1 x i32> @llvm.riscv.tuple.extract.nxv1i32.triscv.vector.tuple_nxv4i8_6t(target("riscv.vector.tuple", <vscale x 4 x i8>, 6) %0, i32 1)
+  ret <vscale x 1 x i32> %1
+}
+
+
+declare target("riscv.vector.tuple", <vscale x 4 x i8>, 7) @llvm.riscv.vlseg7.triscv.vector.tuple_nxv4i8_7t(target("riscv.vector.tuple", <vscale x 4 x i8>, 7), ptr, i64, i64)
+declare target("riscv.vector.tuple", <vscale x 4 x i8>, 7) @llvm.riscv.vlseg7.mask.triscv.vector.tuple_nxv4i8_7t.nxv1i1(target("riscv.vector.tuple", <vscale x 4 x i8>, 7), ptr, <vscale x 1 x i1>, i64, i64, i64)
+
+define <vscale x 1 x i32> @test_vlseg7_nxv1i32(ptr %base, i64 %vl) sanitize_address {
+; CHECK-LABEL: @test_vlseg7_nxv1i32(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i64, ptr @__asan_shadow_memory_dynamic_address, align 8
+; CHECK-NEXT:    [[TMP84:%.*]] = tail call target("riscv.vector.tuple", <vscale x 4 x i8>, 7) @llvm.riscv.vlseg7.triscv.vector.tuple_nxv4i8_7t.p0.i64(target("riscv.vector.tuple", <vscale x 4 x i8>, 7) undef, ptr [[BASE:%.*]], i64 [[VL:%.*]], i64 5)
+; CHECK-NEXT:    [[TMP85:%.*]] = call <vscale x 1 x i32> @llvm.riscv.tuple.extract.nxv1i32.triscv.vector.tuple_nxv4i8_7t(target("riscv.vector.tuple", <vscale x 4 x i8>, 7) [[TMP84]], i32 1)
+; CHECK-NEXT:    ret <vscale x 1 x i32> [[TMP85]]
+;
+entry:
+  %0 = tail call target("riscv.vector.tuple", <vscale x 4 x i8>, 7) @llvm.riscv.vlseg7.triscv.vector.tuple_nxv4i8_7t(target("riscv.vector.tuple", <vscale x 4 x i8>, 7) undef, ptr %base, i64 %vl, i64 5)
+  %1 = call <vscale x 1 x i32> @llvm.riscv.tuple.extract.nxv1i32.triscv.vector.tuple_nxv4i8_7t(target("riscv.vector.tuple", <vscale x 4 x i8>, 7) %0, i32 1)
+  ret <vscale x 1 x i32> %1
+}
+
+define <vscale x 1 x i32> @test_vlseg7_mask_nxv1i32(ptr %base, i64 %vl, <vscale x 1 x i1> %mask) sanitize_address {
+; CHECK-LABEL: @test_vlseg7_mask_nxv1i32(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i64, ptr @__asan_shadow_memory_dynamic_address, align 8
+; CHECK-NEXT:    [[TMP84:%.*]] = tail call target("riscv.vector.tuple", <vscale x 4 x i8>, 7) @llvm.riscv.vlseg7.mask.triscv.vector.tuple_nxv4i8_7t.p0.nxv1i1.i64(target("riscv.vector.tuple", <vscale x 4 x i8>, 7) undef, ptr [[BASE:%.*]], <vscale x 1 x i1> [[MASK:%.*]], i64 [[VL:%.*]], i64 1, i64 5)
+; CHECK-NEXT:    [[TMP85:%.*]] = call <vscale x 1 x i32> @llvm.riscv.tuple.extract.nxv1i32.triscv.vector.tuple_nxv4i8_7t(target("riscv.vector.tuple", <vscale x 4 x i8>, 7) [[TMP84]], i32 1)
+; CHECK-NEXT:    ret <vscale x 1 x i32> [[TMP85]]
+;
+entry:
+  %0 = tail call target("riscv.vector.tuple", <vscale x 4 x i8>, 7) @llvm.riscv.vlseg7.mask.triscv.vector.tuple_nxv4i8_7t.nxv1i1(target("riscv.vector.tuple", <vscale x 4 x i8>, 7) undef, ptr %base, <vscale x 1 x i1> %mask, i64 %vl, i64 1, i64 5)
+  %1 = call <vscale x 1 x i32> @llvm.riscv.tuple.extract.nxv1i32.triscv.vector.tuple_nxv4i8_7t(target("riscv.vector.tuple", <vscale x 4 x i8>, 7) %0, i32 1)
+  ret <vscale x 1 x i32> %1
+}
+
+
+declare target("riscv.vector.tuple", <vscale x 4 x i8>, 8) @llvm.riscv.vlseg8.triscv.vector.tuple_nxv4i8_8t(target("riscv.vector.tuple", <vscale x 4 x i8>, 8), ptr, i64, i64)
+declare target("riscv.vector.tuple", <vscale x 4 x i8>, 8) @llvm.riscv.vlseg8.mask.triscv.vector.tuple_nxv4i8_8t.nxv1i1(target("riscv.vector.tuple", <vscale x 4 x i8>, 8), ptr, <vscale x 1 x i1>, i64, i64, i64)
+
+define <vscale x 1 x i32> @test_vlseg8_nxv1i32(ptr %base, i64 %vl) sanitize_address {
+; CHECK-LABEL: @test_vlseg8_nxv1i32(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i64, ptr @__asan_shadow_memory_dynamic_address, align 8
+; CHECK-NEXT:    [[TMP96:%.*]] = tail call target("riscv.vector.tuple", <vscale x 4 x i8>, 8) @llvm.riscv.vlseg8.triscv.vector.tuple_nxv4i8_8t.p0.i64(target("riscv.vector.tuple", <vscale x 4 x i8>, 8) undef, ptr [[BASE:%.*]], i64 [[VL:%.*]], i64 5)
+; CHECK-NEXT:    [[TMP97:%.*]] = call <vscale x 1 x i32> @llvm.riscv.tuple.extract.nxv1i32.triscv.vector.tuple_nxv4i8_8t(target("riscv.vector.tuple", <vscale x 4 x i8>, 8) [[TMP96]], i32 1)
+; CHECK-NEXT:    ret <vscale x 1 x i32> [[TMP97]]
+;
+entry:
+  %0 = tail call target("riscv.vector.tuple", <vscale x 4 x i8>, 8) @llvm.riscv.vlseg8.triscv.vector.tuple_nxv4i8_8t(target("riscv.vector.tuple", <vscale x 4 x i8>, 8) undef, ptr %base, i64 %vl, i64 5)
+  %1 = call <vscale x 1 x i32> @llvm.riscv.tuple.extract.nxv1i32.triscv.vector.tuple_nxv4i8_8t(target("riscv.vector.tuple", <vscale x 4 x i8>, 8) %0, i32 1)
+  ret <vscale x 1 x i32> %1
+}
+
+define <vscale x 1 x i32> @test_vlseg8_mask_nxv1i32(ptr %base, i64 %vl, <vscale x 1 x i1> %mask) sanitize_address {
+; CHECK-LABEL: @test_vlseg8_mask_nxv1i32(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i64, ptr @__asan_shadow_memory_dynamic_address, align 8
+; CHECK-NEXT:    [[TMP96:%.*]] = tail call target("riscv.vector.tuple", <vscale x 4 x i8>, 8) @llvm.riscv.vlseg8.mask.triscv.vector.tuple_nxv4i8_8t.p0.nxv1i1.i64(target("riscv.vector.tuple", <vscale x 4 x i8>, 8) undef, ptr [[BASE:%.*]], <vscale x 1 x i1> [[MASK:%.*]], i64 [[VL:%.*]], i64 1, i64 5)
+; CHECK-NEXT:    [[TMP97:%.*]] = call <vscale x 1 x i32> @llvm.riscv.tuple.extract.nxv1i32.triscv.vector.tuple_nxv4i8_8t(target("riscv.vector.tuple", <vscale x 4 x i8>, 8) [[TMP96]], i32 1)
+; CHECK-NEXT:    ret <vscale x 1 x i32> [[TMP97]]
+;
+entry:
+  %0 = tail call target("riscv.vector.tuple", <vscale x 4 x i8>, 8) @llvm.riscv.vlseg8.mask.triscv.vector.tuple_nxv4i8_8t.nxv1i1(target("riscv.vector.tuple", <vscale x 4 x i8>, 8) undef, ptr %base, <vscale x 1 x i1> %mask, i64 %vl, i64 1, i64 5)
+  %1 = call <vscale x 1 x i32> @llvm.riscv.tuple.extract.nxv1i32.triscv.vector.tuple_nxv4i8_8t(target("riscv.vector.tuple", <vscale x 4 x i8>, 8) %0, i32 1)
+  ret <vscale x 1 x i32> %1
+}
+
+
+declare void @llvm.riscv.vsseg2.triscv.vector.tuple_nxv4i8_2t(target("riscv.vector.tuple", <vscale x 4 x i8>, 2), ptr, i64, i64)
+declare void @llvm.riscv.vsseg2.mask.triscv.vector.tuple_nxv4i8_2t.nxv1i1(target("riscv.vector.tuple", <vscale x 4 x i8>, 2), ptr, <vscale x 1 x i1>, i64, i64)
+
+define void @test_vsseg2_nxv1i32(target("riscv.vector.tuple", <vscale x 4 x i8>, 2) %val, ptr %base, i64 %vl) sanitize_address {
+; CHECK-LABEL: @test_vsseg2_nxv1i32(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i64, ptr @__asan_shadow_memory_dynamic_address, align 8
+; CHECK-NEXT:    tail call void @llvm.riscv.vsseg2.triscv.vector.tuple_nxv4i8_2t.p0.i64(target("riscv.vector.tuple", <vscale x 4 x i8>, 2) [[VAL:%.*]], ptr [[BASE:%.*]], i64 [[VL:%.*]], i64 5)
+; CHECK-NEXT:    ret void
+;
+entry:
+  tail call void @llvm.riscv.vsseg2.triscv.vector.tuple_nxv4i8_2t(target("riscv.vector.tuple", <vscale x 4 x i8>, 2) %val, ptr %base, i64 %vl, i64 5)
+  ret void
+}
+
+define void @test_vsseg2_mask_nxv1i32(target("riscv.vector.tuple", <vscale x 4 x i8>, 2) %val, ptr %base, i64 %vl, <vscale x 1 x i1> %mask) sanitize_address {
+; CHECK-LABEL: @test_vsseg2_mask_nxv1i32(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i64, ptr @__asan_shadow_memory_dynamic_address, align 8
+; CHECK-NEXT:    tail call void @llvm.riscv.vsseg2.mask.triscv.vector.tuple_nxv4i8_2t.p0.nxv1i1.i64(target("riscv.vector.tuple", <vscale x 4 x i8>, 2) [[VAL:%.*]], ptr [[BASE:%.*]], <vscale x 1 x i1> [[MASK:%.*]], i64 [[VL:%.*]], i64 5)
+; CHECK-NEXT:    ret void
+;
+entry:
+  tail call void @llvm.riscv.vsseg2.mask.triscv.vector.tuple_nxv4i8_2t.nxv1i1(target("riscv.vector.tuple", <vscale x 4 x i8>, 2) %val, ptr %base, <vscale x 1 x i1> %mask, i64 %vl, i64 5)
+  ret void
+}
+
+
+declare void @llvm.riscv.vsseg3.triscv.vector.tuple_nxv4i8_3t(target("riscv.vector.tuple", <vscale x 4 x i8>, 3), ptr, i64, i64)
+declare void @llvm.riscv.vsseg3.mask.triscv.vector.tuple_nxv4i8_3t.nxv1i1(target("riscv.vector.tuple", <vscale x 4 x i8>, 3), ptr, <vscale x 1 x i1>, i64, i64)
+
+define void @test_vsseg3_nxv1i32(target("riscv.vector.tuple", <vscale x 4 x i8>, 3) %val, ptr %base, i64 %vl) sanitize_address {
+; CHECK-LABEL: @test_vsseg3_nxv1i32(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i64, ptr @__asan_shadow_memory_dynamic_address, align 8
+; CHECK-NEXT:    tail call void @llvm.riscv.vsseg3.triscv.vector.tuple_nxv4i8_3t.p0.i64(target("riscv.vector.tuple", <vscale x 4 x i8>, 3) [[VAL:%.*]], ptr [[BASE:%.*]], i64 [[VL:%.*]], i64 5)
+; CHECK-NEXT:    ret void
+;
+entry:
+  tail call void @llvm.riscv.vsseg3.triscv.vector.tuple_nxv4i8_3t(target("riscv.vector.tuple", <vscale x 4 x i8>, 3) %val, ptr %base, i64 %vl, i64 5)
+  ret void
+}
+
+define void @test_vsseg3_mask_nxv1i32(target("riscv.vector.tuple", <vscale x 4 x i8>, 3) %val, ptr %base, i64 %vl, <vscale x 1 x i1> %mask) sanitize_address {
+; CHECK-LABEL: @test_vsseg3_mask_nxv1i32(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i64, ptr @__asan_shadow_memory_dynamic_address, align 8
+; CHECK-NEXT:    tail call void @llvm.riscv.vsseg3.mask.triscv.vector.tuple_nxv4i8_3t.p0.nxv1i1.i64(target("riscv.vector.tuple", <vscale x 4 x i8>, 3) [[VAL:%.*]], ptr [[BASE:%.*]], <vscale x 1 x i1> [[MASK:%.*]], i64 [[VL:%.*]], i64 5)
+; CHECK-NEXT:    ret void
+;
+entry:
+  tail call void @llvm.riscv.vsseg3.mask.triscv.vector.tuple_nxv4i8_3t.nxv1i1(target("riscv.vector.tuple", <vscale x 4 x i8>, 3) %val, ptr %base, <vscale x 1 x i1> %mask, i64 %vl, i64 5)
+  ret void
+}
+
+
+declare void @llvm.riscv.vsseg4.triscv.vector.tuple_nxv4i8_4t(target("riscv.vector.tuple", <vscale x 4 x i8>, 4), ptr, i64, i64)
+declare void @llvm.riscv.vsseg4.mask.triscv.vector.tuple_nxv4i8_4t.nxv1i1(target("riscv.vector.tuple", <vscale x 4 x i8>, 4), ptr, <vscale x 1 x i1>, i64, i64)
+
+define void @test_vsseg4_nxv1i32(target("riscv.vector.tuple", <vscale x 4 x i8>, 4) %val, ptr %base, i64 %vl) sanitize_address {
+; CHECK-LABEL: @test_vsseg4_nxv1i32(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i64, ptr @__asan_shadow_memory_dynamic_address, align 8
+; CHECK-NEXT:    tail call void @llvm.riscv.vsseg4.triscv.vector.tuple_nxv4i8_4t.p0.i64(target("riscv.vector.tuple", <vscale x 4 x i8>, 4) [[VAL:%.*]], ptr [[BASE:%.*]], i64 [[VL:%.*]], i64 5)
+; CHECK-NEXT:    ret void
+;
+entry:
+  tail call void @llvm.riscv.vsseg4.triscv.vector.tuple_nxv4i8_4t(target("riscv.vector.tuple", <vscale x 4 x i8>, 4) %val, ptr %base, i64 %vl, i64 5)
+  ret void
+}
+
+define void @test_vsseg4_mask_nxv1i32(target("riscv.vector.tuple", <vscale x 4 x i8>, 4) %val, ptr %base, i64 %vl, <vscale x 1 x i1> %mask) sanitize_address {
+; CHECK-LABEL: @test_vsseg4_mask_nxv1i32(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i64, ptr @__asan_shadow_memory_dynamic_address, align 8
+; CHECK-NEXT:    tail call void @llvm.riscv.vsseg4.mask.triscv.vector.tuple_nxv4i8_4t.p0.nxv1i1.i64(target("riscv.vector.tuple", <vscale x 4 x i8>, 4) [[VAL:%.*]], ptr [[BASE:%.*]], <vscale x 1 x i1> [[MASK:%.*]], i64 [[VL:%.*]], i64 5)
+; CHECK-NEXT:    ret void
+;
+entry:
+  tail call void @llvm.riscv.vsseg4.mask.triscv.vector.tuple_nxv4i8_4t.nxv1i1(target("riscv.vector.tuple", <vscale x 4 x i8>, 4) %val, ptr %base, <vscale x 1 x i1> %mask, i64 %vl, i64 5)
+  ret void
+}
+
+
+declare void @llvm.riscv.vsseg5.triscv.vector.tuple_nxv4i8_5t(target("riscv.vector.tuple", <vscale x 4 x i8>, 5), ptr, i64, i64)
+declare void @llvm.riscv.vsseg5.mask.triscv.vector.tuple_nxv4i8_5t.nxv1i1(target("riscv.vector.tuple", <vscale x 4 x i8>, 5), ptr, <vscale x 1 x i1>, i64, i64)
+
+define void @test_vsseg5_nxv1i32(target("riscv.vector.tuple", <vscale x 4 x i8>, 5) %val, ptr %base, i64 %vl) sanitize_address {
+; CHECK-LABEL: @test_vsseg5_nxv1i32(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i64, ptr @__asan_shadow_memory_dynamic_address, align 8
+; CHECK-NEXT:    tail call void @llvm.riscv.vsseg5.triscv.vector.tuple_nxv4i8_5t.p0.i64(target("riscv.vector.tuple", <vscale x 4 x i8>, 5) [[VAL:%.*]], ptr [[BASE:%.*]], i64 [[VL:%.*]], i64 5)
+; CHECK-NEXT:    ret void
+;
+entry:
+  tail call void @llvm.riscv.vsseg5.triscv.vector.tuple_nxv4i8_5t(target("riscv.vector.tuple", <vscale x 4 x i8>, 5) %val, ptr %base, i64 %vl, i64 5)
+  ret void
+}
+
+define void @test_vsseg5_mask_nxv1i32(target("riscv.vector.tuple", <vscale x 4 x i8>, 5) %val, ptr %base, i64 %vl, <vscale x 1 x i1> %mask) sanitize_address {
+; CHECK-LABEL: @test_vsseg5_mask_nxv1i32(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i64, ptr @__asan_shadow_memory_dynamic_address, align 8
+; CHECK-NEXT:    tail call void @llvm.riscv.vsseg5.mask.triscv.vector.tuple_nxv4i8_5t.p0.nxv1i1.i64(target("riscv.vector.tuple", <vscale x 4 x i8>, 5) [[VAL:%.*]], ptr [[BASE:%.*]], <vscale x 1 x i1> [[MASK:%.*]], i64 [[VL:%.*]], i64 5)
+; CHECK-NEXT:    ret void
+;
+entry:
+  tail call void @llvm.riscv.vsseg5.mask.triscv.vector.tuple_nxv4i8_5t.nxv1i1(target("riscv.vector.tuple", <vscale x 4 x i8>, 5) %val, ptr %base, <vscale x 1 x i1> %mask, i64 %vl, i64 5)
+  ret void
+}
+
+
+declare void @llvm.riscv.vsseg6.triscv.vector.tuple_nxv4i8_6t(target("riscv.vector.tuple", <vscale x 4 x i8>, 6), ptr, i64, i64)
+declare void @llvm.riscv.vsseg6.mask.triscv.vector.tuple_nxv4i8_6t.nxv1i1(target("riscv.vector.tuple", <vscale x 4 x i8>, 6), ptr, <vscale x 1 x i1>, i64, i64)
+
+define void @test_vsseg6_nxv1i32(target("riscv.vector.tuple", <vscale x 4 x i8>, 6) %val, ptr %base, i64 %vl) sanitize_address {
+; CHECK-LABEL: @test_vsseg6_nxv1i32(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i64, ptr @__asan_shadow_memory_dynamic_address, align 8
+; CHECK-NEXT:    tail call void @llvm.riscv.vsseg6.triscv.vector.tuple_nxv4i8_6t.p0.i64(target("riscv.vector.tuple", <vscale x 4 x i8>, 6) [[VAL:%.*]], ptr [[BASE:%.*]], i64 [[VL:%.*]], i64 5)
+; CHECK-NEXT:    ret void
+;
+entry:
+  tail call void @llvm.riscv.vsseg6.triscv.vector.tuple_nxv4i8_6t(target("riscv.vector.tuple", <vscale x 4 x i8>, 6) %val, ptr %base, i64 %vl, i64 5)
+  ret void
+}
+
+define void @test_vsseg6_mask_nxv1i32(target("riscv.vector.tuple", <vscale x 4 x i8>, 6) %val, ptr %base, i64 %vl, <vscale x 1 x i1> %mask) sanitize_address {
+; CHECK-LABEL: @test_vsseg6_mask_nxv1i32(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i64, ptr @__asan_shadow_memory_dynamic_address, align 8
+; CHECK-NEXT:    tail call void @llvm.riscv.vsseg6.mask.triscv.vector.tuple_nxv4i8_6t.p0.nxv1i1.i64(target("riscv.vector.tuple", <vscale x 4 x i8>, 6) [[VAL:%.*]], ptr [[BASE:%.*]], <vscale x 1 x i1> [[MASK:%.*]], i64 [[VL:%.*]], i64 5)
+; CHECK-NEXT:    ret void
+;
+entry:
+  tail call void @llvm.riscv.vsseg6.mask.triscv.vector.tuple_nxv4i8_6t.nxv1i1(target("riscv.vector.tuple", <vscale x 4 x i8>, 6) %val, ptr %base, <vscale x 1 x i1> %mask, i64 %vl, i64 5)
+  ret void
+}
+
+
+declare void @llvm.riscv.vsseg7.triscv.vector.tuple_nxv4i8_7t(target("riscv.vector.tuple", <vscale x 4 x i8>, 7), ptr, i64, i64)
+declare void @llvm.riscv.vsseg7.mask.triscv.vector.tuple_nxv4i8_7t.nxv1i1(target("riscv.vector.tuple", <vscale x 4 x i8>, 7), ptr, <vscale x 1 x i1>, i64, i64)
+
+define void @test_vsseg7_nxv1i32(target("riscv.vector.tuple", <vscale x 4 x i8>, 7) %val, ptr %base, i64 %vl) sanitize_address {
+; CHECK-LABEL: @test_vsseg7_nxv1i32(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i64, ptr @__asan_shadow_memory_dynamic_address, align 8
+; CHECK-NEXT:    tail call void @llvm.riscv.vsseg7.triscv.vector.tuple_nxv4i8_7t.p0.i64(target("riscv.vector.tuple", <vscale x 4 x i8>, 7) [[VAL:%.*]], ptr [[BASE:%.*]], i64 [[VL:%.*]], i64 5)
+; CHECK-NEXT:    ret void
+;
+entry:
+  tail call void @llvm.riscv.vsseg7.triscv.vector.tuple_nxv4i8_7t(target("riscv.vector.tuple", <vscale x 4 x i8>, 7) %val, ptr %base, i64 %vl, i64 5)
+  ret void
+}
+
+define void @test_vsseg7_mask_nxv1i32(target("riscv.vector.tuple", <vscale x 4 x i8>, 7) %val, ptr %base, i64 %vl, <vscale x 1 x i1> %mask) sanitize_address {
+; CHECK-LABEL: @test_vsseg7_mask_nxv1i32(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i64, ptr @__asan_shadow_memory_dynamic_address, align 8
+; CHECK-NEXT:    tail call void @llvm.riscv.vsseg7.mask.triscv.vector.tuple_nxv4i8_7t.p0.nxv1i1.i64(target("riscv.vector.tuple", <vscale x 4 x i8>, 7) [[VAL:%.*]], ptr [[BASE:%.*]], <vscale x 1 x i1> [[MASK:%.*]], i64 [[VL:%.*]], i64 5)
+; CHECK-NEXT:    ret void
+;
+entry:
+  tail call void @llvm.riscv.vsseg7.mask.triscv.vector.tuple_nxv4i8_7t.nxv1i1(target("riscv.vector.tuple", <vscale x 4 x i8>, 7) %val, ptr %base, <vscale x 1 x i1> %mask, i64 %vl, i64 5)
+  ret void
+}
+
+
+declare void @llvm.riscv.vsseg8.triscv.vector.tuple_nxv4i8_8t(target("riscv.vector.tuple", <vscale x 4 x i8>, 8), ptr, i64, i64)
+declare void @llvm.riscv.vsseg8.mask.triscv.vector.tuple_nxv4i8_8t.nxv1i1(target("riscv.vector.tuple", <vscale x 4 x i8>, 8), ptr, <vscale x 1 x i1>, i64, i64)
+
+define void @test_vsseg8_nxv1i32(target("riscv.vector.tuple", <vscale x 4 x i8>, 8) %val, ptr %base, i64 %vl) sanitize_address {
+; CHECK-LABEL: @test_vsseg8_nxv1i32(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i64, ptr @__asan_shadow_memory_dynamic_address, align 8
+; CHECK-NEXT:    tail call void @llvm.riscv.vsseg8.triscv.vector.tuple_nxv4i8_8t.p0.i64(target("riscv.vector.tuple", <vscale x 4 x i8>, 8) [[VAL:%.*]], ptr [[BASE:%.*]], i64 [[VL:%.*]], i64 5)
+; CHECK-NEXT:    ret void
+;
+entry:
+  tail call void @llvm.riscv.vsseg8.triscv.vector.tuple_nxv4i8_8t(target("riscv.vector.tuple", <vscale x 4 x i8>, 8) %val, ptr %base, i64 %vl, i64 5)
+  ret void
+}
+
+define void @test_vsseg8_mask_nxv1i32(target("riscv.vector.tuple", <vscale x 4 x i8>, 8) %val, ptr %base, i64 %vl, <vscale x 1 x i1> %mask) sanitize_address {
+; CHECK-LABEL: @test_vsseg8_mask_nxv1i32(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i64, ptr @__asan_shadow_memory_dynamic_address, align 8
+; CHECK-NEXT:    tail call void @llvm.riscv.vsseg8.mask.triscv.vector.tuple_nxv4i8_8t.p0.nxv1i1.i64(target("riscv.vector.tuple", <vscale x 4 x i8>, 8) [[VAL:%.*]], ptr [[BASE:%.*]], <vscale x 1 x i1> [[MASK:%.*]], i64 [[VL:%.*]], i64 5)
+; CHECK-NEXT:    ret void
+;
+entry:
+  tail call void @llvm.riscv.vsseg8.mask.triscv.vector.tuple_nxv4i8_8t.nxv1i1(target("riscv.vector.tuple", <vscale x 4 x i8>, 8) %val, ptr %base, <vscale x 1 x i1> %mask, i64 %vl, i64 5)
+  ret void
+}
+
+
+; Test stride load
+declare <vscale x 1 x i32> @llvm.riscv.vlse.nxv1i32(
+  <vscale x 1 x i32>,
+  <vscale x 1 x i32>*,
+  i64,
+  i64);
+
+define <vscale x 1 x i32> @intrinsic_vlse_v_nxv1i32_nxv1i32(<vscale x 1 x i32>* align 4 %0, i64 %1, i64 %2) sanitize_address {
+; CHECK-LABEL: @intrinsic_vlse_v_nxv1i32_nxv1i32(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP3:%.*]] = load i64, ptr @__asan_shadow_memory_dynamic_address, align 8
+; CHECK-NEXT:    [[TMP4:%.*]] = icmp ne i64 [[TMP2:%.*]], 0
+; CHECK-NEXT:    br i1 [[TMP4]], label [[TMP5:%.*]], label [[TMP14:%.*]]
+; CHECK:       5:
+; CHECK-NEXT:    [[TMP6:%.*]] = call i64 @llvm.vscale.i64()
+; CHECK-NEXT:    [[TMP7:%.*]] = call i64 @llvm.umin.i64(i64 [[TMP2]], i64 [[TMP6]])
+; CHECK-NEXT:    br label [[DOTSPLIT:%.*]]
+; CHECK:       .split:
+; CHECK-NEXT:    [[IV:%.*]] = phi i64 [ 0, [[TMP5]] ], [ [[IV_NEXT:%.*]], [[TMP13:%.*]] ]
+; CHECK-NEXT:    [[TMP8:%.*]] = extractelement <vscale x 1 x i1> splat (i1 true), i64 [[IV]]
+; CHECK-NEXT:    br i1 [[TMP8]], label [[TMP9:%.*]], label [[TMP13]]
+; CHECK:       9:
+; CHECK-NEXT:    [[TMP10:%.*]] = mul i64 [[IV]], [[TMP1:%.*]]
+; CHECK-NEXT:    [[TMP11:%.*]] = getelementptr i8, ptr [[TMP0:%.*]], i64 [[TMP10]]
+; CHECK-NEXT:    [[TMP12:%.*]] = ptrtoint ptr [[TMP11]] to i64
+; CHECK-NEXT:    call void @__asan_loadN(i64 [[TMP12]], i64 4)
+; CHECK-NEXT:    br label [[TMP13]]
+; CHECK:       13:
+; CHECK-NEXT:    [[IV_NEXT]] = add nuw nsw i64 [[IV]], 1
+; CHECK-NEXT:    [[IV_CHECK:%.*]] = icmp eq i64 [[IV_NEXT]], [[TMP7]]
+; CHECK-NEXT:    br i1 [[IV_CHECK]], label [[DOTSPLIT_SPLIT:%.*]], label [[DOTSPLIT]]
+; CHECK:       .split.split:
+; CHECK-NEXT:    br label [[TMP14]]
+; CHECK:       14:
+; CHECK-NEXT:    [[A:%.*]] = call <vscale x 1 x i32> @llvm.riscv.vlse.nxv1i32.p0.i64(<vscale x 1 x i32> undef, ptr [[TMP0]], i64 [[TMP1]], i64 [[TMP2]])
+; CHECK-NEXT:    ret <vscale x 1 x i32> [[A]]
+;
+entry:
+  %a = call <vscale x 1 x i32> @llvm.riscv.vlse.nxv1i32(
+  <vscale x 1 x i32> undef,
+  <vscale x 1 x i32>* %0,
+  i64 %1,
+  i64 %2)
+
+  ret <vscale x 1 x i32> %a
+}
+
+declare <vscale x 1 x i32> @llvm.riscv.vlse.mask.nxv1i32(
+  <vscale x 1 x i32>,
+  <vscale x 1 x i32>*,
+  i64,
+  <vscale x 1 x i1>,
+  i64,
+  i64);
+
+define <vscale x 1 x i32> @intrinsic_vlse_mask_v_nxv1i32_nxv1i32(<vscale x 1 x i32> %0, <vscale x 1 x i32>* %1, i64 %2, <vscale x 1 x i1> %3, i64 %4) sanitize_address {
+; CHECK-LABEL: @intrinsic_vlse_mask_v_nxv1i32_nxv1i32(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP5:%.*]] = load i64, ptr @__asan_shadow_memory_dynamic_address, align 8
+; CHECK-NEXT:    [[TMP6:%.*]] = icmp ne i64 [[TMP4:%.*]], 0
+; CHECK-NEXT:    br i1 [[TMP6]], label [[TMP7:%.*]], label [[TMP16:%.*]]
+; CHECK:       7:
+; CHECK-NEXT:    [[TMP8:%.*]] = call i64 @llvm.vscale.i64()
+; CHECK-NEXT:    [[TMP9:%.*]] = call i64 @llvm.umin.i64(i64 [[TMP4]], i64 [[TMP8]])
+; CHECK-NEXT:    br label [[DOTSPLIT:%.*]]
+; CHECK:       .split:
+; CHECK-NEXT:    [[IV:%.*]] = phi i64 [ 0, [[TMP7]] ], [ [[IV_NEXT:%.*]], [[TMP15:%.*]] ]
+; CHECK-NEXT:    [[TMP10:%.*]] = extractelement <vscale x 1 x i1> [[TMP3:%.*]], i64 [[IV]]
+; CHECK-NEXT:    br i1 [[TMP10]], label [[TMP11:%.*]], label [[TMP15]]
+; CHECK:       11:
+; CHECK-NEXT:    [[TMP12:%.*]] = mul i64 [[IV]], [[TMP2:%.*]]
+; CHECK-NEXT:    [[TMP13:%.*]] = getelementptr i8, ptr [[TMP1:%.*]], i64 [[TMP12]]
+; CHECK-NEXT:    [[TMP14:%.*]] = ptrtoint ptr [[TMP13]] to i64
+; CHECK-NEXT:    call void @__asan_loadN(i64 [[TMP14]], i64 4)
+; CHECK-NEXT:    br label [[TMP15]]
+; CHECK:       15:
+; CHECK-NEXT:    [[IV_NEXT]] = add nuw nsw i64 [[IV]], 1
+; CHECK-NEXT:    [[IV_CHECK:%.*]] = icmp eq i64 [[IV_NEXT]], [[TMP9]]
+; CHECK-NEXT:    br i1 [[IV_CHECK]], label [[DOTSPLIT_SPLIT:%.*]], label [[DOTSPLIT]]
+; CHECK:       .split.split:
+; CHECK-NEXT:    br label [[TMP16]]
+; CHECK:       16:
+; CHECK-NEXT:    [[A:%.*]] = call <vscale x 1 x i32> @llvm.riscv.vlse.mask.nxv1i32.p0.i64(<vscale x 1 x i32> [[TMP0:%.*]], ptr [[TMP1]], i64 [[TMP2]], <vscale x 1 x i1> [[TMP3]], i64 [[TMP4]], i64 1)
+; CHECK-NEXT:    ret <vscale x 1 x i32> [[A]]
+;
+entry:
+  %a = call <vscale x 1 x i32> @llvm.riscv.vlse.mask.nxv1i32(
+  <vscale x 1 x i32> %0,
+  <vscale x 1 x i32>* %1,
+  i64 %2,
+  <vscale x 1 x i1> %3,
+  i64 %4, i64 1)
+
+  ret <vscale x 1 x i32> %a
+}
+
+; Test stride store
+declare void @llvm.riscv.vsse.nxv1i32(
+  <vscale x 1 x i32>,
+  <vscale x 1 x i32>*,
+  i64,
+  i64);
+
+define void @intrinsic_vsse_v_nxv1i32_nxv1i32(<vscale x 1 x i32> %0, <vscale x 1 x i32>* %1, i64 %2, i64 %3) sanitize_address {
+; CHECK-LABEL: @intrinsic_vsse_v_nxv1i32_nxv1i32(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP4:%.*]] = load i64, ptr @__asan_shadow_memory_dynamic_address, align 8
+; CHECK-NEXT:    [[TMP5:%.*]] = icmp ne i64 [[TMP3:%.*]], 0
+; CHECK-NEXT:    br i1 [[TMP5]], label [[TMP6:%.*]], label [[TMP15:%.*]]
+; CHECK:       6:
+; CHECK-NEXT:    [[TMP7:%.*]] = call i64 @llvm.vscale.i64()
+; CHECK-NEXT:    [[TMP8:%.*]] = call i64 @llvm.umin.i64(i64 [[TMP3]], i64 [[TMP7]])
+; CHECK-NEXT:    br label [[DOTSPLIT:%.*]]
+; CHECK:       .split:
+; CHECK-NEXT:    [[IV:%.*]] = phi i64 [ 0, [[TMP6]] ], [ [[IV_NEXT:%.*]], [[TMP14:%.*]] ]
+; CHECK-NEXT:    [[TMP9:%.*]] = extractelement <vscale x 1 x i1> splat (i1 true), i64 [[IV]]
+; CHECK-NEXT:    br i1 [[TMP9]], label [[TMP10:%.*]], label [[TMP14]]
+; CHECK:       10:
+; CHECK-NEXT:    [[TMP11:%.*]] = mul i64 [[IV]], [[TMP2:%.*]]
+; CHECK-NEXT:    [[TMP12:%.*]] = getelementptr i8, ptr [[TMP1:%.*]], i64 [[TMP11]]
+; CHECK-NEXT:    [[TMP13:%.*]] = ptrtoint ptr [[TMP12]] to i64
+; CHECK-NEXT:    call void @__asan_storeN(i64 [[TMP13]], i64 4)
+; CHECK-NEXT:    br label [[TMP14]]
+; CHECK:       14:
+; CHECK-NEXT:    [[IV_NEXT]] = add nuw nsw i64 [[IV]], 1
+; CHECK-NEXT:    [[IV_CHECK:%.*]] = icmp eq i64 [[IV_NEXT]], [[TMP8]]
+; CHECK-NEXT:    br i1 [[IV_CHECK]], label [[DOTSPLIT_SPLIT:%.*]], label [[DOTSPLIT]]
+; CHECK:       .split.split:
+; CHECK-NEXT:    br label [[TMP15]]
+; CHECK:       15:
+; CHECK-NEXT:    call void @llvm.riscv.vsse.nxv1i32.p0.i64(<vscale x 1 x i32> [[TMP0:%.*]], ptr [[TMP1]], i64 [[TMP2]], i64 [[TMP3]])
+; CHECK-NEXT:    ret void
+;
+entry:
+  call void @llvm.riscv.vsse.nxv1i32(
+  <vscale x 1 x i32> %0,
+  <vscale x 1 x i32>* %1,
+  i64 %2,
+  i64 %3)
+
+  ret void
+}
+
+declare void @llvm.riscv.vsse.mask.nxv1i32(
+  <vscale x 1 x i32>,
+  <vscale x 1 x i32>*,
+  i64,
+  <vscale x 1 x i1>,
+  i64);
+
+define void @intrinsic_vsse_mask_v_nxv1i32_nxv1i32(<vscale x 1 x i32> %0, <vscale x 1 x i32>* %1, i64 %2, <vscale x 1 x i1> %3, i64 %4) sanitize_address {
+; CHECK-LABEL: @intrinsic_vsse_mask_v_nxv1i32_nxv1i32(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP5:%.*]] = load i64, ptr @__asan_shadow_memory_dynamic_address, align 8
+; CHECK-NEXT:    [[TMP6:%.*]] = icmp ne i64 [[TMP4:%.*]], 0
+; CHECK-NEXT:    br i1 [[TMP6]], label [[TMP7:%.*]], label [[TMP16:%.*]]
+; CHECK:       7:
+; CHECK-NEXT:    [[TMP8:%.*]] = call i64 @llvm.vscale.i64()
+; CHECK-NEXT:    [[TMP9:%.*]] = call i64 @llvm.umin.i64(i64 [[TMP4]], i64 [[TMP8]])
+; CHECK-NEXT:    br label [[DOTSPLIT:%.*]]
+; CHECK:       .split:
+; CHECK-NEXT:    [[IV:%.*]] = phi i64 [ 0, [[TMP7]] ], [ [[IV_NEXT:%.*]], [[TMP15:%.*]] ]
+; CHECK-NEXT:    [[TMP10:%.*]] = extractelement <vscale x 1 x i1> [[TMP3:%.*]], i64 [[IV]]
+; CHECK-NEXT:    br i1 [[TMP10]], label [[TMP11:%.*]], label [[TMP15]]
+; CHECK:       11:
+; CHECK-NEXT:    [[TMP12:%.*]] = mul i64 [[IV]], [[TMP2:%.*]]
+; CHECK-NEXT:    [[TMP13:%.*]] = getelementptr i8, ptr [[TMP1:%.*]], i64 [[TMP12]]
+; CHECK-NEXT:    [[TMP14:%.*]] = ptrtoint ptr [[TMP13]] to i64
+; CHECK-NEXT:    call void @__asan_storeN(i64 [[TMP14]], i64 4)
+; CHECK-NEXT:    br label [[TMP15]]
+; CHECK:       15:
+; CHECK-NEXT:    [[IV_NEXT]] = add nuw nsw i64 [[IV]], 1
+; CHECK-NEXT:    [[IV_CHECK:%.*]] = icmp eq i64 [[IV_NEXT]], [[TMP9]]
+; CHECK-NEXT:    br i1 [[IV_CHECK]], label [[DOTSPLIT_SPLIT:%.*]], label [[DOTSPLIT]]
+; CHECK:       .split.split:
+; CHECK-NEXT:    br label [[TMP16]]
+; CHECK:       16:
+; CHECK-NEXT:    call void @llvm.riscv.vsse.mask.nxv1i32.p0.i64(<vscale x 1 x i32> [[TMP0:%.*]], ptr [[TMP1]], i64 [[TMP2]], <vscale x 1 x i1> [[TMP3]], i64 [[TMP4]])
+; CHECK-NEXT:    ret void
+;
+entry:
+  call void @llvm.riscv.vsse.mask.nxv1i32(
+  <vscale x 1 x i32> %0,
+  <vscale x 1 x i32>* %1,
+  i64 %2,
+  <vscale x 1 x i1> %3,
+  i64 %4)
+
+  ret void
+}
+
+
+declare target("riscv.vector.tuple", <vscale x 4 x i8>, 2) @llvm.riscv.vlsseg2.triscv.vector.tuple_nxv4i8_2t(target("riscv.vector.tuple", <vscale x 4 x i8>, 2), ptr, i64, i64, i64)
+declare target("riscv.vector.tuple", <vscale x 4 x i8>, 2) @llvm.riscv.vlsseg2.mask.triscv.vector.tuple_nxv4i8_2t.nxv1i1(target("riscv.vector.tuple", <vscale x 4 x i8>, 2), ptr, i64, <vscale x 1 x i1>, i64, i64, i64)
+
+define <vscale x 1 x i32> @test_vlsseg2_nxv1i32(ptr %base, i64 %offset, i64 %vl) sanitize_address {
+; CHECK-LABEL: @test_vlsseg2_nxv1i32(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i64, ptr @__asan_shadow_memory_dynamic_address, align 8
+; CHECK-NEXT:    [[TMP24:%.*]] = tail call target("riscv.vector.tuple", <vscale x 4 x i8>, 2) @llvm.riscv.vlsseg2.triscv.vector.tuple_nxv4i8_2t.p0.i64(target("riscv.vector.tuple", <vscale x 4 x i8>, 2) undef, ptr [[BASE:%.*]], i64 [[OFFSET:%.*]], i64 [[VL:%.*]], i64 5)
+; CHECK-NEXT:    [[TMP25:%.*]] = call <vscale x 1 x i32> @llvm.riscv.tuple.extract.nxv1i32.triscv.vector.tuple_nxv4i8_2t(target("riscv.vector.tuple", <vscale x 4 x i8>, 2) [[TMP24]], i32 1)
+; CHECK-NEXT:    ret <vscale x 1 x i32> [[TMP25]]
+;
+entry:
+  %0 = tail call target("riscv.vector.tuple", <vscale x 4 x i8>, 2) @llvm.riscv.vlsseg2.triscv.vector.tuple_nxv4i8_2t(target("riscv.vector.tuple", <vscale x 4 x i8>, 2) undef, ptr %base, i64 %offset, i64 %vl, i64 5)
+  %1 = call <vscale x 1 x i32> @llvm.riscv.tuple.extract.nxv1i32.triscv.vector.tuple_nxv4i8_2t(target("riscv.vector.tuple", <vscale x 4 x i8>, 2) %0, i32 1)
+  ret <vscale x 1 x i32> %1
+}
+
+define <vscale x 1 x i32> @test_vlsseg2_mask_nxv1i32(ptr %base, i64 %offset, i64 %vl, <vscale x 1 x i1> %mask) sanitize_address {
+; CHECK-LABEL: @test_vlsseg2_mask_nxv1i32(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i64, ptr @__asan_shadow_memory_dynamic_address, align 8
+; CHECK-NEXT:    [[TMP24:%.*]] = tail call target("riscv.vector.tuple", <vscale x 4 x i8>, 2) @llvm.riscv.vlsseg2.mask.triscv.vector.tuple_nxv4i8_2t.p0.i64.nxv1i1(target("riscv.vector.tuple", <vscale x 4 x i8>, 2) undef, ptr [[BASE:%.*]], i64 [[OFFSET:%.*]], <vscale x 1 x i1> [[MASK:%.*]], i64 [[VL:%.*]], i64 1, i64 5)
+; CHECK-NEXT:    [[TMP25:%.*]] = call <vscale x 1 x i32> @llvm.riscv.tuple.extract.nxv1i32.triscv.vector.tuple_nxv4i8_2t(target("riscv.vector.tuple", <vscale x 4 x i8>, 2) [[TMP24]], i32 1)
+; CHECK-NEXT:    ret <vscale x 1 x i32> [[TMP25]]
+;
+entry:
+  %0 = tail call target("riscv.vector.tuple", <vscale x 4 x i8>, 2) @llvm.riscv.vlsseg2.mask.triscv.vector.tuple_nxv4i8_2t.nxv1i1(target("riscv.vector.tuple", <vscale x 4 x i8>, 2) undef, ptr %base, i64 %offset, <vscale x 1 x i1> %mask, i64 %vl, i64 1, i64 5)
+  %1 = call <vscale x 1 x i32> @llvm.riscv.tuple.extract.nxv1i32.triscv.vector.tuple_nxv4i8_2t(target("riscv.vector.tuple", <vscale x 4 x i8>, 2) %0, i32 1)
+  ret <vscale x 1 x i32> %1
+}
+
+
+declare target("riscv.vector.tuple", <vscale x 4 x i8>, 3) @llvm.riscv.vlsseg3.triscv.vector.tuple_nxv4i8_3t(target("riscv.vector.tuple", <vscale x 4 x i8>, 3), ptr, i64, i64, i64)
+declare target("riscv.vector.tuple", <vscale x 4 x i8>, 3) @llvm.riscv.vlsseg3.mask.triscv.vector.tuple_nxv4i8_3t.nxv1i1(target("riscv.vector.tuple", <vscale x 4 x i8>, 3), ptr, i64, <vscale x 1 x i1>, i64, i64, i64)
+
+define <vscale x 1 x i32> @test_vlsseg3_nxv1i32(ptr %base, i64 %offset, i64 %vl) sanitize_address {
+; CHECK-LABEL: @test_vlsseg3_nxv1i32(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i64, ptr @__asan_shadow_memory_dynamic_address, align 8
+; CHECK-NEXT:    [[TMP36:%.*]] = tail call target("riscv.vector.tuple", <vscale x 4 x i8>, 3) @llvm.riscv.vlsseg3.triscv.vector.tuple_nxv4i8_3t.p0.i64(target("riscv.vector.tuple", <vscale x 4 x i8>, 3) undef, ptr [[BASE:%.*]], i64 [[OFFSET:%.*]], i64 [[VL:%.*]], i64 5)
+; CHECK-NEXT:    [[TMP37:%.*]] = call <vscale x 1 x i32> @llvm.riscv.tuple.extract.nxv1i32.triscv.vector.tuple_nxv4i8_3t(target("riscv.vector.tuple", <vscale x 4 x i8>, 3) [[TMP36]], i32 1)
+; CHECK-NEXT:    ret <vscale x 1 x i32> [[TMP37]]
+;
+entry:
+  %0 = tail call target("riscv.vector.tuple", <vscale x 4 x i8>, 3) @llvm.riscv.vlsseg3.triscv.vector.tuple_nxv4i8_3t(target("riscv.vector.tuple", <vscale x 4 x i8>, 3) undef, ptr %base, i64 %offset, i64 %vl, i64 5)
+  %1 = call <vscale x 1 x i32> @llvm.riscv.tuple.extract.nxv1i32.triscv.vector.tuple_nxv4i8_3t(target("riscv.vector.tuple", <vscale x 4 x i8>, 3) %0, i32 1)
+  ret <vscale x 1 x i32> %1
+}
+
+define <vscale x 1 x i32> @test_vlsseg3_mask_nxv1i32(ptr %base, i64 %offset, i64 %vl, <vscale x 1 x i1> %mask) sanitize_address {
+; CHECK-LABEL: @test_vlsseg3_mask_nxv1i32(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i64, ptr @__asan_shadow_memory_dynamic_address, align 8
+; CHECK-NEXT:    [[TMP36:%.*]] = tail call target("riscv.vector.tuple", <vscale x 4 x i8>, 3) @llvm.riscv.vlsseg3.mask.triscv.vector.tuple_nxv4i8_3t.p0.i64.nxv1i1(target("riscv.vector.tuple", <vscale x 4 x i8>, 3) undef, ptr [[BASE:%.*]], i64 [[OFFSET:%.*]], <vscale x 1 x i1> [[MASK:%.*]], i64 [[VL:%.*]], i64 1, i64 5)
+; CHECK-NEXT:    [[TMP37:%.*]] = call <vscale x 1 x i32> @llvm.riscv.tuple.extract.nxv1i32.triscv.vector.tuple_nxv4i8_3t(target("riscv.vector.tuple", <vscale x 4 x i8>, 3) [[TMP36]], i32 1)
+; CHECK-NEXT:    ret <vscale x 1 x i32> [[TMP37]]
+;
+entry:
+  %0 = tail call target("riscv.vector.tuple", <vscale x 4 x i8>, 3) @llvm.riscv.vlsseg3.mask.triscv.vector.tuple_nxv4i8_3t.nxv1i1(target("riscv.vector.tuple", <vscale x 4 x i8>, 3) undef, ptr %base, i64 %offset, <vscale x 1 x i1> %mask, i64 %vl, i64 1, i64 5)
+  %1 = call <vscale x 1 x i32> @llvm.riscv.tuple.extract.nxv1i32.triscv.vector.tuple_nxv4i8_3t(target("riscv.vector.tuple", <vscale x 4 x i8>, 3) %0, i32 1)
+  ret <vscale x 1 x i32> %1
+}
+
+
+declare target("riscv.vector.tuple", <vscale x 4 x i8>, 4) @llvm.riscv.vlsseg4.triscv.vector.tuple_nxv4i8_4t(target("riscv.vector.tuple", <vscale x 4 x i8>, 4), ptr, i64, i64, i64)
+declare target("riscv.vector.tuple", <vscale x 4 x i8>, 4) @llvm.riscv.vlsseg4.mask.triscv.vector.tuple_nxv4i8_4t.nxv1i1(target("riscv.vector.tuple", <vscale x 4 x i8>, 4), ptr, i64, <vscale x 1 x i1>, i64, i64, i64)
+
+define <vscale x 1 x i32> @test_vlsseg4_nxv1i32(ptr %base, i64 %offset, i64 %vl) sanitize_address {
+; CHECK-LABEL: @test_vlsseg4_nxv1i32(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i64, ptr @__asan_shadow_memory_dynamic_address, align 8
+; CHECK-NEXT:    [[TMP48:%.*]] = tail call target("riscv.vector.tuple", <vscale x 4 x i8>, 4) @llvm.riscv.vlsseg4.triscv.vector.tuple_nxv4i8_4t.p0.i64(target("riscv.vector.tuple", <vscale x 4 x i8>, 4) undef, ptr [[BASE:%.*]], i64 [[OFFSET:%.*]], i64 [[VL:%.*]], i64 5)
+; CHECK-NEXT:    [[TMP49:%.*]] = call <vscale x 1 x i32> @llvm.riscv.tuple.extract.nxv1i32.triscv.vector.tuple_nxv4i8_4t(target("riscv.vector.tuple", <vscale x 4 x i8>, 4) [[TMP48]], i32 1)
+; CHECK-NEXT:    ret <vscale x 1 x i32> [[TMP49]]
+;
+entry:
+  %0 = tail call target("riscv.vector.tuple", <vscale x 4 x i8>, 4) @llvm.riscv.vlsseg4.triscv.vector.tuple_nxv4i8_4t(target("riscv.vector.tuple", <vscale x 4 x i8>, 4) undef, ptr %base, i64 %offset, i64 %vl, i64 5)
+  %1 = call <vscale x 1 x i32> @llvm.riscv.tuple.extract.nxv1i32.triscv.vector.tuple_nxv4i8_4t(target("riscv.vector.tuple", <vscale x 4 x i8>, 4) %0, i32 1)
+  ret <vscale x 1 x i32> %1
+}
+
+define <vscale x 1 x i32> @test_vlsseg4_mask_nxv1i32(ptr %base, i64 %offset, i64 %vl, <vscale x 1 x i1> %mask) sanitize_address {
+; CHECK-LABEL: @test_vlsseg4_mask_nxv1i32(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i64, ptr @__asan_shadow_memory_dynamic_address, align 8
+; CHECK-NEXT:    [[TMP48:%.*]] = tail call target("riscv.vector.tuple", <vscale x 4 x i8>, 4) @llvm.riscv.vlsseg4.mask.triscv.vector.tuple_nxv4i8_4t.p0.i64.nxv1i1(target("riscv.vector.tuple", <vscale x 4 x i8>, 4) undef, ptr [[BASE:%.*]], i64 [[OFFSET:%.*]], <vscale x 1 x i1> [[MASK:%.*]], i64 [[VL:%.*]], i64 1, i64 5)
+; CHECK-NEXT:    [[TMP49:%.*]] = call <vscale x 1 x i32> @llvm.riscv.tuple.extract.nxv1i32.triscv.vector.tuple_nxv4i8_4t(target("riscv.vector.tuple", <vscale x 4 x i8>, 4) [[TMP48]], i32 1)
+; CHECK-NEXT:    ret <vscale x 1 x i32> [[TMP49]]
+;
+entry:
+  %0 = tail call target("riscv.vector.tuple", <vscale x 4 x i8>, 4) @llvm.riscv.vlsseg4.mask.triscv.vector.tuple_nxv4i8_4t.nxv1i1(target("riscv.vector.tuple", <vscale x 4 x i8>, 4) undef, ptr %base, i64 %offset, <vscale x 1 x i1> %mask, i64 %vl, i64 1, i64 5)
+  %1 = call <vscale x 1 x i32> @llvm.riscv.tuple.extract.nxv1i32.triscv.vector.tuple_nxv4i8_4t(target("riscv.vector.tuple", <vscale x 4 x i8>, 4) %0, i32 1)
+  ret <vscale x 1 x i32> %1
+}
+
+
+declare target("riscv.vector.tuple", <vscale x 4 x i8>, 5) @llvm.riscv.vlsseg5.triscv.vector.tuple_nxv4i8_5t(target("riscv.vector.tuple", <vscale x 4 x i8>, 5), ptr, i64, i64, i64)
+declare target("riscv.vector.tuple", <vscale x 4 x i8>, 5) @llvm.riscv.vlsseg5.mask.triscv.vector.tuple_nxv4i8_5t.nxv1i1(target("riscv.vector.tuple", <vscale x 4 x i8>, 5), ptr, i64, <vscale x 1 x i1>, i64, i64, i64)
+
+define <vscale x 1 x i32> @test_vlsseg5_nxv1i32(ptr %base, i64 %offset, i64 %vl) sanitize_address {
+; CHECK-LABEL: @test_vlsseg5_nxv1i32(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i64, ptr @__asan_shadow_memory_dynamic_address, align 8
+; CHECK-NEXT:    [[TMP60:%.*]] = tail call target("riscv.vector.tuple", <vscale x 4 x i8>, 5) @llvm.riscv.vlsseg5.triscv.vector.tuple_nxv4i8_5t.p0.i64(target("riscv.vector.tuple", <vscale x 4 x i8>, 5) undef, ptr [[BASE:%.*]], i64 [[OFFSET:%.*]], i64 [[VL:%.*]], i64 5)
+; CHECK-NEXT:    [[TMP61:%.*]] = call <vscale x 1 x i32> @llvm.riscv.tuple.extract.nxv1i32.triscv.vector.tuple_nxv4i8_5t(target("riscv.vector.tuple", <vscale x 4 x i8>, 5) [[TMP60]], i32 1)
+; CHECK-NEXT:    ret <vscale x 1 x i32> [[TMP61]]
+;
+entry:
+  %0 = tail call target("riscv.vector.tuple", <vscale x 4 x i8>, 5) @llvm.riscv.vlsseg5.triscv.vector.tuple_nxv4i8_5t(target("riscv.vector.tuple", <vscale x 4 x i8>, 5) undef, ptr %base, i64 %offset, i64 %vl, i64 5)
+  %1 = call <vscale x 1 x i32> @llvm.riscv.tuple.extract.nxv1i32.triscv.vector.tuple_nxv4i8_5t(target("riscv.vector.tuple", <vscale x 4 x i8>, 5) %0, i32 1)
+  ret <vscale x 1 x i32> %1
+}
+
+define <vscale x 1 x i32> @test_vlsseg5_mask_nxv1i32(ptr %base, i64 %offset, i64 %vl, <vscale x 1 x i1> %mask) sanitize_address {
+; CHECK-LABEL: @test_vlsseg5_mask_nxv1i32(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i64, ptr @__asan_shadow_memory_dynamic_address, align 8
+; CHECK-NEXT:    [[TMP60:%.*]] = tail call target("riscv.vector.tuple", <vscale x 4 x i8>, 5) @llvm.riscv.vlsseg5.mask.triscv.vector.tuple_nxv4i8_5t.p0.i64.nxv1i1(target("riscv.vector.tuple", <vscale x 4 x i8>, 5) undef, ptr [[BASE:%.*]], i64 [[OFFSET:%.*]], <vscale x 1 x i1> [[MASK:%.*]], i64 [[VL:%.*]], i64 1, i64 5)
+; CHECK-NEXT:    [[TMP61:%.*]] = call <vscale x 1 x i32> @llvm.riscv.tuple.extract.nxv1i32.triscv.vector.tuple_nxv4i8_5t(target("riscv.vector.tuple", <vscale x 4 x i8>, 5) [[TMP60]], i32 1)
+; CHECK-NEXT:    ret <vscale x 1 x i32> [[TMP61]]
+;
+entry:
+  %0 = tail call target("riscv.vector.tuple", <vscale x 4 x i8>, 5) @llvm.riscv.vlsseg5.mask.triscv.vector.tuple_nxv4i8_5t.nxv1i1(target("riscv.vector.tuple", <vscale x 4 x i8>, 5) undef, ptr %base, i64 %offset, <vscale x 1 x i1> %mask, i64 %vl, i64 1, i64 5)
+  %1 = call <vscale x 1 x i32> @llvm.riscv.tuple.extract.nxv1i32.triscv.vector.tuple_nxv4i8_5t(target("riscv.vector.tuple", <vscale x 4 x i8>, 5) %0, i32 1)
+  ret <vscale x 1 x i32> %1
+}
+
+
+declare target("riscv.vector.tuple", <vscale x 4 x i8>, 6) @llvm.riscv.vlsseg6.triscv.vector.tuple_nxv4i8_6t(target("riscv.vector.tuple", <vscale x 4 x i8>, 6), ptr, i64, i64, i64)
+declare target("riscv.vector.tuple", <vscale x 4 x i8>, 6) @llvm.riscv.vlsseg6.mask.triscv.vector.tuple_nxv4i8_6t.nxv1i1(target("riscv.vector.tuple", <vscale x 4 x i8>, 6), ptr, i64, <vscale x 1 x i1>, i64, i64, i64)
+
+define <vscale x 1 x i32> @test_vlsseg6_nxv1i32(ptr %base, i64 %offset, i64 %vl) sanitize_address {
+; CHECK-LABEL: @test_vlsseg6_nxv1i32(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i64, ptr @__asan_shadow_memory_dynamic_address, align 8
+; CHECK-NEXT:    [[TMP72:%.*]] = tail call target("riscv.vector.tuple", <vscale x 4 x i8>, 6) @llvm.riscv.vlsseg6.triscv.vector.tuple_nxv4i8_6t.p0.i64(target("riscv.vector.tuple", <vscale x 4 x i8>, 6) undef, ptr [[BASE:%.*]], i64 [[OFFSET:%.*]], i64 [[VL:%.*]], i64 5)
+; CHECK-NEXT:    [[TMP73:%.*]] = call <vscale x 1 x i32> @llvm.riscv.tuple.extract.nxv1i32.triscv.vector.tuple_nxv4i8_6t(target("riscv.vector.tuple", <vscale x 4 x i8>, 6) [[TMP72]], i32 1)
+; CHECK-NEXT:    ret <vscale x 1 x i32> [[TMP73]]
+;
+entry:
+  %0 = tail call target("riscv.vector.tuple", <vscale x 4 x i8>, 6) @llvm.riscv.vlsseg6.triscv.vector.tuple_nxv4i8_6t(target("riscv.vector.tuple", <vscale x 4 x i8>, 6) undef, ptr %base, i64 %offset, i64 %vl, i64 5)
+  %1 = call <vscale x 1 x i32> @llvm.riscv.tuple.extract.nxv1i32.triscv.vector.tuple_nxv4i8_6t(target("riscv.vector.tuple", <vscale x 4 x i8>, 6) %0, i32 1)
+  ret <vscale x 1 x i32> %1
+}
+
+define <vscale x 1 x i32> @test_vlsseg6_mask_nxv1i32(ptr %base, i64 %offset, i64 %vl, <vscale x 1 x i1> %mask) sanitize_address {
+; CHECK-LABEL: @test_vlsseg6_mask_nxv1i32(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i64, ptr @__asan_shadow_memory_dynamic_address, align 8
+; CHECK-NEXT:    [[TMP72:%.*]] = tail call target("riscv.vector.tuple", <vscale x 4 x i8>, 6) @llvm.riscv.vlsseg6.mask.triscv.vector.tuple_nxv4i8_6t.p0.i64.nxv1i1(target("riscv.vector.tuple", <vscale x 4 x i8>, 6) undef, ptr [[BASE:%.*]], i64 [[OFFSET:%.*]], <vscale x 1 x i1> [[MASK:%.*]], i64 [[VL:%.*]], i64 1, i64 5)
+; CHECK-NEXT:    [[TMP73:%.*]] = call <vscale x 1 x i32> @llvm.riscv.tuple.extract.nxv1i32.triscv.vector.tuple_nxv4i8_6t(target("riscv.vector.tuple", <vscale x 4 x i8>, 6) [[TMP72]], i32 1)
+; CHECK-NEXT:    ret <vscale x 1 x i32> [[TMP73]]
+;
+entry:
+  %0 = tail call target("riscv.vector.tuple", <vscale x 4 x i8>, 6) @llvm.riscv.vlsseg6.mask.triscv.vector.tuple_nxv4i8_6t.nxv1i1(target("riscv.vector.tuple", <vscale x 4 x i8>, 6) undef, ptr %base, i64 %offset, <vscale x 1 x i1> %mask, i64 %vl, i64 1, i64 5)
+  %1 = call <vscale x 1 x i32> @llvm.riscv.tuple.extract.nxv1i32.triscv.vector.tuple_nxv4i8_6t(target("riscv.vector.tuple", <vscale x 4 x i8>, 6) %0, i32 1)
+  ret <vscale x 1 x i32> %1
+}
+
+
+declare target("riscv.vector.tuple", <vscale x 4 x i8>, 7) @llvm.riscv.vlsseg7.triscv.vector.tuple_nxv4i8_7t(target("riscv.vector.tuple", <vscale x 4 x i8>, 7), ptr, i64, i64, i64)
+declare target("riscv.vector.tuple", <vscale x 4 x i8>, 7) @llvm.riscv.vlsseg7.mask.triscv.vector.tuple_nxv4i8_7t.nxv1i1(target("riscv.vector.tuple", <vscale x 4 x i8>, 7), ptr, i64, <vscale x 1 x i1>, i64, i64, i64)
+
+define <vscale x 1 x i32> @test_vlsseg7_nxv1i32(ptr %base, i64 %offset, i64 %vl) sanitize_address {
+; CHECK-LABEL: @test_vlsseg7_nxv1i32(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i64, ptr @__asan_shadow_memory_dynamic_address, align 8
+; CHECK-NEXT:    [[TMP84:%.*]] = tail call target("riscv.vector.tuple", <vscale x 4 x i8>, 7) @llvm.riscv.vlsseg7.triscv.vector.tuple_nxv4i8_7t.p0.i64(target("riscv.vector.tuple", <vscale x 4 x i8>, 7) undef, ptr [[BASE:%.*]], i64 [[OFFSET:%.*]], i64 [[VL:%.*]], i64 5)
+; CHECK-NEXT:    [[TMP85:%.*]] = call <vscale x 1 x i32> @llvm.riscv.tuple.extract.nxv1i32.triscv.vector.tuple_nxv4i8_7t(target("riscv.vector.tuple", <vscale x 4 x i8>, 7) [[TMP84]], i32 1)
+; CHECK-NEXT:    ret <vscale x 1 x i32> [[TMP85]]
+;
+entry:
+  %0 = tail call target("riscv.vector.tuple", <vscale x 4 x i8>, 7) @llvm.riscv.vlsseg7.triscv.vector.tuple_nxv4i8_7t(target("riscv.vector.tuple", <vscale x 4 x i8>, 7) undef, ptr %base, i64 %offset, i64 %vl, i64 5)
+  %1 = call <vscale x 1 x i32> @llvm.riscv.tuple.extract.nxv1i32.triscv.vector.tuple_nxv4i8_7t(target("riscv.vector.tuple", <vscale x 4 x i8>, 7) %0, i32 1)
+  ret <vscale x 1 x i32> %1
+}
+
+define <vscale x 1 x i32> @test_vlsseg7_mask_nxv1i32(ptr %base, i64 %offset, i64 %vl, <vscale x 1 x i1> %mask) sanitize_address {
+; CHECK-LABEL: @test_vlsseg7_mask_nxv1i32(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i64, ptr @__asan_shadow_memory_dynamic_address, align 8
+; CHECK-NEXT:    [[TMP84:%.*]] = tail call target("riscv.vector.tuple", <vscale x 4 x i8>, 7) @llvm.riscv.vlsseg7.mask.triscv.vector.tuple_nxv4i8_7t.p0.i64.nxv1i1(target("riscv.vector.tuple", <vscale x 4 x i8>, 7) undef, ptr [[BASE:%.*]], i64 [[OFFSET:%.*]], <vscale x 1 x i1> [[MASK:%.*]], i64 [[VL:%.*]], i64 1, i64 5)
+; CHECK-NEXT:    [[TMP85:%.*]] = call <vscale x 1 x i32> @llvm.riscv.tuple.extract.nxv1i32.triscv.vector.tuple_nxv4i8_7t(target("riscv.vector.tuple", <vscale x 4 x i8>, 7) [[TMP84]], i32 1)
+; CHECK-NEXT:    ret <vscale x 1 x i32> [[TMP85]]
+;
+entry:
+  %0 = tail call target("riscv.vector.tuple", <vscale x 4 x i8>, 7) @llvm.riscv.vlsseg7.mask.triscv.vector.tuple_nxv4i8_7t.nxv1i1(target("riscv.vector.tuple", <vscale x 4 x i8>, 7) undef, ptr %base, i64 %offset, <vscale x 1 x i1> %mask, i64 %vl, i64 1, i64 5)
+  %1 = call <vscale x 1 x i32> @llvm.riscv.tuple.extract.nxv1i32.triscv.vector.tuple_nxv4i8_7t(target("riscv.vector.tuple", <vscale x 4 x i8>, 7) %0, i32 1)
+  ret <vscale x 1 x i32> %1
+}
+
+
+declare target("riscv.vector.tuple", <vscale x 4 x i8>, 8) @llvm.riscv.vlsseg8.triscv.vector.tuple_nxv4i8_8t(target("riscv.vector.tuple", <vscale x 4 x i8>, 8), ptr, i64, i64, i64)
+declare target("riscv.vector.tuple", <vscale x 4 x i8>, 8) @llvm.riscv.vlsseg8.mask.triscv.vector.tuple_nxv4i8_8t.nxv1i1(target("riscv.vector.tuple", <vscale x 4 x i8>, 8), ptr, i64, <vscale x 1 x i1>, i64, i64, i64)
+
+define <vscale x 1 x i32> @test_vlsseg8_nxv1i32(ptr %base, i64 %offset, i64 %vl) sanitize_address {
+; CHECK-LABEL: @test_vlsseg8_nxv1i32(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i64, ptr @__asan_shadow_memory_dynamic_address, align 8
+; CHECK-NEXT:    [[TMP96:%.*]] = tail call target("riscv.vector.tuple", <vscale x 4 x i8>, 8) @llvm.riscv.vlsseg8.triscv.vector.tuple_nxv4i8_8t.p0.i64(target("riscv.vector.tuple", <vscale x 4 x i8>, 8) undef, ptr [[BASE:%.*]], i64 [[OFFSET:%.*]], i64 [[VL:%.*]], i64 5)
+; CHECK-NEXT:    [[TMP97:%.*]] = call <vscale x 1 x i32> @llvm.riscv.tuple.extract.nxv1i32.triscv.vector.tuple_nxv4i8_8t(target("riscv.vector.tuple", <vscale x 4 x i8>, 8) [[TMP96]], i32 1)
+; CHECK-NEXT:    ret <vscale x 1 x i32> [[TMP97]]
+;
+entry:
+  %0 = tail call target("riscv.vector.tuple", <vscale x 4 x i8>, 8) @llvm.riscv.vlsseg8.triscv.vector.tuple_nxv4i8_8t(target("riscv.vector.tuple", <vscale x 4 x i8>, 8) undef, ptr %base, i64 %offset, i64 %vl, i64 5)
+  %1 = call <vscale x 1 x i32> @llvm.riscv.tuple.extract.nxv1i32.triscv.vector.tuple_nxv4i8_8t(target("riscv.vector.tuple", <vscale x 4 x i8>, 8) %0, i32 1)
+  ret <vscale x 1 x i32> %1
+}
+
+define <vscale x 1 x i32> @test_vlsseg8_mask_nxv1i32(ptr %base, i64 %offset, i64 %vl, <vscale x 1 x i1> %mask) sanitize_address {
+; CHECK-LABEL: @test_vlsseg8_mask_nxv1i32(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i64, ptr @__asan_shadow_memory_dynamic_address, align 8
+; CHECK-NEXT:    [[TMP96:%.*]] = tail call target("riscv.vector.tuple", <vscale x 4 x i8>, 8) @llvm.riscv.vlsseg8.mask.triscv.vector.tuple_nxv4i8_8t.p0.i64.nxv1i1(target("riscv.vector.tuple", <vscale x 4 x i8>, 8) undef, ptr [[BASE:%.*]], i64 [[OFFSET:%.*]], <vscale x 1 x i1> [[MASK:%.*]], i64 [[VL:%.*]], i64 1, i64 5)
+; CHECK-NEXT:    [[TMP97:%.*]] = call <vscale x 1 x i32> @llvm.riscv.tuple.extract.nxv1i32.triscv.vector.tuple_nxv4i8_8t(target("riscv.vector.tuple", <vscale x 4 x i8>, 8) [[TMP96]], i32 1)
+; CHECK-NEXT:    ret <vscale x 1 x i32> [[TMP97]]
+;
+entry:
+  %0 = tail call target("riscv.vector.tuple", <vscale x 4 x i8>, 8) @llvm.riscv.vlsseg8.mask.triscv.vector.tuple_nxv4i8_8t.nxv1i1(target("riscv.vector.tuple", <vscale x 4 x i8>, 8) undef, ptr %base, i64 %offset, <vscale x 1 x i1> %mask, i64 %vl, i64 1, i64 5)
+  %1 = call <vscale x 1 x i32> @llvm.riscv.tuple.extract.nxv1i32.triscv.vector.tuple_nxv4i8_8t(target("riscv.vector.tuple", <vscale x 4 x i8>, 8) %0, i32 1)
+  ret <vscale x 1 x i32> %1
+}
+
+
+declare void @llvm.riscv.vssseg2.triscv.vector.tuple_nxv4i8_2t(target("riscv.vector.tuple", <vscale x 4 x i8>, 2), ptr, i64, i64, i64)
+declare void @llvm.riscv.vssseg2.mask.triscv.vector.tuple_nxv4i8_2t.nxv1i1(target("riscv.vector.tuple", <vscale x 4 x i8>, 2), ptr, i64, <vscale x 1 x i1>, i64, i64)
+
+define void @test_vssseg2_nxv1i32(target("riscv.vector.tuple", <vscale x 4 x i8>, 2) %val, ptr %base, i64 %offset, i64 %vl) sanitize_address {
+; CHECK-LABEL: @test_vssseg2_nxv1i32(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i64, ptr @__asan_shadow_memory_dynamic_address, align 8
+; CHECK-NEXT:    tail call void @llvm.riscv.vssseg2.triscv.vector.tuple_nxv4i8_2t.p0.i64(target("riscv.vector.tuple", <vscale x 4 x i8>, 2) [[VAL:%.*]], ptr [[BASE:%.*]], i64 [[OFFSET:%.*]], i64 [[VL:%.*]], i64 5)
+; CHECK-NEXT:    ret void
+;
+entry:
+  tail call void @llvm.riscv.vssseg2.triscv.vector.tuple_nxv4i8_2t(target("riscv.vector.tuple", <vscale x 4 x i8>, 2) %val, ptr %base, i64 %offset, i64 %vl, i64 5)
+  ret void
+}
+
+define void @test_vssseg2_mask_nxv1i32(target("riscv.vector.tuple", <vscale x 4 x i8>, 2) %val, ptr %base, i64 %offset, i64 %vl, <vscale x 1 x i1> %mask) sanitize_address {
+; CHECK-LABEL: @test_vssseg2_mask_nxv1i32(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i64, ptr @__asan_shadow_memory_dynamic_address, align 8
+; CHECK-NEXT:    tail call void @llvm.riscv.vssseg2.mask.triscv.vector.tuple_nxv4i8_2t.p0.i64.nxv1i1(target("riscv.vector.tuple", <vscale x 4 x i8>, 2) [[VAL:%.*]], ptr [[BASE:%.*]], i64 [[OFFSET:%.*]], <vscale x 1 x i1> [[MASK:%.*]], i64 [[VL:%.*]], i64 5)
+; CHECK-NEXT:    ret void
+;
+entry:
+  tail call void @llvm.riscv.vssseg2.mask.triscv.vector.tuple_nxv4i8_2t.nxv1i1(target("riscv.vector.tuple", <vscale x 4 x i8>, 2) %val, ptr %base, i64 %offset, <vscale x 1 x i1> %mask, i64 %vl, i64 5)
+  ret void
+}
+
+
+declare void @llvm.riscv.vssseg3.triscv.vector.tuple_nxv4i8_3t(target("riscv.vector.tuple", <vscale x 4 x i8>, 3), ptr, i64, i64, i64)
+declare void @llvm.riscv.vssseg3.mask.triscv.vector.tuple_nxv4i8_3t.nxv1i1(target("riscv.vector.tuple", <vscale x 4 x i8>, 3), ptr, i64, <vscale x 1 x i1>, i64, i64)
+
+define void @test_vssseg3_nxv1i32(target("riscv.vector.tuple", <vscale x 4 x i8>, 3) %val, ptr %base, i64 %offset, i64 %vl) sanitize_address {
+; CHECK-LABEL: @test_vssseg3_nxv1i32(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i64, ptr @__asan_shadow_memory_dynamic_address, align 8
+; CHECK-NEXT:    tail call void @llvm.riscv.vssseg3.triscv.vector.tuple_nxv4i8_3t.p0.i64(target("riscv.vector.tuple", <vscale x 4 x i8>, 3) [[VAL:%.*]], ptr [[BASE:%.*]], i64 [[OFFSET:%.*]], i64 [[VL:%.*]], i64 5)
+; CHECK-NEXT:    ret void
+;
+entry:
+  tail call void @llvm.riscv.vssseg3.triscv.vector.tuple_nxv4i8_3t(target("riscv.vector.tuple", <vscale x 4 x i8>, 3) %val, ptr %base, i64 %offset, i64 %vl, i64 5)
+  ret void
+}
+
+define void @test_vssseg3_mask_nxv1i32(target("riscv.vector.tuple", <vscale x 4 x i8>, 3) %val, ptr %base, i64 %offset, i64 %vl, <vscale x 1 x i1> %mask) sanitize_address {
+; CHECK-LABEL: @test_vssseg3_mask_nxv1i32(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i64, ptr @__asan_shadow_memory_dynamic_address, align 8
+; CHECK-NEXT:    tail call void @llvm.riscv.vssseg3.mask.triscv.vector.tuple_nxv4i8_3t.p0.i64.nxv1i1(target("riscv.vector.tuple", <vscale x 4 x i8>, 3) [[VAL:%.*]], ptr [[BASE:%.*]], i64 [[OFFSET:%.*]], <vscale x 1 x i1> [[MASK:%.*]], i64 [[VL:%.*]], i64 5)
+; CHECK-NEXT:    ret void
+;
+entry:
+  tail call void @llvm.riscv.vssseg3.mask.triscv.vector.tuple_nxv4i8_3t.nxv1i1(target("riscv.vector.tuple", <vscale x 4 x i8>, 3) %val, ptr %base, i64 %offset, <vscale x 1 x i1> %mask, i64 %vl, i64 5)
+  ret void
+}
+
+
+declare void @llvm.riscv.vssseg4.triscv.vector.tuple_nxv4i8_4t(target("riscv.vector.tuple", <vscale x 4 x i8>, 4), ptr, i64, i64, i64)
+declare void @llvm.riscv.vssseg4.mask.triscv.vector.tuple_nxv4i8_4t.nxv1i1(target("riscv.vector.tuple", <vscale x 4 x i8>, 4), ptr, i64, <vscale x 1 x i1>, i64, i64)
+
+define void @test_vssseg4_nxv1i32(target("riscv.vector.tuple", <vscale x 4 x i8>, 4) %val, ptr %base, i64 %offset, i64 %vl) sanitize_address {
+; CHECK-LABEL: @test_vssseg4_nxv1i32(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i64, ptr @__asan_shadow_memory_dynamic_address, align 8
+; CHECK-NEXT:    tail call void @llvm.riscv.vssseg4.triscv.vector.tuple_nxv4i8_4t.p0.i64(target("riscv.vector.tuple", <vscale x 4 x i8>, 4) [[VAL:%.*]], ptr [[BASE:%.*]], i64 [[OFFSET:%.*]], i64 [[VL:%.*]], i64 5)
+; CHECK-NEXT:    ret void
+;
+entry:
+  tail call void @llvm.riscv.vssseg4.triscv.vector.tuple_nxv4i8_4t(target("riscv.vector.tuple", <vscale x 4 x i8>, 4) %val, ptr %base, i64 %offset, i64 %vl, i64 5)
+  ret void
+}
+
+define void @test_vssseg4_mask_nxv1i32(target("riscv.vector.tuple", <vscale x 4 x i8>, 4) %val, ptr %base, i64 %offset, i64 %vl, <vscale x 1 x i1> %mask) sanitize_address {
+; CHECK-LABEL: @test_vssseg4_mask_nxv1i32(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i64, ptr @__asan_shadow_memory_dynamic_address, align 8
+; CHECK-NEXT:    tail call void @llvm.riscv.vssseg4.mask.triscv.vector.tuple_nxv4i8_4t.p0.i64.nxv1i1(target("riscv.vector.tuple", <vscale x 4 x i8>, 4) [[VAL:%.*]], ptr [[BASE:%.*]], i64 [[OFFSET:%.*]], <vscale x 1 x i1> [[MASK:%.*]], i64 [[VL:%.*]], i64 5)
+; CHECK-NEXT:    ret void
+;
+entry:
+  tail call void @llvm.riscv.vssseg4.mask.triscv.vector.tuple_nxv4i8_4t.nxv1i1(target("riscv.vector.tuple", <vscale x 4 x i8>, 4) %val, ptr %base, i64 %offset, <vscale x 1 x i1> %mask, i64 %vl, i64 5)
+  ret void
+}
+
+
+declare void @llvm.riscv.vssseg5.triscv.vector.tuple_nxv4i8_5t(target("riscv.vector.tuple", <vscale x 4 x i8>, 5), ptr, i64, i64, i64)
+declare void @llvm.riscv.vssseg5.mask.triscv.vector.tuple_nxv4i8_5t.nxv1i1(target("riscv.vector.tuple", <vscale x 4 x i8>, 5), ptr, i64, <vscale x 1 x i1>, i64, i64)
+
+define void @test_vssseg5_nxv1i32(target("riscv.vector.tuple", <vscale x 4 x i8>, 5) %val, ptr %base, i64 %offset, i64 %vl) sanitize_address {
+; CHECK-LABEL: @test_vssseg5_nxv1i32(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i64, ptr @__asan_shadow_memory_dynamic_address, align 8
+; CHECK-NEXT:    tail call void @llvm.riscv.vssseg5.triscv.vector.tuple_nxv4i8_5t.p0.i64(target("riscv.vector.tuple", <vscale x 4 x i8>, 5) [[VAL:%.*]], ptr [[BASE:%.*]], i64 [[OFFSET:%.*]], i64 [[VL:%.*]], i64 5)
+; CHECK-NEXT:    ret void
+;
+entry:
+  tail call void @llvm.riscv.vssseg5.triscv.vector.tuple_nxv4i8_5t(target("riscv.vector.tuple", <vscale x 4 x i8>, 5) %val, ptr %base, i64 %offset, i64 %vl, i64 5)
+  ret void
+}
+
+define void @test_vssseg5_mask_nxv1i32(target("riscv.vector.tuple", <vscale x 4 x i8>, 5) %val, ptr %base, i64 %offset, i64 %vl, <vscale x 1 x i1> %mask) sanitize_address {
+; CHECK-LABEL: @test_vssseg5_mask_nxv1i32(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i64, ptr @__asan_shadow_memory_dynamic_address, align 8
+; CHECK-NEXT:    tail call void @llvm.riscv.vssseg5.mask.triscv.vector.tuple_nxv4i8_5t.p0.i64.nxv1i1(target("riscv.vector.tuple", <vscale x 4 x i8>, 5) [[VAL:%.*]], ptr [[BASE:%.*]], i64 [[OFFSET:%.*]], <vscale x 1 x i1> [[MASK:%.*]], i64 [[VL:%.*]], i64 5)
+; CHECK-NEXT:    ret void
+;
+entry:
+  tail call void @llvm.riscv.vssseg5.mask.triscv.vector.tuple_nxv4i8_5t.nxv1i1(target("riscv.vector.tuple", <vscale x 4 x i8>, 5) %val, ptr %base, i64 %offset, <vscale x 1 x i1> %mask, i64 %vl, i64 5)
+  ret void
+}
+
+
+declare void @llvm.riscv.vssseg6.triscv.vector.tuple_nxv4i8_6t(target("riscv.vector.tuple", <vscale x 4 x i8>, 6), ptr, i64, i64, i64)
+declare void @llvm.riscv.vssseg6.mask.triscv.vector.tuple_nxv4i8_6t.nxv1i1(target("riscv.vector.tuple", <vscale x 4 x i8>, 6), ptr, i64, <vscale x 1 x i1>, i64, i64)
+
+define void @test_vssseg6_nxv1i32(target("riscv.vector.tuple", <vscale x 4 x i8>, 6) %val, ptr %base, i64 %offset, i64 %vl) sanitize_address {
+; CHECK-LABEL: @test_vssseg6_nxv1i32(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i64, ptr @__asan_shadow_memory_dynamic_address, align 8
+; CHECK-NEXT:    tail call void @llvm.riscv.vssseg6.triscv.vector.tuple_nxv4i8_6t.p0.i64(target("riscv.vector.tuple", <vscale x 4 x i8>, 6) [[VAL:%.*]], ptr [[BASE:%.*]], i64 [[OFFSET:%.*]], i64 [[VL:%.*]], i64 5)
+; CHECK-NEXT:    ret void
+;
+entry:
+  tail call void @llvm.riscv.vssseg6.triscv.vector.tuple_nxv4i8_6t(target("riscv.vector.tuple", <vscale x 4 x i8>, 6) %val, ptr %base, i64 %offset, i64 %vl, i64 5)
+  ret void
+}
+
+define void @test_vssseg6_mask_nxv1i32(target("riscv.vector.tuple", <vscale x 4 x i8>, 6) %val, ptr %base, i64 %offset, i64 %vl, <vscale x 1 x i1> %mask) sanitize_address {
+; CHECK-LABEL: @test_vssseg6_mask_nxv1i32(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i64, ptr @__asan_shadow_memory_dynamic_address, align 8
+; CHECK-NEXT:    tail call void @llvm.riscv.vssseg6.mask.triscv.vector.tuple_nxv4i8_6t.p0.i64.nxv1i1(target("riscv.vector.tuple", <vscale x 4 x i8>, 6) [[VAL:%.*]], ptr [[BASE:%.*]], i64 [[OFFSET:%.*]], <vscale x 1 x i1> [[MASK:%.*]], i64 [[VL:%.*]], i64 5)
+; CHECK-NEXT:    ret void
+;
+entry:
+  tail call void @llvm.riscv.vssseg6.mask.triscv.vector.tuple_nxv4i8_6t.nxv1i1(target("riscv.vector.tuple", <vscale x 4 x i8>, 6) %val, ptr %base, i64 %offset, <vscale x 1 x i1> %mask, i64 %vl, i64 5)
+  ret void
+}
+
+
+declare void @llvm.riscv.vssseg7.triscv.vector.tuple_nxv4i8_7t(target("riscv.vector.tuple", <vscale x 4 x i8>, 7), ptr, i64, i64, i64)
+declare void @llvm.riscv.vssseg7.mask.triscv.vector.tuple_nxv4i8_7t.nxv1i1(target("riscv.vector.tuple", <vscale x 4 x i8>, 7), ptr, i64, <vscale x 1 x i1>, i64, i64)
+
+define void @test_vssseg7_nxv1i32(target("riscv.vector.tuple", <vscale x 4 x i8>, 7) %val, ptr %base, i64 %offset, i64 %vl) sanitize_address {
+; CHECK-LABEL: @test_vssseg7_nxv1i32(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i64, ptr @__asan_shadow_memory_dynamic_address, align 8
+; CHECK-NEXT:    tail call void @llvm.riscv.vssseg7.triscv.vector.tuple_nxv4i8_7t.p0.i64(target("riscv.vector.tuple", <vscale x 4 x i8>, 7) [[VAL:%.*]], ptr [[BASE:%.*]], i64 [[OFFSET:%.*]], i64 [[VL:%.*]], i64 5)
+; CHECK-NEXT:    ret void
+;
+entry:
+  tail call void @llvm.riscv.vssseg7.triscv.vector.tuple_nxv4i8_7t(target("riscv.vector.tuple", <vscale x 4 x i8>, 7) %val, ptr %base, i64 %offset, i64 %vl, i64 5)
+  ret void
+}
+
+define void @test_vssseg7_mask_nxv1i32(target("riscv.vector.tuple", <vscale x 4 x i8>, 7) %val, ptr %base, i64 %offset, i64 %vl, <vscale x 1 x i1> %mask) sanitize_address {
+; CHECK-LABEL: @test_vssseg7_mask_nxv1i32(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i64, ptr @__asan_shadow_memory_dynamic_address, align 8
+; CHECK-NEXT:    tail call void @llvm.riscv.vssseg7.mask.triscv.vector.tuple_nxv4i8_7t.p0.i64.nxv1i1(target("riscv.vector.tuple", <vscale x 4 x i8>, 7) [[VAL:%.*]], ptr [[BASE:%.*]], i64 [[OFFSET:%.*]], <vscale x 1 x i1> [[MASK:%.*]], i64 [[VL:%.*]], i64 5)
+; CHECK-NEXT:    ret void
+;
+entry:
+  tail call void @llvm.riscv.vssseg7.mask.triscv.vector.tuple_nxv4i8_7t.nxv1i1(target("riscv.vector.tuple", <vscale x 4 x i8>, 7) %val, ptr %base, i64 %offset, <vscale x 1 x i1> %mask, i64 %vl, i64 5)
+  ret void
+}
+
+
+declare void @llvm.riscv.vssseg8.triscv.vector.tuple_nxv4i8_8t(target("riscv.vector.tuple", <vscale x 4 x i8>, 8), ptr, i64, i64, i64)
+declare void @llvm.riscv.vssseg8.mask.triscv.vector.tuple_nxv4i8_8t.nxv1i1(target("riscv.vector.tuple", <vscale x 4 x i8>, 8), ptr, i64, <vscale x 1 x i1>, i64, i64)
+
+define void @test_vssseg8_nxv1i32(target("riscv.vector.tuple", <vscale x 4 x i8>, 8) %val, ptr %base, i64 %offset, i64 %vl) sanitize_address {
+; CHECK-LABEL: @test_vssseg8_nxv1i32(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i64, ptr @__asan_shadow_memory_dynamic_address, align 8
+; CHECK-NEXT:    tail call void @llvm.riscv.vssseg8.triscv.vector.tuple_nxv4i8_8t.p0.i64(target("riscv.vector.tuple", <vscale x 4 x i8>, 8) [[VAL:%.*]], ptr [[BASE:%.*]], i64 [[OFFSET:%.*]], i64 [[VL:%.*]], i64 5)
+; CHECK-NEXT:    ret void
+;
+entry:
+  tail call void @llvm.riscv.vssseg8.triscv.vector.tuple_nxv4i8_8t(target("riscv.vector.tuple", <vscale x 4 x i8>, 8) %val, ptr %base, i64 %offset, i64 %vl, i64 5)
+  ret void
+}
+
+define void @test_vssseg8_mask_nxv1i32(target("riscv.vector.tuple", <vscale x 4 x i8>, 8) %val, ptr %base, i64 %offset, i64 %vl, <vscale x 1 x i1> %mask) sanitize_address {
+; CHECK-LABEL: @test_vssseg8_mask_nxv1i32(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i64, ptr @__asan_shadow_memory_dynamic_address, align 8
+; CHECK-NEXT:    tail call void @llvm.riscv.vssseg8.mask.triscv.vector.tuple_nxv4i8_8t.p0.i64.nxv1i1(target("riscv.vector.tuple", <vscale x 4 x i8>, 8) [[VAL:%.*]], ptr [[BASE:%.*]], i64 [[OFFSET:%.*]], <vscale x 1 x i1> [[MASK:%.*]], i64 [[VL:%.*]], i64 5)
+; CHECK-NEXT:    ret void
+;
+entry:
+  tail call void @llvm.riscv.vssseg8.mask.triscv.vector.tuple_nxv4i8_8t.nxv1i1(target("riscv.vector.tuple", <vscale x 4 x i8>, 8) %val, ptr %base, i64 %offset, <vscale x 1 x i1> %mask, i64 %vl, i64 5)
+  ret void
+}
+
+
+; Test stride value is a multiple of pointer alignment.
+define <vscale x 1 x i32> @intrinsic_vlse_v_nxv1i32_nxv1i32_align(<vscale x 1 x i32>* align 4 %0, i64 %1, i64 %2) sanitize_address {
+; CHECK-LABEL: @intrinsic_vlse_v_nxv1i32_nxv1i32_align(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP3:%.*]] = load i64, ptr @__asan_shadow_memory_dynamic_address, align 8
+; CHECK-NEXT:    [[TMP4:%.*]] = icmp ne i64 [[TMP2:%.*]], 0
+; CHECK-NEXT:    br i1 [[TMP4]], label [[TMP5:%.*]], label [[TMP14:%.*]]
+; CHECK:       5:
+; CHECK-NEXT:    [[TMP6:%.*]] = call i64 @llvm.vscale.i64()
+; CHECK-NEXT:    [[TMP7:%.*]] = call i64 @llvm.umin.i64(i64 [[TMP2]], i64 [[TMP6]])
+; CHECK-NEXT:    br label [[DOTSPLIT:%.*]]
+; CHECK:       .split:
+; CHECK-NEXT:    [[IV:%.*]] = phi i64 [ 0, [[TMP5]] ], [ [[IV_NEXT:%.*]], [[TMP13:%.*]] ]
+; CHECK-NEXT:    [[TMP8:%.*]] = extractelement <vscale x 1 x i1> splat (i1 true), i64 [[IV]]
+; CHECK-NEXT:    br i1 [[TMP8]], label [[TMP9:%.*]], label [[TMP13]]
+; CHECK:       9:
+; CHECK-NEXT:    [[TMP10:%.*]] = mul i64 [[IV]], 4
+; CHECK-NEXT:    [[TMP11:%.*]] = getelementptr i8, ptr [[TMP0:%.*]], i64 [[TMP10]]
+; CHECK-NEXT:    [[TMP12:%.*]] = ptrtoint ptr [[TMP11]] to i64
+; CHECK-NEXT:    call void @__asan_load4(i64 [[TMP12]])
+; CHECK-NEXT:    br label [[TMP13]]
+; CHECK:       13:
+; CHECK-NEXT:    [[IV_NEXT]] = add nuw nsw i64 [[IV]], 1
+; CHECK-NEXT:    [[IV_CHECK:%.*]] = icmp eq i64 [[IV_NEXT]], [[TMP7]]
+; CHECK-NEXT:    br i1 [[IV_CHECK]], label [[DOTSPLIT_SPLIT:%.*]], label [[DOTSPLIT]]
+; CHECK:       .split.split:
+; CHECK-NEXT:    br label [[TMP14]]
+; CHECK:       14:
+; CHECK-NEXT:    [[A:%.*]] = call <vscale x 1 x i32> @llvm.riscv.vlse.nxv1i32.p0.i64(<vscale x 1 x i32> undef, ptr [[TMP0]], i64 4, i64 [[TMP2]])
+; CHECK-NEXT:    ret <vscale x 1 x i32> [[A]]
+;
+entry:
+  %a = call <vscale x 1 x i32> @llvm.riscv.vlse.nxv1i32(
+  <vscale x 1 x i32> undef,
+  <vscale x 1 x i32>* %0,
+  i64 4,
+  i64 %2)
+
+  ret <vscale x 1 x i32> %a
+}
+
+declare <vscale x 1 x i32> @llvm.riscv.vloxei.nxv1i32.nxv1i16(
+  <vscale x 1 x i32>,
+  <vscale x 1 x i32>*,
+  <vscale x 1 x i16>,
+  i64);
+
+define <vscale x 1 x i32> @intrinsic_vloxei_v_nxv1i32_nxv1i32_nxv1i16(<vscale x 1 x i32>* %0, <vscale x 1 x i16> %1, i64 %2) sanitize_address {
+; CHECK-LABEL: @intrinsic_vloxei_v_nxv1i32_nxv1i32_nxv1i16(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP3:%.*]] = load i64, ptr @__asan_shadow_memory_dynamic_address, align 8
+; CHECK-NEXT:    [[A:%.*]] = call <vscale x 1 x i32> @llvm.riscv.vloxei.nxv1i32.p0.nxv1i16.i64(<vscale x 1 x i32> undef, ptr [[TMP0:%.*]], <vscale x 1 x i16> [[TMP1:%.*]], i64 [[TMP2:%.*]])
+; CHECK-NEXT:    ret <vscale x 1 x i32> [[A]]
+;
+entry:
+  %a = call <vscale x 1 x i32> @llvm.riscv.vloxei.nxv1i32.nxv1i16(
+  <vscale x 1 x i32> undef,
+  <vscale x 1 x i32>* %0,
+  <vscale x 1 x i16> %1,
+  i64 %2)
+
+  ret <vscale x 1 x i32> %a
+}
+
+declare <vscale x 1 x i32> @llvm.riscv.vloxei.mask.nxv1i32.nxv1i16(
+  <vscale x 1 x i32>,
+  <vscale x 1 x i32>*,
+  <vscale x 1 x i16>,
+  <vscale x 1 x i1>,
+  i64,
+  i64);
+
+define <vscale x 1 x i32> @intrinsic_vloxei_mask_v_nxv1i32_nxv1i32_nxv1i16(<vscale x 1 x i32> %0, <vscale x 1 x i32>* %1, <vscale x 1 x i16> %2, <vscale x 1 x i1> %3, i64 %4) sanitize_address {
+; CHECK-LABEL: @intrinsic_vloxei_mask_v_nxv1i32_nxv1i32_nxv1i16(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP5:%.*]] = load i64, ptr @__asan_shadow_memory_dynamic_address, align 8
+; CHECK-NEXT:    [[A:%.*]] = call <vscale x 1 x i32> @llvm.riscv.vloxei.mask.nxv1i32.p0.nxv1i16.i64(<vscale x 1 x i32> [[TMP0:%.*]], ptr [[TMP1:%.*]], <vscale x 1 x i16> [[TMP2:%.*]], <vscale x 1 x i1> [[TMP3:%.*]], i64 [[TMP4:%.*]], i64 1)
+; CHECK-NEXT:    ret <vscale x 1 x i32> [[A]]
+;
+entry:
+  %a = call <vscale x 1 x i32> @llvm.riscv.vloxei.mask.nxv1i32.nxv1i16(
+  <vscale x 1 x i32> %0,
+  <vscale x 1 x i32>* %1,
+  <vscale x 1 x i16> %2,
+  <vscale x 1 x i1> %3,
+  i64 %4, i64 1)
+
+  ret <vscale x 1 x i32> %a
+}
+
+declare <vscale x 1 x float> @llvm.riscv.vloxei.nxv1f32.nxv1i16(
+  <vscale x 1 x float>,
+  <vscale x 1 x float>*,
+  <vscale x 1 x i16>,
+  i64);
+
+define <vscale x 1 x float> @intrinsic_vloxei_v_nxv1f32_nxv1f32_nxv1i16(<vscale x 1 x float>* %0, <vscale x 1 x i16> %1, i64 %2) sanitize_address {
+; CHECK-LABEL: @intrinsic_vloxei_v_nxv1f32_nxv1f32_nxv1i16(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP3:%.*]] = load i64, ptr @__asan_shadow_memory_dynamic_address, align 8
+; CHECK-NEXT:    [[A:%.*]] = call <vscale x 1 x float> @llvm.riscv.vloxei.nxv1f32.p0.nxv1i16.i64(<vscale x 1 x float> undef, ptr [[TMP0:%.*]], <vscale x 1 x i16> [[TMP1:%.*]], i64 [[TMP2:%.*]])
+; CHECK-NEXT:    ret <vscale x 1 x float> [[A]]
+;
+entry:
+  %a = call <vscale x 1 x float> @llvm.riscv.vloxei.nxv1f32.nxv1i16(
+  <vscale x 1 x float> undef,
+  <vscale x 1 x float>* %0,
+  <vscale x 1 x i16> %1,
+  i64 %2)
+
+  ret <vscale x 1 x float> %a
+}
+
+declare <vscale x 1 x i32> @llvm.riscv.vluxei.nxv1i32.nxv1i16(
+  <vscale x 1 x i32>,
+  <vscale x 1 x i32>*,
+  <vscale x 1 x i16>,
+  i64);
+
+define <vscale x 1 x i32> @intrinsic_vluxei_v_nxv1i32_nxv1i32_nxv1i16(<vscale x 1 x i32>* %0, <vscale x 1 x i16> %1, i64 %2) sanitize_address {
+; CHECK-LABEL: @intrinsic_vluxei_v_nxv1i32_nxv1i32_nxv1i16(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP3:%.*]] = load i64, ptr @__asan_shadow_memory_dynamic_address, align 8
+; CHECK-NEXT:    [[A:%.*]] = call <vscale x 1 x i32> @llvm.riscv.vluxei.nxv1i32.p0.nxv1i16.i64(<vscale x 1 x i32> undef, ptr [[TMP0:%.*]], <vscale x 1 x i16> [[TMP1:%.*]], i64 [[TMP2:%.*]])
+; CHECK-NEXT:    ret <vscale x 1 x i32> [[A]]
+;
+entry:
+  %a = call <vscale x 1 x i32> @llvm.riscv.vluxei.nxv1i32.nxv1i16(
+  <vscale x 1 x i32> undef,
+  <vscale x 1 x i32>* %0,
+  <vscale x 1 x i16> %1,
+  i64 %2)
+
+  ret <vscale x 1 x i32> %a
+}
+
+declare <vscale x 1 x i32> @llvm.riscv.vluxei.mask.nxv1i32.nxv1i16(
+  <vscale x 1 x i32>,
+  <vscale x 1 x i32>*,
+  <vscale x 1 x i16>,
+  <vscale x 1 x i1>,
+  i64,
+  i64);
+
+define <vscale x 1 x i32> @intrinsic_vluxei_mask_v_nxv1i32_nxv1i32_nxv1i16(<vscale x 1 x i32> %0, <vscale x 1 x i32>* %1, <vscale x 1 x i16> %2, <vscale x 1 x i1> %3, i64 %4) sanitize_address {
+; CHECK-LABEL: @intrinsic_vluxei_mask_v_nxv1i32_nxv1i32_nxv1i16(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP5:%.*]] = load i64, ptr @__asan_shadow_memory_dynamic_address, align 8
+; CHECK-NEXT:    [[A:%.*]] = call <vscale x 1 x i32> @llvm.riscv.vluxei.mask.nxv1i32.p0.nxv1i16.i64(<vscale x 1 x i32> [[TMP0:%.*]], ptr [[TMP1:%.*]], <vscale x 1 x i16> [[TMP2:%.*]], <vscale x 1 x i1> [[TMP3:%.*]], i64 [[TMP4:%.*]], i64 1)
+; CHECK-NEXT:    ret <vscale x 1 x i32> [[A]]
+;
+entry:
+  %a = call <vscale x 1 x i32> @llvm.riscv.vluxei.mask.nxv1i32.nxv1i16(
+  <vscale x 1 x i32> %0,
+  <vscale x 1 x i32>* %1,
+  <vscale x 1 x i16> %2,
+  <vscale x 1 x i1> %3,
+  i64 %4, i64 1)
+
+  ret <vscale x 1 x i32> %a
+}
+
+declare void @llvm.riscv.vsoxei.nxv1i32.nxv1i16(
+  <vscale x 1 x i32>,
+  <vscale x 1 x i32>*,
+  <vscale x 1 x i16>,
+  i64);
+
+define void @intrinsic_vsoxei_v_nxv1i32_nxv1i32_nxv1i16(<vscale x 1 x i32> %0, <vscale x 1 x i32>* %1, <vscale x 1 x i16> %2, i64 %3) sanitize_address {
+; CHECK-LABEL: @intrinsic_vsoxei_v_nxv1i32_nxv1i32_nxv1i16(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP4:%.*]] = load i64, ptr @__asan_shadow_memory_dynamic_address, align 8
+; CHECK-NEXT:    call void @llvm.riscv.vsoxei.nxv1i32.p0.nxv1i16.i64(<vscale x 1 x i32> [[TMP0:%.*]], ptr [[TMP1:%.*]], <vscale x 1 x i16> [[TMP2:%.*]], i64 [[TMP3:%.*]])
+; CHECK-NEXT:    ret void
+;
+entry:
+  call void @llvm.riscv.vsoxei.nxv1i32.nxv1i16(
+  <vscale x 1 x i32> %0,
+  <vscale x 1 x i32>* %1,
+  <vscale x 1 x i16> %2,
+  i64 %3)
+
+  ret void
+}
+
+declare void @llvm.riscv.vsoxei.mask.nxv1i32.nxv1i16(
+  <vscale x 1 x i32>,
+  <vscale x 1 x i32>*,
+  <vscale x 1 x i16>,
+  <vscale x 1 x i1>,
+  i64);
+
+define void @intrinsic_vsoxei_mask_v_nxv1i32_nxv1i32_nxv1i16(<vscale x 1 x i32> %0, <vscale x 1 x i32>* %1, <vscale x 1 x i16> %2, <vscale x 1 x i1> %3, i64 %4) sanitize_address {
+; CHECK-LABEL: @intrinsic_vsoxei_mask_v_nxv1i32_nxv1i32_nxv1i16(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP5:%.*]] = load i64, ptr @__asan_shadow_memory_dynamic_address, align 8
+; CHECK-NEXT:    call void @llvm.riscv.vsoxei.mask.nxv1i32.p0.nxv1i16.i64(<vscale x 1 x i32> [[TMP0:%.*]], ptr [[TMP1:%.*]], <vscale x 1 x i16> [[TMP2:%.*]], <vscale x 1 x i1> [[TMP3:%.*]], i64 [[TMP4:%.*]])
+; CHECK-NEXT:    ret void
+;
+entry:
+  call void @llvm.riscv.vsoxei.mask.nxv1i32.nxv1i16(
+  <vscale x 1 x i32> %0,
+  <vscale x 1 x i32>* %1,
+  <vscale x 1 x i16> %2,
+  <vscale x 1 x i1> %3,
+  i64 %4)
+
+  ret void
+}
+
+declare void @llvm.riscv.vsuxei.nxv1i32.nxv1i16(
+  <vscale x 1 x i32>,
+  <vscale x 1 x i32>*,
+  <vscale x 1 x i16>,
+  i64);
+
+define void @intrinsic_vsuxei_v_nxv1i32_nxv1i32_nxv1i16(<vscale x 1 x i32> %0, <vscale x 1 x i32>* %1, <vscale x 1 x i16> %2, i64 %3) sanitize_address {
+; CHECK-LABEL: @intrinsic_vsuxei_v_nxv1i32_nxv1i32_nxv1i16(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP4:%.*]] = load i64, ptr @__asan_shadow_memory_dynamic_address, align 8
+; CHECK-NEXT:    call void @llvm.riscv.vsuxei.nxv1i32.p0.nxv1i16.i64(<vscale x 1 x i32> [[TMP0:%.*]], ptr [[TMP1:%.*]], <vscale x 1 x i16> [[TMP2:%.*]], i64 [[TMP3:%.*]])
+; CHECK-NEXT:    ret void
+;
+entry:
+  call void @llvm.riscv.vsuxei.nxv1i32.nxv1i16(
+  <vscale x 1 x i32> %0,
+  <vscale x 1 x i32>* %1,
+  <vscale x 1 x i16> %2,
+  i64 %3)
+
+  ret void
+}
+
+declare void @llvm.riscv.vsuxei.mask.nxv1i32.nxv1i16(
+  <vscale x 1 x i32>,
+  <vscale x 1 x i32>*,
+  <vscale x 1 x i16>,
+  <vscale x 1 x i1>,
+  i64);
+
+define void @intrinsic_vsuxei_mask_v_nxv1i32_nxv1i32_nxv1i16(<vscale x 1 x i32> %0, <vscale x 1 x i32>* %1, <vscale x 1 x i16> %2, <vscale x 1 x i1> %3, i64 %4) sanitize_address {
+; CHECK-LABEL: @intrinsic_vsuxei_mask_v_nxv1i32_nxv1i32_nxv1i16(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP5:%.*]] = load i64, ptr @__asan_shadow_memory_dynamic_address, align 8
+; CHECK-NEXT:    call void @llvm.riscv.vsuxei.mask.nxv1i32.p0.nxv1i16.i64(<vscale x 1 x i32> [[TMP0:%.*]], ptr [[TMP1:%.*]], <vscale x 1 x i16> [[TMP2:%.*]], <vscale x 1 x i1> [[TMP3:%.*]], i64 [[TMP4:%.*]])
+; CHECK-NEXT:    ret void
+;
+entry:
+  call void @llvm.riscv.vsuxei.mask.nxv1i32.nxv1i16(
+  <vscale x 1 x i32> %0,
+  <vscale x 1 x i32>* %1,
+  <vscale x 1 x i16> %2,
+  <vscale x 1 x i1> %3,
+  i64 %4)
+
+  ret void
+}
+
+
+declare target("riscv.vector.tuple", <vscale x 4 x i8>, 2) @llvm.riscv.vloxseg2.triscv.vector.tuple_nxv4i8_2t.nxv4i8(target("riscv.vector.tuple", <vscale x 4 x i8>, 2), ptr, <vscale x 1 x i16>, i64, i64)
+declare target("riscv.vector.tuple", <vscale x 4 x i8>, 2) @llvm.riscv.vloxseg2.mask.triscv.vector.tuple_nxv4i8_2t.nxv1i16.nxv1i1(target("riscv.vector.tuple", <vscale x 4 x i8>, 2), ptr, <vscale x 1 x i16>, <vscale x 1 x i1>, i64, i64, i64)
+
+define <vscale x 1 x i32> @test_vloxseg2_nxv1i32_nxv1i16(ptr %base, <vscale x 1 x i16> %index, i64 %vl) sanitize_address {
+; CHECK-LABEL: @test_vloxseg2_nxv1i32_nxv1i16(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i64, ptr @__asan_shadow_memory_dynamic_address, align 8
+; CHECK-NEXT:    [[TMP25:%.*]] = tail call target("riscv.vector.tuple", <vscale x 4 x i8>, 2) @llvm.riscv.vloxseg2.triscv.vector.tuple_nxv4i8_2t.p0.nxv1i16.i64(target("riscv.vector.tuple", <vscale x 4 x i8>, 2) undef, ptr [[BASE:%.*]], <vscale x 1 x i16> [[INDEX:%.*]], i64 [[VL:%.*]], i64 5)
+; CHECK-NEXT:    [[TMP26:%.*]] = call <vscale x 1 x i32> @llvm.riscv.tuple.extract.nxv1i32.triscv.vector.tuple_nxv4i8_2t(target("riscv.vector.tuple", <vscale x 4 x i8>, 2) [[TMP25]], i32 1)
+; CHECK-NEXT:    ret <vscale x 1 x i32> [[TMP26]]
+;
+entry:
+  %0 = tail call target("riscv.vector.tuple", <vscale x 4 x i8>, 2) @llvm.riscv.vloxseg2.triscv.vector.tuple_nxv4i8_2t.nxv1i16(target("riscv.vector.tuple", <vscale x 4 x i8>, 2) undef, ptr %base, <vscale x 1 x i16> %index, i64 %vl, i64 5)
+  %1 = call <vscale x 1 x i32> @llvm.riscv.tuple.extract.nxv1i32.triscv.vector.tuple_nxv4i8_2t(target("riscv.vector.tuple", <vscale x 4 x i8>, 2) %0, i32 1)
+  ret <vscale x 1 x i32> %1
+}
+
+define <vscale x 1 x i32> @test_vloxseg2_mask_nxv1i32_nxv1i16(ptr %base, <vscale x 1 x i16> %index, i64 %vl, <vscale x 1 x i1> %mask) sanitize_address {
+; CHECK-LABEL: @test_vloxseg2_mask_nxv1i32_nxv1i16(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i64, ptr @__asan_shadow_memory_dynamic_address, align 8
+; CHECK-NEXT:    [[TMP25:%.*]] = tail call target("riscv.vector.tuple", <vscale x 4 x i8>, 2) @llvm.riscv.vloxseg2.mask.triscv.vector.tuple_nxv4i8_2t.p0.nxv1i16.nxv1i1.i64(target("riscv.vector.tuple", <vscale x 4 x i8>, 2) undef, ptr [[BASE:%.*]], <vscale x 1 x i16> [[INDEX:%.*]], <vscale x 1 x i1> [[MASK:%.*]], i64 [[VL:%.*]], i64 1, i64 5)
+; CHECK-NEXT:    [[TMP26:%.*]] = call <vscale x 1 x i32> @llvm.riscv.tuple.extract.nxv1i32.triscv.vector.tuple_nxv4i8_2t(target("riscv.vector.tuple", <vscale x 4 x i8>, 2) [[TMP25]], i32 1)
+; CHECK-NEXT:    ret <vscale x 1 x i32> [[TMP26]]
+;
+entry:
+  %0 = tail call target("riscv.vector.tuple", <vscale x 4 x i8>, 2) @llvm.riscv.vloxseg2.mask.triscv.vector.tuple_nxv4i8_2t.nxv4i1.nxv1i32(target("riscv.vector.tuple", <vscale x 4 x i8>, 2) undef, ptr %base, <vscale x 1 x i16> %index, <vscale x 1 x i1> %mask, i64 %vl, i64 1, i64 5)
+  %1 = call <vscale x 1 x i32> @llvm.riscv.tuple.extract.nxv1i32.triscv.vector.tuple_nxv4i8_2t(target("riscv.vector.tuple", <vscale x 4 x i8>, 2) %0, i32 1)
+  ret <vscale x 1 x i32> %1
+}
+
+
+declare target("riscv.vector.tuple", <vscale x 4 x i8>, 3) @llvm.riscv.vloxseg3.triscv.vector.tuple_nxv4i8_3t.nxv4i8(target("riscv.vector.tuple", <vscale x 4 x i8>, 3), ptr, <vscale x 1 x i16>, i64, i64)
+declare target("riscv.vector.tuple", <vscale x 4 x i8>, 3) @llvm.riscv.vloxseg3.mask.triscv.vector.tuple_nxv4i8_3t.nxv1i16.nxv1i1(target("riscv.vector.tuple", <vscale x 4 x i8>, 3), ptr, <vscale x 1 x i16>, <vscale x 1 x i1>, i64, i64, i64)
+
+define <vscale x 1 x i32> @test_vloxseg3_nxv1i32_nxv1i16(ptr %base, <vscale x 1 x i16> %index, i64 %vl) sanitize_address {
+; CHECK-LABEL: @test_vloxseg3_nxv1i32_nxv1i16(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i64, ptr @__asan_shadow_memory_dynamic_address, align 8
+; CHECK-NEXT:    [[TMP37:%.*]] = tail call target("riscv.vector.tuple", <vscale x 4 x i8>, 3) @llvm.riscv.vloxseg3.triscv.vector.tuple_nxv4i8_3t.p0.nxv1i16.i64(target("riscv.vector.tuple", <vscale x 4 x i8>, 3) undef, ptr [[BASE:%.*]], <vscale x 1 x i16> [[INDEX:%.*]], i64 [[VL:%.*]], i64 5)
+; CHECK-NEXT:    [[TMP38:%.*]] = call <vscale x 1 x i32> @llvm.riscv.tuple.extract.nxv1i32.triscv.vector.tuple_nxv4i8_3t(target("riscv.vector.tuple", <vscale x 4 x i8>, 3) [[TMP37]], i32 1)
+; CHECK-NEXT:    ret <vscale x 1 x i32> [[TMP38]]
+;
+entry:
+  %0 = tail call target("riscv.vector.tuple", <vscale x 4 x i8>, 3) @llvm.riscv.vloxseg3.triscv.vector.tuple_nxv4i8_3t.nxv1i16(target("riscv.vector.tuple", <vscale x 4 x i8>, 3) undef, ptr %base, <vscale x 1 x i16> %index, i64 %vl, i64 5)
+  %1 = call <vscale x 1 x i32> @llvm.riscv.tuple.extract.nxv1i32.triscv.vector.tuple_nxv4i8_3t(target("riscv.vector.tuple", <vscale x 4 x i8>, 3) %0, i32 1)
+  ret <vscale x 1 x i32> %1
+}
+
+define <vscale x 1 x i32> @test_vloxseg3_mask_nxv1i32_nxv1i16(ptr %base, <vscale x 1 x i16> %index, i64 %vl, <vscale x 1 x i1> %mask) sanitize_address {
+; CHECK-LABEL: @test_vloxseg3_mask_nxv1i32_nxv1i16(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i64, ptr @__asan_shadow_memory_dynamic_address, align 8
+; CHECK-NEXT:    [[TMP37:%.*]] = tail call target("riscv.vector.tuple", <vscale x 4 x i8>, 3) @llvm.riscv.vloxseg3.mask.triscv.vector.tuple_nxv4i8_3t.p0.nxv1i16.nxv1i1.i64(target("riscv.vector.tuple", <vscale x 4 x i8>, 3) undef, ptr [[BASE:%.*]], <vscale x 1 x i16> [[INDEX:%.*]], <vscale x 1 x i1> [[MASK:%.*]], i64 [[VL:%.*]], i64 1, i64 5)
+; CHECK-NEXT:    [[TMP38:%.*]] = call <vscale x 1 x i32> @llvm.riscv.tuple.extract.nxv1i32.triscv.vector.tuple_nxv4i8_3t(target("riscv.vector.tuple", <vscale x 4 x i8>, 3) [[TMP37]], i32 1)
+; CHECK-NEXT:    ret <vscale x 1 x i32> [[TMP38]]
+;
+entry:
+  %0 = tail call target("riscv.vector.tuple", <vscale x 4 x i8>, 3) @llvm.riscv.vloxseg3.mask.triscv.vector.tuple_nxv4i8_3t.nxv4i1.nxv1i32(target("riscv.vector.tuple", <vscale x 4 x i8>, 3) undef, ptr %base, <vscale x 1 x i16> %index, <vscale x 1 x i1> %mask, i64 %vl, i64 1, i64 5)
+  %1 = call <vscale x 1 x i32> @llvm.riscv.tuple.extract.nxv1i32.triscv.vector.tuple_nxv4i8_3t(target("riscv.vector.tuple", <vscale x 4 x i8>, 3) %0, i32 1)
+  ret <vscale x 1 x i32> %1
+}
+
+
+declare target("riscv.vector.tuple", <vscale x 4 x i8>, 4) @llvm.riscv.vloxseg4.triscv.vector.tuple_nxv4i8_4t.nxv4i8(target("riscv.vector.tuple", <vscale x 4 x i8>, 4), ptr, <vscale x 1 x i16>, i64, i64)
+declare target("riscv.vector.tuple", <vscale x 4 x i8>, 4) @llvm.riscv.vloxseg4.mask.triscv.vector.tuple_nxv4i8_4t.nxv1i16.nxv1i1(target("riscv.vector.tuple", <vscale x 4 x i8>, 4), ptr, <vscale x 1 x i16>, <vscale x 1 x i1>, i64, i64, i64)
+
+define <vscale x 1 x i32> @test_vloxseg4_nxv1i32_nxv1i16(ptr %base, <vscale x 1 x i16> %index, i64 %vl) sanitize_address {
+; CHECK-LABEL: @test_vloxseg4_nxv1i32_nxv1i16(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i64, ptr @__asan_shadow_memory_dynamic_address, align 8
+; CHECK-NEXT:    [[TMP49:%.*]] = tail call target("riscv.vector.tuple", <vscale x 4 x i8>, 4) @llvm.riscv.vloxseg4.triscv.vector.tuple_nxv4i8_4t.p0.nxv1i16.i64(target("riscv.vector.tuple", <vscale x 4 x i8>, 4) undef, ptr [[BASE:%.*]], <vscale x 1 x i16> [[INDEX:%.*]], i64 [[VL:%.*]], i64 5)
+; CHECK-NEXT:    [[TMP50:%.*]] = call <vscale x 1 x i32> @llvm.riscv.tuple.extract.nxv1i32.triscv.vector.tuple_nxv4i8_4t(target("riscv.vector.tuple", <vscale x 4 x i8>, 4) [[TMP49]], i32 1)
+; CHECK-NEXT:    ret <vscale x 1 x i32> [[TMP50]]
+;
+entry:
+  %0 = tail call target("riscv.vector.tuple", <vscale x 4 x i8>, 4) @llvm.riscv.vloxseg4.triscv.vector.tuple_nxv4i8_4t.nxv1i16(target("riscv.vector.tuple", <vscale x 4 x i8>, 4) undef, ptr %base, <vscale x 1 x i16> %index, i64 %vl, i64 5)
+  %1 = call <vscale x 1 x i32> @llvm.riscv.tuple.extract.nxv1i32.triscv.vector.tuple_nxv4i8_4t(target("riscv.vector.tuple", <vscale x 4 x i8>, 4) %0, i32 1)
+  ret <vscale x 1 x i32> %1
+}
+
+define <vscale x 1 x i32> @test_vloxseg4_mask_nxv1i32_nxv1i16(ptr %base, <vscale x 1 x i16> %index, i64 %vl, <vscale x 1 x i1> %mask) sanitize_address {
+; CHECK-LABEL: @test_vloxseg4_mask_nxv1i32_nxv1i16(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i64, ptr @__asan_shadow_memory_dynamic_address, align 8
+; CHECK-NEXT:    [[TMP49:%.*]] = tail call target("riscv.vector.tuple", <vscale x 4 x i8>, 4) @llvm.riscv.vloxseg4.mask.triscv.vector.tuple_nxv4i8_4t.p0.nxv1i16.nxv1i1.i64(target("riscv.vector.tuple", <vscale x 4 x i8>, 4) undef, ptr [[BASE:%.*]], <vscale x 1 x i16> [[INDEX:%.*]], <vscale x 1 x i1> [[MASK:%.*]], i64 [[VL:%.*]], i64 1, i64 5)
+; CHECK-NEXT:    [[TMP50:%.*]] = call <vscale x 1 x i32> @llvm.riscv.tuple.extract.nxv1i32.triscv.vector.tuple_nxv4i8_4t(target("riscv.vector.tuple", <vscale x 4 x i8>, 4) [[TMP49]], i32 1)
+; CHECK-NEXT:    ret <vscale x 1 x i32> [[TMP50]]
+;
+entry:
+  %0 = tail call target("riscv.vector.tuple", <vscale x 4 x i8>, 4) @llvm.riscv.vloxseg4.mask.triscv.vector.tuple_nxv4i8_4t.nxv4i1.nxv1i32(target("riscv.vector.tuple", <vscale x 4 x i8>, 4) undef, ptr %base, <vscale x 1 x i16> %index, <vscale x 1 x i1> %mask, i64 %vl, i64 1, i64 5)
+  %1 = call <vscale x 1 x i32> @llvm.riscv.tuple.extract.nxv1i32.triscv.vector.tuple_nxv4i8_4t(target("riscv.vector.tuple", <vscale x 4 x i8>, 4) %0, i32 1)
+  ret <vscale x 1 x i32> %1
+}
+
+
+declare target("riscv.vector.tuple", <vscale x 4 x i8>, 5) @llvm.riscv.vloxseg5.triscv.vector.tuple_nxv4i8_5t.nxv4i8(target("riscv.vector.tuple", <vscale x 4 x i8>, 5), ptr, <vscale x 1 x i16>, i64, i64)
+declare target("riscv.vector.tuple", <vscale x 4 x i8>, 5) @llvm.riscv.vloxseg5.mask.triscv.vector.tuple_nxv4i8_5t.nxv1i16.nxv1i1(target("riscv.vector.tuple", <vscale x 4 x i8>, 5), ptr, <vscale x 1 x i16>, <vscale x 1 x i1>, i64, i64, i64)
+
+define <vscale x 1 x i32> @test_vloxseg5_nxv1i32_nxv1i16(ptr %base, <vscale x 1 x i16> %index, i64 %vl) sanitize_address {
+; CHECK-LABEL: @test_vloxseg5_nxv1i32_nxv1i16(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i64, ptr @__asan_shadow_memory_dynamic_address, align 8
+; CHECK-NEXT:    [[TMP61:%.*]] = tail call target("riscv.vector.tuple", <vscale x 4 x i8>, 5) @llvm.riscv.vloxseg5.triscv.vector.tuple_nxv4i8_5t.p0.nxv1i16.i64(target("riscv.vector.tuple", <vscale x 4 x i8>, 5) undef, ptr [[BASE:%.*]], <vscale x 1 x i16> [[INDEX:%.*]], i64 [[VL:%.*]], i64 5)
+; CHECK-NEXT:    [[TMP62:%.*]] = call <vscale x 1 x i32> @llvm.riscv.tuple.extract.nxv1i32.triscv.vector.tuple_nxv4i8_5t(target("riscv.vector.tuple", <vscale x 4 x i8>, 5) [[TMP61]], i32 1)
+; CHECK-NEXT:    ret <vscale x 1 x i32> [[TMP62]]
+;
+entry:
+  %0 = tail call target("riscv.vector.tuple", <vscale x 4 x i8>, 5) @llvm.riscv.vloxseg5.triscv.vector.tuple_nxv4i8_5t.nxv1i16(target("riscv.vector.tuple", <vscale x 4 x i8>, 5) undef, ptr %base, <vscale x 1 x i16> %index, i64 %vl, i64 5)
+  %1 = call <vscale x 1 x i32> @llvm.riscv.tuple.extract.nxv1i32.triscv.vector.tuple_nxv4i8_5t(target("riscv.vector.tuple", <vscale x 4 x i8>, 5) %0, i32 1)
+  ret <vscale x 1 x i32> %1
+}
+
+define <vscale x 1 x i32> @test_vloxseg5_mask_nxv1i32_nxv1i16(ptr %base, <vscale x 1 x i16> %index, i64 %vl, <vscale x 1 x i1> %mask) sanitize_address {
+; CHECK-LABEL: @test_vloxseg5_mask_nxv1i32_nxv1i16(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i64, ptr @__asan_shadow_memory_dynamic_address, align 8
+; CHECK-NEXT:    [[TMP61:%.*]] = tail call target("riscv.vector.tuple", <vscale x 4 x i8>, 5) @llvm.riscv.vloxseg5.mask.triscv.vector.tuple_nxv4i8_5t.p0.nxv1i16.nxv1i1.i64(target("riscv.vector.tuple", <vscale x 4 x i8>, 5) undef, ptr [[BASE:%.*]], <vscale x 1 x i16> [[INDEX:%.*]], <vscale x 1 x i1> [[MASK:%.*]], i64 [[VL:%.*]], i64 1, i64 5)
+; CHECK-NEXT:    [[TMP62:%.*]] = call <vscale x 1 x i32> @llvm.riscv.tuple.extract.nxv1i32.triscv.vector.tuple_nxv4i8_5t(target("riscv.vector.tuple", <vscale x 4 x i8>, 5) [[TMP61]], i32 1)
+; CHECK-NEXT:    ret <vscale x 1 x i32> [[TMP62]]
+;
+entry:
+  %0 = tail call target("riscv.vector.tuple", <vscale x 4 x i8>, 5) @llvm.riscv.vloxseg5.mask.triscv.vector.tuple_nxv4i8_5t.nxv4i1.nxv1i32(target("riscv.vector.tuple", <vscale x 4 x i8>, 5) undef, ptr %base, <vscale x 1 x i16> %index, <vscale x 1 x i1> %mask, i64 %vl, i64 1, i64 5)
+  %1 = call <vscale x 1 x i32> @llvm.riscv.tuple.extract.nxv1i32.triscv.vector.tuple_nxv4i8_5t(target("riscv.vector.tuple", <vscale x 4 x i8>, 5) %0, i32 1)
+  ret <vscale x 1 x i32> %1
+}
+
+
+declare target("riscv.vector.tuple", <vscale x 4 x i8>, 6) @llvm.riscv.vloxseg6.triscv.vector.tuple_nxv4i8_6t.nxv4i8(target("riscv.vector.tuple", <vscale x 4 x i8>, 6), ptr, <vscale x 1 x i16>, i64, i64)
+declare target("riscv.vector.tuple", <vscale x 4 x i8>, 6) @llvm.riscv.vloxseg6.mask.triscv.vector.tuple_nxv4i8_6t.nxv1i16.nxv1i1(target("riscv.vector.tuple", <vscale x 4 x i8>, 6), ptr, <vscale x 1 x i16>, <vscale x 1 x i1>, i64, i64, i64)
+
+define <vscale x 1 x i32> @test_vloxseg6_nxv1i32_nxv1i16(ptr %base, <vscale x 1 x i16> %index, i64 %vl) sanitize_address {
+; CHECK-LABEL: @test_vloxseg6_nxv1i32_nxv1i16(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i64, ptr @__asan_shadow_memory_dynamic_address, align 8
+; CHECK-NEXT:    [[TMP73:%.*]] = tail call target("riscv.vector.tuple", <vscale x 4 x i8>, 6) @llvm.riscv.vloxseg6.triscv.vector.tuple_nxv4i8_6t.p0.nxv1i16.i64(target("riscv.vector.tuple", <vscale x 4 x i8>, 6) undef, ptr [[BASE:%.*]], <vscale x 1 x i16> [[INDEX:%.*]], i64 [[VL:%.*]], i64 5)
+; CHECK-NEXT:    [[TMP74:%.*]] = call <vscale x 1 x i32> @llvm.riscv.tuple.extract.nxv1i32.triscv.vector.tuple_nxv4i8_6t(target("riscv.vector.tuple", <vscale x 4 x i8>, 6) [[TMP73]], i32 1)
+; CHECK-NEXT:    ret <vscale x 1 x i32> [[TMP74]]
+;
+entry:
+  %0 = tail call target("riscv.vector.tuple", <vscale x 4 x i8>, 6) @llvm.riscv.vloxseg6.triscv.vector.tuple_nxv4i8_6t.nxv1i16(target("riscv.vector.tuple", <vscale x 4 x i8>, 6) undef, ptr %base, <vscale x 1 x i16> %index, i64 %vl, i64 5)
+  %1 = call <vscale x 1 x i32> @llvm.riscv.tuple.extract.nxv1i32.triscv.vector.tuple_nxv4i8_6t(target("riscv.vector.tuple", <vscale x 4 x i8>, 6) %0, i32 1)
+  ret <vscale x 1 x i32> %1
+}
+
+define <vscale x 1 x i32> @test_vloxseg6_mask_nxv1i32_nxv1i16(ptr %base, <vscale x 1 x i16> %index, i64 %vl, <vscale x 1 x i1> %mask) sanitize_address {
+; CHECK-LABEL: @test_vloxseg6_mask_nxv1i32_nxv1i16(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i64, ptr @__asan_shadow_memory_dynamic_address, align 8
+; CHECK-NEXT:    [[TMP73:%.*]] = tail call target("riscv.vector.tuple", <vscale x 4 x i8>, 6) @llvm.riscv.vloxseg6.mask.triscv.vector.tuple_nxv4i8_6t.p0.nxv1i16.nxv1i1.i64(target("riscv.vector.tuple", <vscale x 4 x i8>, 6) undef, ptr [[BASE:%.*]], <vscale x 1 x i16> [[INDEX:%.*]], <vscale x 1 x i1> [[MASK:%.*]], i64 [[VL:%.*]], i64 1, i64 5)
+; CHECK-NEXT:    [[TMP74:%.*]] = call <vscale x 1 x i32> @llvm.riscv.tuple.extract.nxv1i32.triscv.vector.tuple_nxv4i8_6t(target("riscv.vector.tuple", <vscale x 4 x i8>, 6) [[TMP73]], i32 1)
+; CHECK-NEXT:    ret <vscale x 1 x i32> [[TMP74]]
+;
+entry:
+  %0 = tail call target("riscv.vector.tuple", <vscale x 4 x i8>, 6) @llvm.riscv.vloxseg6.mask.triscv.vector.tuple_nxv4i8_6t.nxv4i1.nxv1i32(target("riscv.vector.tuple", <vscale x 4 x i8>, 6) undef, ptr %base, <vscale x 1 x i16> %index, <vscale x 1 x i1> %mask, i64 %vl, i64 1, i64 5)
+  %1 = call <vscale x 1 x i32> @llvm.riscv.tuple.extract.nxv1i32.triscv.vector.tuple_nxv4i8_6t(target("riscv.vector.tuple", <vscale x 4 x i8>, 6) %0, i32 1)
+  ret <vscale x 1 x i32> %1
+}
+
+
+declare target("riscv.vector.tuple", <vscale x 4 x i8>, 7) @llvm.riscv.vloxseg7.triscv.vector.tuple_nxv4i8_7t.nxv4i8(target("riscv.vector.tuple", <vscale x 4 x i8>, 7), ptr, <vscale x 1 x i16>, i64, i64)
+declare target("riscv.vector.tuple", <vscale x 4 x i8>, 7) @llvm.riscv.vloxseg7.mask.triscv.vector.tuple_nxv4i8_7t.nxv1i16.nxv1i1(target("riscv.vector.tuple", <vscale x 4 x i8>, 7), ptr, <vscale x 1 x i16>, <vscale x 1 x i1>, i64, i64, i64)
+
+define <vscale x 1 x i32> @test_vloxseg7_nxv1i32_nxv1i16(ptr %base, <vscale x 1 x i16> %index, i64 %vl) sanitize_address {
+; CHECK-LABEL: @test_vloxseg7_nxv1i32_nxv1i16(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i64, ptr @__asan_shadow_memory_dynamic_address, align 8
+; CHECK-NEXT:    [[TMP85:%.*]] = tail call target("riscv.vector.tuple", <vscale x 4 x i8>, 7) @llvm.riscv.vloxseg7.triscv.vector.tuple_nxv4i8_7t.p0.nxv1i16.i64(target("riscv.vector.tuple", <vscale x 4 x i8>, 7) undef, ptr [[BASE:%.*]], <vscale x 1 x i16> [[INDEX:%.*]], i64 [[VL:%.*]], i64 5)
+; CHECK-NEXT:    [[TMP86:%.*]] = call <vscale x 1 x i32> @llvm.riscv.tuple.extract.nxv1i32.triscv.vector.tuple_nxv4i8_7t(target("riscv.vector.tuple", <vscale x 4 x i8>, 7) [[TMP85]], i32 1)
+; CHECK-NEXT:    ret <vscale x 1 x i32> [[TMP86]]
+;
+entry:
+  %0 = tail call target("riscv.vector.tuple", <vscale x 4 x i8>, 7) @llvm.riscv.vloxseg7.triscv.vector.tuple_nxv4i8_7t.nxv1i16(target("riscv.vector.tuple", <vscale x 4 x i8>, 7) undef, ptr %base, <vscale x 1 x i16> %index, i64 %vl, i64 5)
+  %1 = call <vscale x 1 x i32> @llvm.riscv.tuple.extract.nxv1i32.triscv.vector.tuple_nxv4i8_7t(target("riscv.vector.tuple", <vscale x 4 x i8>, 7) %0, i32 1)
+  ret <vscale x 1 x i32> %1
+}
+
+define <vscale x 1 x i32> @test_vloxseg7_mask_nxv1i32_nxv1i16(ptr %base, <vscale x 1 x i16> %index, i64 %vl, <vscale x 1 x i1> %mask) sanitize_address {
+; CHECK-LABEL: @test_vloxseg7_mask_nxv1i32_nxv1i16(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i64, ptr @__asan_shadow_memory_dynamic_address, align 8
+; CHECK-NEXT:    [[TMP85:%.*]] = tail call target("riscv.vector.tuple", <vscale x 4 x i8>, 7) @llvm.riscv.vloxseg7.mask.triscv.vector.tuple_nxv4i8_7t.p0.nxv1i16.nxv1i1.i64(target("riscv.vector.tuple", <vscale x 4 x i8>, 7) undef, ptr [[BASE:%.*]], <vscale x 1 x i16> [[INDEX:%.*]], <vscale x 1 x i1> [[MASK:%.*]], i64 [[VL:%.*]], i64 1, i64 5)
+; CHECK-NEXT:    [[TMP86:%.*]] = call <vscale x 1 x i32> @llvm.riscv.tuple.extract.nxv1i32.triscv.vector.tuple_nxv4i8_7t(target("riscv.vector.tuple", <vscale x 4 x i8>, 7) [[TMP85]], i32 1)
+; CHECK-NEXT:    ret <vscale x 1 x i32> [[TMP86]]
+;
+entry:
+  %0 = tail call target("riscv.vector.tuple", <vscale x 4 x i8>, 7) @llvm.riscv.vloxseg7.mask.triscv.vector.tuple_nxv4i8_7t.nxv4i1.nxv1i32(target("riscv.vector.tuple", <vscale x 4 x i8>, 7) undef, ptr %base, <vscale x 1 x i16> %index, <vscale x 1 x i1> %mask, i64 %vl, i64 1, i64 5)
+  %1 = call <vscale x 1 x i32> @llvm.riscv.tuple.extract.nxv1i32.triscv.vector.tuple_nxv4i8_7t(target("riscv.vector.tuple", <vscale x 4 x i8>, 7) %0, i32 1)
+  ret <vscale x 1 x i32> %1
+}
+
+
+declare target("riscv.vector.tuple", <vscale x 4 x i8>, 8) @llvm.riscv.vloxseg8.triscv.vector.tuple_nxv4i8_8t.nxv4i8(target("riscv.vector.tuple", <vscale x 4 x i8>, 8), ptr, <vscale x 1 x i16>, i64, i64)
+declare target("riscv.vector.tuple", <vscale x 4 x i8>, 8) @llvm.riscv.vloxseg8.mask.triscv.vector.tuple_nxv4i8_8t.nxv1i16.nxv1i1(target("riscv.vector.tuple", <vscale x 4 x i8>, 8), ptr, <vscale x 1 x i16>, <vscale x 1 x i1>, i64, i64, i64)
+
+define <vscale x 1 x i32> @test_vloxseg8_nxv1i32_nxv1i16(ptr %base, <vscale x 1 x i16> %index, i64 %vl) sanitize_address {
+; CHECK-LABEL: @test_vloxseg8_nxv1i32_nxv1i16(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i64, ptr @__asan_shadow_memory_dynamic_address, align 8
+; CHECK-NEXT:    [[TMP97:%.*]] = tail call target("riscv.vector.tuple", <vscale x 4 x i8>, 8) @llvm.riscv.vloxseg8.triscv.vector.tuple_nxv4i8_8t.p0.nxv1i16.i64(target("riscv.vector.tuple", <vscale x 4 x i8>, 8) undef, ptr [[BASE:%.*]], <vscale x 1 x i16> [[INDEX:%.*]], i64 [[VL:%.*]], i64 5)
+; CHECK-NEXT:    [[TMP98:%.*]] = call <vscale x 1 x i32> @llvm.riscv.tuple.extract.nxv1i32.triscv.vector.tuple_nxv4i8_8t(target("riscv.vector.tuple", <vscale x 4 x i8>, 8) [[TMP97]], i32 1)
+; CHECK-NEXT:    ret <vscale x 1 x i32> [[TMP98]]
+;
+entry:
+  %0 = tail call target("riscv.vector.tuple", <vscale x 4 x i8>, 8) @llvm.riscv.vloxseg8.triscv.vector.tuple_nxv4i8_8t.nxv1i16(target("riscv.vector.tuple", <vscale x 4 x i8>, 8) undef, ptr %base, <vscale x 1 x i16> %index, i64 %vl, i64 5)
+  %1 = call <vscale x 1 x i32> @llvm.riscv.tuple.extract.nxv1i32.triscv.vector.tuple_nxv4i8_8t(target("riscv.vector.tuple", <vscale x 4 x i8>, 8) %0, i32 1)
+  ret <vscale x 1 x i32> %1
+}
+
+define <vscale x 1 x i32> @test_vloxseg8_mask_nxv1i32_nxv1i16(ptr %base, <vscale x 1 x i16> %index, i64 %vl, <vscale x 1 x i1> %mask) sanitize_address {
+; CHECK-LABEL: @test_vloxseg8_mask_nxv1i32_nxv1i16(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i64, ptr @__asan_shadow_memory_dynamic_address, align 8
+; CHECK-NEXT:    [[TMP97:%.*]] = tail call target("riscv.vector.tuple", <vscale x 4 x i8>, 8) @llvm.riscv.vloxseg8.mask.triscv.vector.tuple_nxv4i8_8t.p0.nxv1i16.nxv1i1.i64(target("riscv.vector.tuple", <vscale x 4 x i8>, 8) undef, ptr [[BASE:%.*]], <vscale x 1 x i16> [[INDEX:%.*]], <vscale x 1 x i1> [[MASK:%.*]], i64 [[VL:%.*]], i64 1, i64 5)
+; CHECK-NEXT:    [[TMP98:%.*]] = call <vscale x 1 x i32> @llvm.riscv.tuple.extract.nxv1i32.triscv.vector.tuple_nxv4i8_8t(target("riscv.vector.tuple", <vscale x 4 x i8>, 8) [[TMP97]], i32 1)
+; CHECK-NEXT:    ret <vscale x 1 x i32> [[TMP98]]
+;
+entry:
+  %0 = tail call target("riscv.vector.tuple", <vscale x 4 x i8>, 8) @llvm.riscv.vloxseg8.mask.triscv.vector.tuple_nxv4i8_8t.nxv4i1.nxv1i32(target("riscv.vector.tuple", <vscale x 4 x i8>, 8) undef, ptr %base, <vscale x 1 x i16> %index, <vscale x 1 x i1> %mask, i64 %vl, i64 1, i64 5)
+  %1 = call <vscale x 1 x i32> @llvm.riscv.tuple.extract.nxv1i32.triscv.vector.tuple_nxv4i8_8t(target("riscv.vector.tuple", <vscale x 4 x i8>, 8) %0, i32 1)
+  ret <vscale x 1 x i32> %1
+}
+
+
+declare target("riscv.vector.tuple", <vscale x 4 x i8>, 2) @llvm.riscv.vluxseg2.triscv.vector.tuple_nxv4i8_2t.nxv4i8(target("riscv.vector.tuple", <vscale x 4 x i8>, 2), ptr, <vscale x 1 x i16>, i64, i64)
+declare target("riscv.vector.tuple", <vscale x 4 x i8>, 2) @llvm.riscv.vluxseg2.mask.triscv.vector.tuple_nxv4i8_2t.nxv1i16.nxv1i1(target("riscv.vector.tuple", <vscale x 4 x i8>, 2), ptr, <vscale x 1 x i16>, <vscale x 1 x i1>, i64, i64, i64)
+
+define <vscale x 1 x i32> @test_vluxseg2_nxv1i32_nxv1i16(ptr %base, <vscale x 1 x i16> %index, i64 %vl) sanitize_address {
+; CHECK-LABEL: @test_vluxseg2_nxv1i32_nxv1i16(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i64, ptr @__asan_shadow_memory_dynamic_address, align 8
+; CHECK-NEXT:    [[TMP25:%.*]] = tail call target("riscv.vector.tuple", <vscale x 4 x i8>, 2) @llvm.riscv.vluxseg2.triscv.vector.tuple_nxv4i8_2t.p0.nxv1i16.i64(target("riscv.vector.tuple", <vscale x 4 x i8>, 2) undef, ptr [[BASE:%.*]], <vscale x 1 x i16> [[INDEX:%.*]], i64 [[VL:%.*]], i64 5)
+; CHECK-NEXT:    [[TMP26:%.*]] = call <vscale x 1 x i32> @llvm.riscv.tuple.extract.nxv1i32.triscv.vector.tuple_nxv4i8_2t(target("riscv.vector.tuple", <vscale x 4 x i8>, 2) [[TMP25]], i32 1)
+; CHECK-NEXT:    ret <vscale x 1 x i32> [[TMP26]]
+;
+entry:
+  %0 = tail call target("riscv.vector.tuple", <vscale x 4 x i8>, 2) @llvm.riscv.vluxseg2.triscv.vector.tuple_nxv4i8_2t.nxv1i16(target("riscv.vector.tuple", <vscale x 4 x i8>, 2) undef, ptr %base, <vscale x 1 x i16> %index, i64 %vl, i64 5)
+  %1 = call <vscale x 1 x i32> @llvm.riscv.tuple.extract.nxv1i32.triscv.vector.tuple_nxv4i8_2t(target("riscv.vector.tuple", <vscale x 4 x i8>, 2) %0, i32 1)
+  ret <vscale x 1 x i32> %1
+}
+
+define <vscale x 1 x i32> @test_vluxseg2_mask_nxv1i32_nxv1i16(ptr %base, <vscale x 1 x i16> %index, i64 %vl, <vscale x 1 x i1> %mask) sanitize_address {
+; CHECK-LABEL: @test_vluxseg2_mask_nxv1i32_nxv1i16(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i64, ptr @__asan_shadow_memory_dynamic_address, align 8
+; CHECK-NEXT:    [[TMP25:%.*]] = tail call target("riscv.vector.tuple", <vscale x 4 x i8>, 2) @llvm.riscv.vluxseg2.mask.triscv.vector.tuple_nxv4i8_2t.p0.nxv1i16.nxv1i1.i64(target("riscv.vector.tuple", <vscale x 4 x i8>, 2) undef, ptr [[BASE:%.*]], <vscale x 1 x i16> [[INDEX:%.*]], <vscale x 1 x i1> [[MASK:%.*]], i64 [[VL:%.*]], i64 1, i64 5)
+; CHECK-NEXT:    [[TMP26:%.*]] = call <vscale x 1 x i32> @llvm.riscv.tuple.extract.nxv1i32.triscv.vector.tuple_nxv4i8_2t(target("riscv.vector.tuple", <vscale x 4 x i8>, 2) [[TMP25]], i32 1)
+; CHECK-NEXT:    ret <vscale x 1 x i32> [[TMP26]]
+;
+entry:
+  %0 = tail call target("riscv.vector.tuple", <vscale x 4 x i8>, 2) @llvm.riscv.vluxseg2.mask.triscv.vector.tuple_nxv4i8_2t.nxv4i1.nxv1i32(target("riscv.vector.tuple", <vscale x 4 x i8>, 2) undef, ptr %base, <vscale x 1 x i16> %index, <vscale x 1 x i1> %mask, i64 %vl, i64 1, i64 5)
+  %1 = call <vscale x 1 x i32> @llvm.riscv.tuple.extract.nxv1i32.triscv.vector.tuple_nxv4i8_2t(target("riscv.vector.tuple", <vscale x 4 x i8>, 2) %0, i32 1)
+  ret <vscale x 1 x i32> %1
+}
+
+
+declare target("riscv.vector.tuple", <vscale x 4 x i8>, 3) @llvm.riscv.vluxseg3.triscv.vector.tuple_nxv4i8_3t.nxv4i8(target("riscv.vector.tuple", <vscale x 4 x i8>, 3), ptr, <vscale x 1 x i16>, i64, i64)
+declare target("riscv.vector.tuple", <vscale x 4 x i8>, 3) @llvm.riscv.vluxseg3.mask.triscv.vector.tuple_nxv4i8_3t.nxv1i16.nxv1i1(target("riscv.vector.tuple", <vscale x 4 x i8>, 3), ptr, <vscale x 1 x i16>, <vscale x 1 x i1>, i64, i64, i64)
+
+define <vscale x 1 x i32> @test_vluxseg3_nxv1i32_nxv1i16(ptr %base, <vscale x 1 x i16> %index, i64 %vl) sanitize_address {
+; CHECK-LABEL: @test_vluxseg3_nxv1i32_nxv1i16(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i64, ptr @__asan_shadow_memory_dynamic_address, align 8
+; CHECK-NEXT:    [[TMP37:%.*]] = tail call target("riscv.vector.tuple", <vscale x 4 x i8>, 3) @llvm.riscv.vluxseg3.triscv.vector.tuple_nxv4i8_3t.p0.nxv1i16.i64(target("riscv.vector.tuple", <vscale x 4 x i8>, 3) undef, ptr [[BASE:%.*]], <vscale x 1 x i16> [[INDEX:%.*]], i64 [[VL:%.*]], i64 5)
+; CHECK-NEXT:    [[TMP38:%.*]] = call <vscale x 1 x i32> @llvm.riscv.tuple.extract.nxv1i32.triscv.vector.tuple_nxv4i8_3t(target("riscv.vector.tuple", <vscale x 4 x i8>, 3) [[TMP37]], i32 1)
+; CHECK-NEXT:    ret <vscale x 1 x i32> [[TMP38]]
+;
+entry:
+  %0 = tail call target("riscv.vector.tuple", <vscale x 4 x i8>, 3) @llvm.riscv.vluxseg3.triscv.vector.tuple_nxv4i8_3t.nxv1i16(target("riscv.vector.tuple", <vscale x 4 x i8>, 3) undef, ptr %base, <vscale x 1 x i16> %index, i64 %vl, i64 5)
+  %1 = call <vscale x 1 x i32> @llvm.riscv.tuple.extract.nxv1i32.triscv.vector.tuple_nxv4i8_3t(target("riscv.vector.tuple", <vscale x 4 x i8>, 3) %0, i32 1)
+  ret <vscale x 1 x i32> %1
+}
+
+define <vscale x 1 x i32> @test_vluxseg3_mask_nxv1i32_nxv1i16(ptr %base, <vscale x 1 x i16> %index, i64 %vl, <vscale x 1 x i1> %mask) sanitize_address {
+; CHECK-LABEL: @test_vluxseg3_mask_nxv1i32_nxv1i16(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i64, ptr @__asan_shadow_memory_dynamic_address, align 8
+; CHECK-NEXT:    [[TMP37:%.*]] = tail call target("riscv.vector.tuple", <vscale x 4 x i8>, 3) @llvm.riscv.vluxseg3.mask.triscv.vector.tuple_nxv4i8_3t.p0.nxv1i16.nxv1i1.i64(target("riscv.vector.tuple", <vscale x 4 x i8>, 3) undef, ptr [[BASE:%.*]], <vscale x 1 x i16> [[INDEX:%.*]], <vscale x 1 x i1> [[MASK:%.*]], i64 [[VL:%.*]], i64 1, i64 5)
+; CHECK-NEXT:    [[TMP38:%.*]] = call <vscale x 1 x i32> @llvm.riscv.tuple.extract.nxv1i32.triscv.vector.tuple_nxv4i8_3t(target("riscv.vector.tuple", <vscale x 4 x i8>, 3) [[TMP37]], i32 1)
+; CHECK-NEXT:    ret <vscale x 1 x i32> [[TMP38]]
+;
+entry:
+  %0 = tail call target("riscv.vector.tuple", <vscale x 4 x i8>, 3) @llvm.riscv.vluxseg3.mask.triscv.vector.tuple_nxv4i8_3t.nxv4i1.nxv1i32(target("riscv.vector.tuple", <vscale x 4 x i8>, 3) undef, ptr %base, <vscale x 1 x i16> %index, <vscale x 1 x i1> %mask, i64 %vl, i64 1, i64 5)
+  %1 = call <vscale x 1 x i32> @llvm.riscv.tuple.extract.nxv1i32.triscv.vector.tuple_nxv4i8_3t(target("riscv.vector.tuple", <vscale x 4 x i8>, 3) %0, i32 1)
+  ret <vscale x 1 x i32> %1
+}
+
+
+declare target("riscv.vector.tuple", <vscale x 4 x i8>, 4) @llvm.riscv.vluxseg4.triscv.vector.tuple_nxv4i8_4t.nxv4i8(target("riscv.vector.tuple", <vscale x 4 x i8>, 4), ptr, <vscale x 1 x i16>, i64, i64)
+declare target("riscv.vector.tuple", <vscale x 4 x i8>, 4) @llvm.riscv.vluxseg4.mask.triscv.vector.tuple_nxv4i8_4t.nxv1i16.nxv1i1(target("riscv.vector.tuple", <vscale x 4 x i8>, 4), ptr, <vscale x 1 x i16>, <vscale x 1 x i1>, i64, i64, i64)
+
+define <vscale x 1 x i32> @test_vluxseg4_nxv1i32_nxv1i16(ptr %base, <vscale x 1 x i16> %index, i64 %vl) sanitize_address {
+; CHECK-LABEL: @test_vluxseg4_nxv1i32_nxv1i16(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i64, ptr @__asan_shadow_memory_dynamic_address, align 8
+; CHECK-NEXT:    [[TMP49:%.*]] = tail call target("riscv.vector.tuple", <vscale x 4 x i8>, 4) @llvm.riscv.vluxseg4.triscv.vector.tuple_nxv4i8_4t.p0.nxv1i16.i64(target("riscv.vector.tuple", <vscale x 4 x i8>, 4) undef, ptr [[BASE:%.*]], <vscale x 1 x i16> [[INDEX:%.*]], i64 [[VL:%.*]], i64 5)
+; CHECK-NEXT:    [[TMP50:%.*]] = call <vscale x 1 x i32> @llvm.riscv.tuple.extract.nxv1i32.triscv.vector.tuple_nxv4i8_4t(target("riscv.vector.tuple", <vscale x 4 x i8>, 4) [[TMP49]], i32 1)
+; CHECK-NEXT:    ret <vscale x 1 x i32> [[TMP50]]
+;
+entry:
+  %0 = tail call target("riscv.vector.tuple", <vscale x 4 x i8>, 4) @llvm.riscv.vluxseg4.triscv.vector.tuple_nxv4i8_4t.nxv1i16(target("riscv.vector.tuple", <vscale x 4 x i8>, 4) undef, ptr %base, <vscale x 1 x i16> %index, i64 %vl, i64 5)
+  %1 = call <vscale x 1 x i32> @llvm.riscv.tuple.extract.nxv1i32.triscv.vector.tuple_nxv4i8_4t(target("riscv.vector.tuple", <vscale x 4 x i8>, 4) %0, i32 1)
+  ret <vscale x 1 x i32> %1
+}
+
+define <vscale x 1 x i32> @test_vluxseg4_mask_nxv1i32_nxv1i16(ptr %base, <vscale x 1 x i16> %index, i64 %vl, <vscale x 1 x i1> %mask) sanitize_address {
+; CHECK-LABEL: @test_vluxseg4_mask_nxv1i32_nxv1i16(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i64, ptr @__asan_shadow_memory_dynamic_address, align 8
+; CHECK-NEXT:    [[TMP49:%.*]] = tail call target("riscv.vector.tuple", <vscale x 4 x i8>, 4) @llvm.riscv.vluxseg4.mask.triscv.vector.tuple_nxv4i8_4t.p0.nxv1i16.nxv1i1.i64(target("riscv.vector.tuple", <vscale x 4 x i8>, 4) undef, ptr [[BASE:%.*]], <vscale x 1 x i16> [[INDEX:%.*]], <vscale x 1 x i1> [[MASK:%.*]], i64 [[VL:%.*]], i64 1, i64 5)
+; CHECK-NEXT:    [[TMP50:%.*]] = call <vscale x 1 x i32> @llvm.riscv.tuple.extract.nxv1i32.triscv.vector.tuple_nxv4i8_4t(target("riscv.vector.tuple", <vscale x 4 x i8>, 4) [[TMP49]], i32 1)
+; CHECK-NEXT:    ret <vscale x 1 x i32> [[TMP50]]
+;
+entry:
+  %0 = tail call target("riscv.vector.tuple", <vscale x 4 x i8>, 4) @llvm.riscv.vluxseg4.mask.triscv.vector.tuple_nxv4i8_4t.nxv4i1.nxv1i32(target("riscv.vector.tuple", <vscale x 4 x i8>, 4) undef, ptr %base, <vscale x 1 x i16> %index, <vscale x 1 x i1> %mask, i64 %vl, i64 1, i64 5)
+  %1 = call <vscale x 1 x i32> @llvm.riscv.tuple.extract.nxv1i32.triscv.vector.tuple_nxv4i8_4t(target("riscv.vector.tuple", <vscale x 4 x i8>, 4) %0, i32 1)
+  ret <vscale x 1 x i32> %1
+}
+
+
+declare target("riscv.vector.tuple", <vscale x 4 x i8>, 5) @llvm.riscv.vluxseg5.triscv.vector.tuple_nxv4i8_5t.nxv4i8(target("riscv.vector.tuple", <vscale x 4 x i8>, 5), ptr, <vscale x 1 x i16>, i64, i64)
+declare target("riscv.vector.tuple", <vscale x 4 x i8>, 5) @llvm.riscv.vluxseg5.mask.triscv.vector.tuple_nxv4i8_5t.nxv1i16.nxv1i1(target("riscv.vector.tuple", <vscale x 4 x i8>, 5), ptr, <vscale x 1 x i16>, <vscale x 1 x i1>, i64, i64, i64)
+
+define <vscale x 1 x i32> @test_vluxseg5_nxv1i32_nxv1i16(ptr %base, <vscale x 1 x i16> %index, i64 %vl) sanitize_address {
+; CHECK-LABEL: @test_vluxseg5_nxv1i32_nxv1i16(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i64, ptr @__asan_shadow_memory_dynamic_address, align 8
+; CHECK-NEXT:    [[TMP61:%.*]] = tail call target("riscv.vector.tuple", <vscale x 4 x i8>, 5) @llvm.riscv.vluxseg5.triscv.vector.tuple_nxv4i8_5t.p0.nxv1i16.i64(target("riscv.vector.tuple", <vscale x 4 x i8>, 5) undef, ptr [[BASE:%.*]], <vscale x 1 x i16> [[INDEX:%.*]], i64 [[VL:%.*]], i64 5)
+; CHECK-NEXT:    [[TMP62:%.*]] = call <vscale x 1 x i32> @llvm.riscv.tuple.extract.nxv1i32.triscv.vector.tuple_nxv4i8_5t(target("riscv.vector.tuple", <vscale x 4 x i8>, 5) [[TMP61]], i32 1)
+; CHECK-NEXT:    ret <vscale x 1 x i32> [[TMP62]]
+;
+entry:
+  %0 = tail call target("riscv.vector.tuple", <vscale x 4 x i8>, 5) @llvm.riscv.vluxseg5.triscv.vector.tuple_nxv4i8_5t.nxv1i16(target("riscv.vector.tuple", <vscale x 4 x i8>, 5) undef, ptr %base, <vscale x 1 x i16> %index, i64 %vl, i64 5)
+  %1 = call <vscale x 1 x i32> @llvm.riscv.tuple.extract.nxv1i32.triscv.vector.tuple_nxv4i8_5t(target("riscv.vector.tuple", <vscale x 4 x i8>, 5) %0, i32 1)
+  ret <vscale x 1 x i32> %1
+}
+
+define <vscale x 1 x i32> @test_vluxseg5_mask_nxv1i32_nxv1i16(ptr %base, <vscale x 1 x i16> %index, i64 %vl, <vscale x 1 x i1> %mask) sanitize_address {
+; CHECK-LABEL: @test_vluxseg5_mask_nxv1i32_nxv1i16(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i64, ptr @__asan_shadow_memory_dynamic_address, align 8
+; CHECK-NEXT:    [[TMP61:%.*]] = tail call target("riscv.vector.tuple", <vscale x 4 x i8>, 5) @llvm.riscv.vluxseg5.mask.triscv.vector.tuple_nxv4i8_5t.p0.nxv1i16.nxv1i1.i64(target("riscv.vector.tuple", <vscale x 4 x i8>, 5) undef, ptr [[BASE:%.*]], <vscale x 1 x i16> [[INDEX:%.*]], <vscale x 1 x i1> [[MASK:%.*]], i64 [[VL:%.*]], i64 1, i64 5)
+; CHECK-NEXT:    [[TMP62:%.*]] = call <vscale x 1 x i32> @llvm.riscv.tuple.extract.nxv1i32.triscv.vector.tuple_nxv4i8_5t(target("riscv.vector.tuple", <vscale x 4 x i8>, 5) [[TMP61]], i32 1)
+; CHECK-NEXT:    ret <vscale x 1 x i32> [[TMP62]]
+;
+entry:
+  %0 = tail call target("riscv.vector.tuple", <vscale x 4 x i8>, 5) @llvm.riscv.vluxseg5.mask.triscv.vector.tuple_nxv4i8_5t.nxv4i1.nxv1i32(target("riscv.vector.tuple", <vscale x 4 x i8>, 5) undef, ptr %base, <vscale x 1 x i16> %index, <vscale x 1 x i1> %mask, i64 %vl, i64 1, i64 5)
+  %1 = call <vscale x 1 x i32> @llvm.riscv.tuple.extract.nxv1i32.triscv.vector.tuple_nxv4i8_5t(target("riscv.vector.tuple", <vscale x 4 x i8>, 5) %0, i32 1)
+  ret <vscale x 1 x i32> %1
+}
+
+
+declare target("riscv.vector.tuple", <vscale x 4 x i8>, 6) @llvm.riscv.vluxseg6.triscv.vector.tuple_nxv4i8_6t.nxv4i8(target("riscv.vector.tuple", <vscale x 4 x i8>, 6), ptr, <vscale x 1 x i16>, i64, i64)
+declare target("riscv.vector.tuple", <vscale x 4 x i8>, 6) @llvm.riscv.vluxseg6.mask.triscv.vector.tuple_nxv4i8_6t.nxv1i16.nxv1i1(target("riscv.vector.tuple", <vscale x 4 x i8>, 6), ptr, <vscale x 1 x i16>, <vscale x 1 x i1>, i64, i64, i64)
+
+define <vscale x 1 x i32> @test_vluxseg6_nxv1i32_nxv1i16(ptr %base, <vscale x 1 x i16> %index, i64 %vl) sanitize_address {
+; CHECK-LABEL: @test_vluxseg6_nxv1i32_nxv1i16(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i64, ptr @__asan_shadow_memory_dynamic_address, align 8
+; CHECK-NEXT:    [[TMP73:%.*]] = tail call target("riscv.vector.tuple", <vscale x 4 x i8>, 6) @llvm.riscv.vluxseg6.triscv.vector.tuple_nxv4i8_6t.p0.nxv1i16.i64(target("riscv.vector.tuple", <vscale x 4 x i8>, 6) undef, ptr [[BASE:%.*]], <vscale x 1 x i16> [[INDEX:%.*]], i64 [[VL:%.*]], i64 5)
+; CHECK-NEXT:    [[TMP74:%.*]] = call <vscale x 1 x i32> @llvm.riscv.tuple.extract.nxv1i32.triscv.vector.tuple_nxv4i8_6t(target("riscv.vector.tuple", <vscale x 4 x i8>, 6) [[TMP73]], i32 1)
+; CHECK-NEXT:    ret <vscale x 1 x i32> [[TMP74]]
+;
+entry:
+  %0 = tail call target("riscv.vector.tuple", <vscale x 4 x i8>, 6) @llvm.riscv.vluxseg6.triscv.vector.tuple_nxv4i8_6t.nxv1i16(target("riscv.vector.tuple", <vscale x 4 x i8>, 6) undef, ptr %base, <vscale x 1 x i16> %index, i64 %vl, i64 5)
+  %1 = call <vscale x 1 x i32> @llvm.riscv.tuple.extract.nxv1i32.triscv.vector.tuple_nxv4i8_6t(target("riscv.vector.tuple", <vscale x 4 x i8>, 6) %0, i32 1)
+  ret <vscale x 1 x i32> %1
+}
+
+define <vscale x 1 x i32> @test_vluxseg6_mask_nxv1i32_nxv1i16(ptr %base, <vscale x 1 x i16> %index, i64 %vl, <vscale x 1 x i1> %mask) sanitize_address {
+; CHECK-LABEL: @test_vluxseg6_mask_nxv1i32_nxv1i16(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i64, ptr @__asan_shadow_memory_dynamic_address, align 8
+; CHECK-NEXT:    [[TMP73:%.*]] = tail call target("riscv.vector.tuple", <vscale x 4 x i8>, 6) @llvm.riscv.vluxseg6.mask.triscv.vector.tuple_nxv4i8_6t.p0.nxv1i16.nxv1i1.i64(target("riscv.vector.tuple", <vscale x 4 x i8>, 6) undef, ptr [[BASE:%.*]], <vscale x 1 x i16> [[INDEX:%.*]], <vscale x 1 x i1> [[MASK:%.*]], i64 [[VL:%.*]], i64 1, i64 5)
+; CHECK-NEXT:    [[TMP74:%.*]] = call <vscale x 1 x i32> @llvm.riscv.tuple.extract.nxv1i32.triscv.vector.tuple_nxv4i8_6t(target("riscv.vector.tuple", <vscale x 4 x i8>, 6) [[TMP73]], i32 1)
+; CHECK-NEXT:    ret <vscale x 1 x i32> [[TMP74]]
+;
+entry:
+  %0 = tail call target("riscv.vector.tuple", <vscale x 4 x i8>, 6) @llvm.riscv.vluxseg6.mask.triscv.vector.tuple_nxv4i8_6t.nxv4i1.nxv1i32(target("riscv.vector.tuple", <vscale x 4 x i8>, 6) undef, ptr %base, <vscale x 1 x i16> %index, <vscale x 1 x i1> %mask, i64 %vl, i64 1, i64 5)
+  %1 = call <vscale x 1 x i32> @llvm.riscv.tuple.extract.nxv1i32.triscv.vector.tuple_nxv4i8_6t(target("riscv.vector.tuple", <vscale x 4 x i8>, 6) %0, i32 1)
+  ret <vscale x 1 x i32> %1
+}
+
+
+declare target("riscv.vector.tuple", <vscale x 4 x i8>, 7) @llvm.riscv.vluxseg7.triscv.vector.tuple_nxv4i8_7t.nxv4i8(target("riscv.vector.tuple", <vscale x 4 x i8>, 7), ptr, <vscale x 1 x i16>, i64, i64)
+declare target("riscv.vector.tuple", <vscale x 4 x i8>, 7) @llvm.riscv.vluxseg7.mask.triscv.vector.tuple_nxv4i8_7t.nxv1i16.nxv1i1(target("riscv.vector.tuple", <vscale x 4 x i8>, 7), ptr, <vscale x 1 x i16>, <vscale x 1 x i1>, i64, i64, i64)
+
+define <vscale x 1 x i32> @test_vluxseg7_nxv1i32_nxv1i16(ptr %base, <vscale x 1 x i16> %index, i64 %vl) sanitize_address {
+; CHECK-LABEL: @test_vluxseg7_nxv1i32_nxv1i16(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i64, ptr @__asan_shadow_memory_dynamic_address, align 8
+; CHECK-NEXT:    [[TMP85:%.*]] = tail call target("riscv.vector.tuple", <vscale x 4 x i8>, 7) @llvm.riscv.vluxseg7.triscv.vector.tuple_nxv4i8_7t.p0.nxv1i16.i64(target("riscv.vector.tuple", <vscale x 4 x i8>, 7) undef, ptr [[BASE:%.*]], <vscale x 1 x i16> [[INDEX:%.*]], i64 [[VL:%.*]], i64 5)
+; CHECK-NEXT:    [[TMP86:%.*]] = call <vscale x 1 x i32> @llvm.riscv.tuple.extract.nxv1i32.triscv.vector.tuple_nxv4i8_7t(target("riscv.vector.tuple", <vscale x 4 x i8>, 7) [[TMP85]], i32 1)
+; CHECK-NEXT:    ret <vscale x 1 x i32> [[TMP86]]
+;
+entry:
+  %0 = tail call target("riscv.vector.tuple", <vscale x 4 x i8>, 7) @llvm.riscv.vluxseg7.triscv.vector.tuple_nxv4i8_7t.nxv1i16(target("riscv.vector.tuple", <vscale x 4 x i8>, 7) undef, ptr %base, <vscale x 1 x i16> %index, i64 %vl, i64 5)
+  %1 = call <vscale x 1 x i32> @llvm.riscv.tuple.extract.nxv1i32.triscv.vector.tuple_nxv4i8_7t(target("riscv.vector.tuple", <vscale x 4 x i8>, 7) %0, i32 1)
+  ret <vscale x 1 x i32> %1
+}
+
+define <vscale x 1 x i32> @test_vluxseg7_mask_nxv1i32_nxv1i16(ptr %base, <vscale x 1 x i16> %index, i64 %vl, <vscale x 1 x i1> %mask) sanitize_address {
+; CHECK-LABEL: @test_vluxseg7_mask_nxv1i32_nxv1i16(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i64, ptr @__asan_shadow_memory_dynamic_address, align 8
+; CHECK-NEXT:    [[TMP85:%.*]] = tail call target("riscv.vector.tuple", <vscale x 4 x i8>, 7) @llvm.riscv.vluxseg7.mask.triscv.vector.tuple_nxv4i8_7t.p0.nxv1i16.nxv1i1.i64(target("riscv.vector.tuple", <vscale x 4 x i8>, 7) undef, ptr [[BASE:%.*]], <vscale x 1 x i16> [[INDEX:%.*]], <vscale x 1 x i1> [[MASK:%.*]], i64 [[VL:%.*]], i64 1, i64 5)
+; CHECK-NEXT:    [[TMP86:%.*]] = call <vscale x 1 x i32> @llvm.riscv.tuple.extract.nxv1i32.triscv.vector.tuple_nxv4i8_7t(target("riscv.vector.tuple", <vscale x 4 x i8>, 7) [[TMP85]], i32 1)
+; CHECK-NEXT:    ret <vscale x 1 x i32> [[TMP86]]
+;
+entry:
+  %0 = tail call target("riscv.vector.tuple", <vscale x 4 x i8>, 7) @llvm.riscv.vluxseg7.mask.triscv.vector.tuple_nxv4i8_7t.nxv4i1.nxv1i32(target("riscv.vector.tuple", <vscale x 4 x i8>, 7) undef, ptr %base, <vscale x 1 x i16> %index, <vscale x 1 x i1> %mask, i64 %vl, i64 1, i64 5)
+  %1 = call <vscale x 1 x i32> @llvm.riscv.tuple.extract.nxv1i32.triscv.vector.tuple_nxv4i8_7t(target("riscv.vector.tuple", <vscale x 4 x i8>, 7) %0, i32 1)
+  ret <vscale x 1 x i32> %1
+}
+
+
+declare target("riscv.vector.tuple", <vscale x 4 x i8>, 8) @llvm.riscv.vluxseg8.triscv.vector.tuple_nxv4i8_8t.nxv4i8(target("riscv.vector.tuple", <vscale x 4 x i8>, 8), ptr, <vscale x 1 x i16>, i64, i64)
+declare target("riscv.vector.tuple", <vscale x 4 x i8>, 8) @llvm.riscv.vluxseg8.mask.triscv.vector.tuple_nxv4i8_8t.nxv1i16.nxv1i1(target("riscv.vector.tuple", <vscale x 4 x i8>, 8), ptr, <vscale x 1 x i16>, <vscale x 1 x i1>, i64, i64, i64)
+
+define <vscale x 1 x i32> @test_vluxseg8_nxv1i32_nxv1i16(ptr %base, <vscale x 1 x i16> %index, i64 %vl) sanitize_address {
+; CHECK-LABEL: @test_vluxseg8_nxv1i32_nxv1i16(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i64, ptr @__asan_shadow_memory_dynamic_address, align 8
+; CHECK-NEXT:    [[TMP97:%.*]] = tail call target("riscv.vector.tuple", <vscale x 4 x i8>, 8) @llvm.riscv.vluxseg8.triscv.vector.tuple_nxv4i8_8t.p0.nxv1i16.i64(target("riscv.vector.tuple", <vscale x 4 x i8>, 8) undef, ptr [[BASE:%.*]], <vscale x 1 x i16> [[INDEX:%.*]], i64 [[VL:%.*]], i64 5)
+; CHECK-NEXT:    [[TMP98:%.*]] = call <vscale x 1 x i32> @llvm.riscv.tuple.extract.nxv1i32.triscv.vector.tuple_nxv4i8_8t(target("riscv.vector.tuple", <vscale x 4 x i8>, 8) [[TMP97]], i32 1)
+; CHECK-NEXT:    ret <vscale x 1 x i32> [[TMP98]]
+;
+entry:
+  %0 = tail call target("riscv.vector.tuple", <vscale x 4 x i8>, 8) @llvm.riscv.vluxseg8.triscv.vector.tuple_nxv4i8_8t.nxv1i16(target("riscv.vector.tuple", <vscale x 4 x i8>, 8) undef, ptr %base, <vscale x 1 x i16> %index, i64 %vl, i64 5)
+  %1 = call <vscale x 1 x i32> @llvm.riscv.tuple.extract.nxv1i32.triscv.vector.tuple_nxv4i8_8t(target("riscv.vector.tuple", <vscale x 4 x i8>, 8) %0, i32 1)
+  ret <vscale x 1 x i32> %1
+}
+
+define <vscale x 1 x i32> @test_vluxseg8_mask_nxv1i32_nxv1i16(ptr %base, <vscale x 1 x i16> %index, i64 %vl, <vscale x 1 x i1> %mask) sanitize_address {
+; CHECK-LABEL: @test_vluxseg8_mask_nxv1i32_nxv1i16(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i64, ptr @__asan_shadow_memory_dynamic_address, align 8
+; CHECK-NEXT:    [[TMP97:%.*]] = tail call target("riscv.vector.tuple", <vscale x 4 x i8>, 8) @llvm.riscv.vluxseg8.mask.triscv.vector.tuple_nxv4i8_8t.p0.nxv1i16.nxv1i1.i64(target("riscv.vector.tuple", <vscale x 4 x i8>, 8) undef, ptr [[BASE:%.*]], <vscale x 1 x i16> [[INDEX:%.*]], <vscale x 1 x i1> [[MASK:%.*]], i64 [[VL:%.*]], i64 1, i64 5)
+; CHECK-NEXT:    [[TMP98:%.*]] = call <vscale x 1 x i32> @llvm.riscv.tuple.extract.nxv1i32.triscv.vector.tuple_nxv4i8_8t(target("riscv.vector.tuple", <vscale x 4 x i8>, 8) [[TMP97]], i32 1)
+; CHECK-NEXT:    ret <vscale x 1 x i32> [[TMP98]]
+;
+entry:
+  %0 = tail call target("riscv.vector.tuple", <vscale x 4 x i8>, 8) @llvm.riscv.vluxseg8.mask.triscv.vector.tuple_nxv4i8_8t.nxv4i1.nxv1i32(target("riscv.vector.tuple", <vscale x 4 x i8>, 8) undef, ptr %base, <vscale x 1 x i16> %index, <vscale x 1 x i1> %mask, i64 %vl, i64 1, i64 5)
+  %1 = call <vscale x 1 x i32> @llvm.riscv.tuple.extract.nxv1i32.triscv.vector.tuple_nxv4i8_8t(target("riscv.vector.tuple", <vscale x 4 x i8>, 8) %0, i32 1)
+  ret <vscale x 1 x i32> %1
+}
+
+
+declare void @llvm.riscv.vsoxseg2.triscv.vector.tuple_nxv4i8_2t.nxv4i8(target("riscv.vector.tuple", <vscale x 4 x i8>, 2), ptr, <vscale x 4 x i8>, i64, i64)
+declare void @llvm.riscv.vsoxseg2.mask.triscv.vector.tuple_nxv4i8_2t.nxv4i8.nxv4i1(target("riscv.vector.tuple", <vscale x 4 x i8>, 2), ptr, <vscale x 4 x i8>, <vscale x 4 x i1>, i64, i64)
+
+define void @test_vsoxseg2_nxv1i32_nxv1i16(target("riscv.vector.tuple", <vscale x 4 x i8>, 2) %val, ptr %base, <vscale x 1 x i16> %index, i64 %vl) sanitize_address {
+; CHECK-LABEL: @test_vsoxseg2_nxv1i32_nxv1i16(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i64, ptr @__asan_shadow_memory_dynamic_address, align 8
+; CHECK-NEXT:    tail call void @llvm.riscv.vsoxseg2.triscv.vector.tuple_nxv4i8_2t.p0.nxv1i16.i64(target("riscv.vector.tuple", <vscale x 4 x i8>, 2) [[VAL:%.*]], ptr [[BASE:%.*]], <vscale x 1 x i16> [[INDEX:%.*]], i64 [[VL:%.*]], i64 5)
+; CHECK-NEXT:    ret void
+;
+entry:
+  tail call void @llvm.riscv.vsoxseg2.triscv.vector.tuple_nxv4i8_2t.nxv1i16(target("riscv.vector.tuple", <vscale x 4 x i8>, 2) %val, ptr %base, <vscale x 1 x i16> %index, i64 %vl, i64 5)
+  ret void
+}
+
+define void @test_vsoxseg2_mask_nxv1i32_nxv1i16(target("riscv.vector.tuple", <vscale x 4 x i8>, 2) %val, ptr %base, <vscale x 1 x i16> %index, i64 %vl, <vscale x 1 x i1> %mask) sanitize_address {
+; CHECK-LABEL: @test_vsoxseg2_mask_nxv1i32_nxv1i16(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i64, ptr @__asan_shadow_memory_dynamic_address, align 8
+; CHECK-NEXT:    tail call void @llvm.riscv.vsoxseg2.mask.triscv.vector.tuple_nxv4i8_2t.p0.nxv1i16.nxv1i1.i64(target("riscv.vector.tuple", <vscale x 4 x i8>, 2) [[VAL:%.*]], ptr [[BASE:%.*]], <vscale x 1 x i16> [[INDEX:%.*]], <vscale x 1 x i1> [[MASK:%.*]], i64 [[VL:%.*]], i64 5)
+; CHECK-NEXT:    ret void
+;
+entry:
+  tail call void @llvm.riscv.vsoxseg2.mask.triscv.vector.tuple_nxv4i8_2t.nxv1i16.nxv1i1(target("riscv.vector.tuple", <vscale x 4 x i8>, 2) %val, ptr %base, <vscale x 1 x i16> %index, <vscale x 1 x i1> %mask, i64 %vl, i64 5)
+  ret void
+}
+
+
+declare void @llvm.riscv.vsoxseg3.triscv.vector.tuple_nxv4i8_3t.nxv4i8(target("riscv.vector.tuple", <vscale x 4 x i8>, 3), ptr, <vscale x 4 x i8>, i64, i64)
+declare void @llvm.riscv.vsoxseg3.mask.triscv.vector.tuple_nxv4i8_3t.nxv4i8.nxv4i1(target("riscv.vector.tuple", <vscale x 4 x i8>, 3), ptr, <vscale x 4 x i8>, <vscale x 4 x i1>, i64, i64)
+
+define void @test_vsoxseg3_nxv1i32_nxv1i16(target("riscv.vector.tuple", <vscale x 4 x i8>, 3) %val, ptr %base, <vscale x 1 x i16> %index, i64 %vl) sanitize_address {
+; CHECK-LABEL: @test_vsoxseg3_nxv1i32_nxv1i16(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i64, ptr @__asan_shadow_memory_dynamic_address, align 8
+; CHECK-NEXT:    tail call void @llvm.riscv.vsoxseg3.triscv.vector.tuple_nxv4i8_3t.p0.nxv1i16.i64(target("riscv.vector.tuple", <vscale x 4 x i8>, 3) [[VAL:%.*]], ptr [[BASE:%.*]], <vscale x 1 x i16> [[INDEX:%.*]], i64 [[VL:%.*]], i64 5)
+; CHECK-NEXT:    ret void
+;
+entry:
+  tail call void @llvm.riscv.vsoxseg3.triscv.vector.tuple_nxv4i8_3t.nxv1i16(target("riscv.vector.tuple", <vscale x 4 x i8>, 3) %val, ptr %base, <vscale x 1 x i16> %index, i64 %vl, i64 5)
+  ret void
+}
+
+define void @test_vsoxseg3_mask_nxv1i32_nxv1i16(target("riscv.vector.tuple", <vscale x 4 x i8>, 3) %val, ptr %base, <vscale x 1 x i16> %index, i64 %vl, <vscale x 1 x i1> %mask) sanitize_address {
+; CHECK-LABEL: @test_vsoxseg3_mask_nxv1i32_nxv1i16(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i64, ptr @__asan_shadow_memory_dynamic_address, align 8
+; CHECK-NEXT:    tail call void @llvm.riscv.vsoxseg3.mask.triscv.vector.tuple_nxv4i8_3t.p0.nxv1i16.nxv1i1.i64(target("riscv.vector.tuple", <vscale x 4 x i8>, 3) [[VAL:%.*]], ptr [[BASE:%.*]], <vscale x 1 x i16> [[INDEX:%.*]], <vscale x 1 x i1> [[MASK:%.*]], i64 [[VL:%.*]], i64 5)
+; CHECK-NEXT:    ret void
+;
+entry:
+  tail call void @llvm.riscv.vsoxseg3.mask.triscv.vector.tuple_nxv4i8_3t.nxv1i16.nxv1i1(target("riscv.vector.tuple", <vscale x 4 x i8>, 3) %val, ptr %base, <vscale x 1 x i16> %index, <vscale x 1 x i1> %mask, i64 %vl, i64 5)
+  ret void
+}
+
+
+declare void @llvm.riscv.vsoxseg4.triscv.vector.tuple_nxv4i8_4t.nxv4i8(target("riscv.vector.tuple", <vscale x 4 x i8>, 4), ptr, <vscale x 4 x i8>, i64, i64)
+declare void @llvm.riscv.vsoxseg4.mask.triscv.vector.tuple_nxv4i8_4t.nxv4i8.nxv4i1(target("riscv.vector.tuple", <vscale x 4 x i8>, 4), ptr, <vscale x 4 x i8>, <vscale x 4 x i1>, i64, i64)
+
+define void @test_vsoxseg4_nxv1i32_nxv1i16(target("riscv.vector.tuple", <vscale x 4 x i8>, 4) %val, ptr %base, <vscale x 1 x i16> %index, i64 %vl) sanitize_address {
+; CHECK-LABEL: @test_vsoxseg4_nxv1i32_nxv1i16(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i64, ptr @__asan_shadow_memory_dynamic_address, align 8
+; CHECK-NEXT:    tail call void @llvm.riscv.vsoxseg4.triscv.vector.tuple_nxv4i8_4t.p0.nxv1i16.i64(target("riscv.vector.tuple", <vscale x 4 x i8>, 4) [[VAL:%.*]], ptr [[BASE:%.*]], <vscale x 1 x i16> [[INDEX:%.*]], i64 [[VL:%.*]], i64 5)
+; CHECK-NEXT:    ret void
+;
+entry:
+  tail call void @llvm.riscv.vsoxseg4.triscv.vector.tuple_nxv4i8_4t.nxv1i16(target("riscv.vector.tuple", <vscale x 4 x i8>, 4) %val, ptr %base, <vscale x 1 x i16> %index, i64 %vl, i64 5)
+  ret void
+}
+
+define void @test_vsoxseg4_mask_nxv1i32_nxv1i16(target("riscv.vector.tuple", <vscale x 4 x i8>, 4) %val, ptr %base, <vscale x 1 x i16> %index, i64 %vl, <vscale x 1 x i1> %mask) sanitize_address {
+; CHECK-LABEL: @test_vsoxseg4_mask_nxv1i32_nxv1i16(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i64, ptr @__asan_shadow_memory_dynamic_address, align 8
+; CHECK-NEXT:    tail call void @llvm.riscv.vsoxseg4.mask.triscv.vector.tuple_nxv4i8_4t.p0.nxv1i16.nxv1i1.i64(target("riscv.vector.tuple", <vscale x 4 x i8>, 4) [[VAL:%.*]], ptr [[BASE:%.*]], <vscale x 1 x i16> [[INDEX:%.*]], <vscale x 1 x i1> [[MASK:%.*]], i64 [[VL:%.*]], i64 5)
+; CHECK-NEXT:    ret void
+;
+entry:
+  tail call void @llvm.riscv.vsoxseg4.mask.triscv.vector.tuple_nxv4i8_4t.nxv1i16.nxv1i1(target("riscv.vector.tuple", <vscale x 4 x i8>, 4) %val, ptr %base, <vscale x 1 x i16> %index, <vscale x 1 x i1> %mask, i64 %vl, i64 5)
+  ret void
+}
+
+
+declare void @llvm.riscv.vsoxseg5.triscv.vector.tuple_nxv4i8_5t.nxv4i8(target("riscv.vector.tuple", <vscale x 4 x i8>, 5), ptr, <vscale x 4 x i8>, i64, i64)
+declare void @llvm.riscv.vsoxseg5.mask.triscv.vector.tuple_nxv4i8_5t.nxv4i8.nxv4i1(target("riscv.vector.tuple", <vscale x 4 x i8>, 5), ptr, <vscale x 4 x i8>, <vscale x 4 x i1>, i64, i64)
+
+define void @test_vsoxseg5_nxv1i32_nxv1i16(target("riscv.vector.tuple", <vscale x 4 x i8>, 5) %val, ptr %base, <vscale x 1 x i16> %index, i64 %vl) sanitize_address {
+; CHECK-LABEL: @test_vsoxseg5_nxv1i32_nxv1i16(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i64, ptr @__asan_shadow_memory_dynamic_address, align 8
+; CHECK-NEXT:    tail call void @llvm.riscv.vsoxseg5.triscv.vector.tuple_nxv4i8_5t.p0.nxv1i16.i64(target("riscv.vector.tuple", <vscale x 4 x i8>, 5) [[VAL:%.*]], ptr [[BASE:%.*]], <vscale x 1 x i16> [[INDEX:%.*]], i64 [[VL:%.*]], i64 5)
+; CHECK-NEXT:    ret void
+;
+entry:
+  tail call void @llvm.riscv.vsoxseg5.triscv.vector.tuple_nxv4i8_5t.nxv1i16(target("riscv.vector.tuple", <vscale x 4 x i8>, 5) %val, ptr %base, <vscale x 1 x i16> %index, i64 %vl, i64 5)
+  ret void
+}
+
+define void @test_vsoxseg5_mask_nxv1i32_nxv1i16(target("riscv.vector.tuple", <vscale x 4 x i8>, 5) %val, ptr %base, <vscale x 1 x i16> %index, i64 %vl, <vscale x 1 x i1> %mask) sanitize_address {
+; CHECK-LABEL: @test_vsoxseg5_mask_nxv1i32_nxv1i16(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i64, ptr @__asan_shadow_memory_dynamic_address, align 8
+; CHECK-NEXT:    tail call void @llvm.riscv.vsoxseg5.mask.triscv.vector.tuple_nxv4i8_5t.p0.nxv1i16.nxv1i1.i64(target("riscv.vector.tuple", <vscale x 4 x i8>, 5) [[VAL:%.*]], ptr [[BASE:%.*]], <vscale x 1 x i16> [[INDEX:%.*]], <vscale x 1 x i1> [[MASK:%.*]], i64 [[VL:%.*]], i64 5)
+; CHECK-NEXT:    ret void
+;
+entry:
+  tail call void @llvm.riscv.vsoxseg5.mask.triscv.vector.tuple_nxv4i8_5t.nxv1i16.nxv1i1(target("riscv.vector.tuple", <vscale x 4 x i8>, 5) %val, ptr %base, <vscale x 1 x i16> %index, <vscale x 1 x i1> %mask, i64 %vl, i64 5)
+  ret void
+}
+
+
+declare void @llvm.riscv.vsoxseg6.triscv.vector.tuple_nxv4i8_6t.nxv4i8(target("riscv.vector.tuple", <vscale x 4 x i8>, 6), ptr, <vscale x 4 x i8>, i64, i64)
+declare void @llvm.riscv.vsoxseg6.mask.triscv.vector.tuple_nxv4i8_6t.nxv4i8.nxv4i1(target("riscv.vector.tuple", <vscale x 4 x i8>, 6), ptr, <vscale x 4 x i8>, <vscale x 4 x i1>, i64, i64)
+
+define void @test_vsoxseg6_nxv1i32_nxv1i16(target("riscv.vector.tuple", <vscale x 4 x i8>, 6) %val, ptr %base, <vscale x 1 x i16> %index, i64 %vl) sanitize_address {
+; CHECK-LABEL: @test_vsoxseg6_nxv1i32_nxv1i16(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i64, ptr @__asan_shadow_memory_dynamic_address, align 8
+; CHECK-NEXT:    tail call void @llvm.riscv.vsoxseg6.triscv.vector.tuple_nxv4i8_6t.p0.nxv1i16.i64(target("riscv.vector.tuple", <vscale x 4 x i8>, 6) [[VAL:%.*]], ptr [[BASE:%.*]], <vscale x 1 x i16> [[INDEX:%.*]], i64 [[VL:%.*]], i64 5)
+; CHECK-NEXT:    ret void
+;
+entry:
+  tail call void @llvm.riscv.vsoxseg6.triscv.vector.tuple_nxv4i8_6t.nxv1i16(target("riscv.vector.tuple", <vscale x 4 x i8>, 6) %val, ptr %base, <vscale x 1 x i16> %index, i64 %vl, i64 5)
+  ret void
+}
+
+define void @test_vsoxseg6_mask_nxv1i32_nxv1i16(target("riscv.vector.tuple", <vscale x 4 x i8>, 6) %val, ptr %base, <vscale x 1 x i16> %index, i64 %vl, <vscale x 1 x i1> %mask) sanitize_address {
+; CHECK-LABEL: @test_vsoxseg6_mask_nxv1i32_nxv1i16(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i64, ptr @__asan_shadow_memory_dynamic_address, align 8
+; CHECK-NEXT:    tail call void @llvm.riscv.vsoxseg6.mask.triscv.vector.tuple_nxv4i8_6t.p0.nxv1i16.nxv1i1.i64(target("riscv.vector.tuple", <vscale x 4 x i8>, 6) [[VAL:%.*]], ptr [[BASE:%.*]], <vscale x 1 x i16> [[INDEX:%.*]], <vscale x 1 x i1> [[MASK:%.*]], i64 [[VL:%.*]], i64 5)
+; CHECK-NEXT:    ret void
+;
+entry:
+  tail call void @llvm.riscv.vsoxseg6.mask.triscv.vector.tuple_nxv4i8_6t.nxv1i16.nxv1i1(target("riscv.vector.tuple", <vscale x 4 x i8>, 6) %val, ptr %base, <vscale x 1 x i16> %index, <vscale x 1 x i1> %mask, i64 %vl, i64 5)
+  ret void
+}
+
+
+declare void @llvm.riscv.vsoxseg7.triscv.vector.tuple_nxv4i8_7t.nxv4i8(target("riscv.vector.tuple", <vscale x 4 x i8>, 7), ptr, <vscale x 4 x i8>, i64, i64)
+declare void @llvm.riscv.vsoxseg7.mask.triscv.vector.tuple_nxv4i8_7t.nxv4i8.nxv4i1(target("riscv.vector.tuple", <vscale x 4 x i8>, 7), ptr, <vscale x 4 x i8>, <vscale x 4 x i1>, i64, i64)
+
+define void @test_vsoxseg7_nxv1i32_nxv1i16(target("riscv.vector.tuple", <vscale x 4 x i8>, 7) %val, ptr %base, <vscale x 1 x i16> %index, i64 %vl) sanitize_address {
+; CHECK-LABEL: @test_vsoxseg7_nxv1i32_nxv1i16(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i64, ptr @__asan_shadow_memory_dynamic_address, align 8
+; CHECK-NEXT:    tail call void @llvm.riscv.vsoxseg7.triscv.vector.tuple_nxv4i8_7t.p0.nxv1i16.i64(target("riscv.vector.tuple", <vscale x 4 x i8>, 7) [[VAL:%.*]], ptr [[BASE:%.*]], <vscale x 1 x i16> [[INDEX:%.*]], i64 [[VL:%.*]], i64 5)
+; CHECK-NEXT:    ret void
+;
+entry:
+  tail call void @llvm.riscv.vsoxseg7.triscv.vector.tuple_nxv4i8_7t.nxv1i16(target("riscv.vector.tuple", <vscale x 4 x i8>, 7) %val, ptr %base, <vscale x 1 x i16> %index, i64 %vl, i64 5)
+  ret void
+}
+
+define void @test_vsoxseg7_mask_nxv1i32_nxv1i16(target("riscv.vector.tuple", <vscale x 4 x i8>, 7) %val, ptr %base, <vscale x 1 x i16> %index, i64 %vl, <vscale x 1 x i1> %mask) sanitize_address {
+; CHECK-LABEL: @test_vsoxseg7_mask_nxv1i32_nxv1i16(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i64, ptr @__asan_shadow_memory_dynamic_address, align 8
+; CHECK-NEXT:    tail call void @llvm.riscv.vsoxseg7.mask.triscv.vector.tuple_nxv4i8_7t.p0.nxv1i16.nxv1i1.i64(target("riscv.vector.tuple", <vscale x 4 x i8>, 7) [[VAL:%.*]], ptr [[BASE:%.*]], <vscale x 1 x i16> [[INDEX:%.*]], <vscale x 1 x i1> [[MASK:%.*]], i64 [[VL:%.*]], i64 5)
+; CHECK-NEXT:    ret void
+;
+entry:
+  tail call void @llvm.riscv.vsoxseg7.mask.triscv.vector.tuple_nxv4i8_7t.nxv1i16.nxv1i1(target("riscv.vector.tuple", <vscale x 4 x i8>, 7) %val, ptr %base, <vscale x 1 x i16> %index, <vscale x 1 x i1> %mask, i64 %vl, i64 5)
+  ret void
+}
+
+
+declare void @llvm.riscv.vsoxseg8.triscv.vector.tuple_nxv4i8_8t.nxv4i8(target("riscv.vector.tuple", <vscale x 4 x i8>, 8), ptr, <vscale x 4 x i8>, i64, i64)
+declare void @llvm.riscv.vsoxseg8.mask.triscv.vector.tuple_nxv4i8_8t.nxv4i8.nxv4i1(target("riscv.vector.tuple", <vscale x 4 x i8>, 8), ptr, <vscale x 4 x i8>, <vscale x 4 x i1>, i64, i64)
+
+define void @test_vsoxseg8_nxv1i32_nxv1i16(target("riscv.vector.tuple", <vscale x 4 x i8>, 8) %val, ptr %base, <vscale x 1 x i16> %index, i64 %vl) sanitize_address {
+; CHECK-LABEL: @test_vsoxseg8_nxv1i32_nxv1i16(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i64, ptr @__asan_shadow_memory_dynamic_address, align 8
+; CHECK-NEXT:    tail call void @llvm.riscv.vsoxseg8.triscv.vector.tuple_nxv4i8_8t.p0.nxv1i16.i64(target("riscv.vector.tuple", <vscale x 4 x i8>, 8) [[VAL:%.*]], ptr [[BASE:%.*]], <vscale x 1 x i16> [[INDEX:%.*]], i64 [[VL:%.*]], i64 5)
+; CHECK-NEXT:    ret void
+;
+entry:
+  tail call void @llvm.riscv.vsoxseg8.triscv.vector.tuple_nxv4i8_8t.nxv1i16(target("riscv.vector.tuple", <vscale x 4 x i8>, 8) %val, ptr %base, <vscale x 1 x i16> %index, i64 %vl, i64 5)
+  ret void
+}
+
+define void @test_vsoxseg8_mask_nxv1i32_nxv1i16(target("riscv.vector.tuple", <vscale x 4 x i8>, 8) %val, ptr %base, <vscale x 1 x i16> %index, i64 %vl, <vscale x 1 x i1> %mask) sanitize_address {
+; CHECK-LABEL: @test_vsoxseg8_mask_nxv1i32_nxv1i16(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i64, ptr @__asan_shadow_memory_dynamic_address, align 8
+; CHECK-NEXT:    tail call void @llvm.riscv.vsoxseg8.mask.triscv.vector.tuple_nxv4i8_8t.p0.nxv1i16.nxv1i1.i64(target("riscv.vector.tuple", <vscale x 4 x i8>, 8) [[VAL:%.*]], ptr [[BASE:%.*]], <vscale x 1 x i16> [[INDEX:%.*]], <vscale x 1 x i1> [[MASK:%.*]], i64 [[VL:%.*]], i64 5)
+; CHECK-NEXT:    ret void
+;
+entry:
+  tail call void @llvm.riscv.vsoxseg8.mask.triscv.vector.tuple_nxv4i8_8t.nxv1i16.nxv1i1(target("riscv.vector.tuple", <vscale x 4 x i8>, 8) %val, ptr %base, <vscale x 1 x i16> %index, <vscale x 1 x i1> %mask, i64 %vl, i64 5)
+  ret void
+}
+
+
+declare void @llvm.riscv.vsuxseg2.triscv.vector.tuple_nxv4i8_2t.nxv4i8(target("riscv.vector.tuple", <vscale x 4 x i8>, 2), ptr, <vscale x 4 x i8>, i64, i64)
+declare void @llvm.riscv.vsuxseg2.mask.triscv.vector.tuple_nxv4i8_2t.nxv4i8.nxv4i1(target("riscv.vector.tuple", <vscale x 4 x i8>, 2), ptr, <vscale x 4 x i8>, <vscale x 4 x i1>, i64, i64)
+
+define void @test_vsuxseg2_nxv1i32_nxv1i16(target("riscv.vector.tuple", <vscale x 4 x i8>, 2) %val, ptr %base, <vscale x 1 x i16> %index, i64 %vl) sanitize_address {
+; CHECK-LABEL: @test_vsuxseg2_nxv1i32_nxv1i16(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i64, ptr @__asan_shadow_memory_dynamic_address, align 8
+; CHECK-NEXT:    tail call void @llvm.riscv.vsuxseg2.triscv.vector.tuple_nxv4i8_2t.p0.nxv1i16.i64(target("riscv.vector.tuple", <vscale x 4 x i8>, 2) [[VAL:%.*]], ptr [[BASE:%.*]], <vscale x 1 x i16> [[INDEX:%.*]], i64 [[VL:%.*]], i64 5)
+; CHECK-NEXT:    ret void
+;
+entry:
+  tail call void @llvm.riscv.vsuxseg2.triscv.vector.tuple_nxv4i8_2t.nxv1i16(target("riscv.vector.tuple", <vscale x 4 x i8>, 2) %val, ptr %base, <vscale x 1 x i16> %index, i64 %vl, i64 5)
+  ret void
+}
+
+define void @test_vsuxseg2_mask_nxv1i32_nxv1i16(target("riscv.vector.tuple", <vscale x 4 x i8>, 2) %val, ptr %base, <vscale x 1 x i16> %index, i64 %vl, <vscale x 1 x i1> %mask) sanitize_address {
+; CHECK-LABEL: @test_vsuxseg2_mask_nxv1i32_nxv1i16(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i64, ptr @__asan_shadow_memory_dynamic_address, align 8
+; CHECK-NEXT:    tail call void @llvm.riscv.vsoxseg2.mask.triscv.vector.tuple_nxv4i8_2t.p0.nxv1i16.nxv1i1.i64(target("riscv.vector.tuple", <vscale x 4 x i8>, 2) [[VAL:%.*]], ptr [[BASE:%.*]], <vscale x 1 x i16> [[INDEX:%.*]], <vscale x 1 x i1> [[MASK:%.*]], i64 [[VL:%.*]], i64 5)
+; CHECK-NEXT:    ret void
+;
+entry:
+  tail call void @llvm.riscv.vsoxseg2.mask.triscv.vector.tuple_nxv4i8_2t.nxv1i16.nxv1i1(target("riscv.vector.tuple", <vscale x 4 x i8>, 2) %val, ptr %base, <vscale x 1 x i16> %index, <vscale x 1 x i1> %mask, i64 %vl, i64 5)
+  ret void
+}
+
+
+declare void @llvm.riscv.vsuxseg3.triscv.vector.tuple_nxv4i8_3t.nxv4i8(target("riscv.vector.tuple", <vscale x 4 x i8>, 3), ptr, <vscale x 4 x i8>, i64, i64)
+declare void @llvm.riscv.vsuxseg3.mask.triscv.vector.tuple_nxv4i8_3t.nxv4i8.nxv4i1(target("riscv.vector.tuple", <vscale x 4 x i8>, 3), ptr, <vscale x 4 x i8>, <vscale x 4 x i1>, i64, i64)
+
+define void @test_vsuxseg3_nxv1i32_nxv1i16(target("riscv.vector.tuple", <vscale x 4 x i8>, 3) %val, ptr %base, <vscale x 1 x i16> %index, i64 %vl) sanitize_address {
+; CHECK-LABEL: @test_vsuxseg3_nxv1i32_nxv1i16(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i64, ptr @__asan_shadow_memory_dynamic_address, align 8
+; CHECK-NEXT:    tail call void @llvm.riscv.vsuxseg3.triscv.vector.tuple_nxv4i8_3t.p0.nxv1i16.i64(target("riscv.vector.tuple", <vscale x 4 x i8>, 3) [[VAL:%.*]], ptr [[BASE:%.*]], <vscale x 1 x i16> [[INDEX:%.*]], i64 [[VL:%.*]], i64 5)
+; CHECK-NEXT:    ret void
+;
+entry:
+  tail call void @llvm.riscv.vsuxseg3.triscv.vector.tuple_nxv4i8_3t.nxv1i16(target("riscv.vector.tuple", <vscale x 4 x i8>, 3) %val, ptr %base, <vscale x 1 x i16> %index, i64 %vl, i64 5)
+  ret void
+}
+
+define void @test_vsuxseg3_mask_nxv1i32_nxv1i16(target("riscv.vector.tuple", <vscale x 4 x i8>, 3) %val, ptr %base, <vscale x 1 x i16> %index, i64 %vl, <vscale x 1 x i1> %mask) sanitize_address {
+; CHECK-LABEL: @test_vsuxseg3_mask_nxv1i32_nxv1i16(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i64, ptr @__asan_shadow_memory_dynamic_address, align 8
+; CHECK-NEXT:    tail call void @llvm.riscv.vsoxseg3.mask.triscv.vector.tuple_nxv4i8_3t.p0.nxv1i16.nxv1i1.i64(target("riscv.vector.tuple", <vscale x 4 x i8>, 3) [[VAL:%.*]], ptr [[BASE:%.*]], <vscale x 1 x i16> [[INDEX:%.*]], <vscale x 1 x i1> [[MASK:%.*]], i64 [[VL:%.*]], i64 5)
+; CHECK-NEXT:    ret void
+;
+entry:
+  tail call void @llvm.riscv.vsoxseg3.mask.triscv.vector.tuple_nxv4i8_3t.nxv1i16.nxv1i1(target("riscv.vector.tuple", <vscale x 4 x i8>, 3) %val, ptr %base, <vscale x 1 x i16> %index, <vscale x 1 x i1> %mask, i64 %vl, i64 5)
+  ret void
+}
+
+
+declare void @llvm.riscv.vsuxseg4.triscv.vector.tuple_nxv4i8_4t.nxv4i8(target("riscv.vector.tuple", <vscale x 4 x i8>, 4), ptr, <vscale x 4 x i8>, i64, i64)
+declare void @llvm.riscv.vsuxseg4.mask.triscv.vector.tuple_nxv4i8_4t.nxv4i8.nxv4i1(target("riscv.vector.tuple", <vscale x 4 x i8>, 4), ptr, <vscale x 4 x i8>, <vscale x 4 x i1>, i64, i64)
+
+define void @test_vsuxseg4_nxv1i32_nxv1i16(target("riscv.vector.tuple", <vscale x 4 x i8>, 4) %val, ptr %base, <vscale x 1 x i16> %index, i64 %vl) sanitize_address {
+; CHECK-LABEL: @test_vsuxseg4_nxv1i32_nxv1i16(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i64, ptr @__asan_shadow_memory_dynamic_address, align 8
+; CHECK-NEXT:    tail call void @llvm.riscv.vsuxseg4.triscv.vector.tuple_nxv4i8_4t.p0.nxv1i16.i64(target("riscv.vector.tuple", <vscale x 4 x i8>, 4) [[VAL:%.*]], ptr [[BASE:%.*]], <vscale x 1 x i16> [[INDEX:%.*]], i64 [[VL:%.*]], i64 5)
+; CHECK-NEXT:    ret void
+;
+entry:
+  tail call void @llvm.riscv.vsuxseg4.triscv.vector.tuple_nxv4i8_4t.nxv1i16(target("riscv.vector.tuple", <vscale x 4 x i8>, 4) %val, ptr %base, <vscale x 1 x i16> %index, i64 %vl, i64 5)
+  ret void
+}
+
+define void @test_vsuxseg4_mask_nxv1i32_nxv1i16(target("riscv.vector.tuple", <vscale x 4 x i8>, 4) %val, ptr %base, <vscale x 1 x i16> %index, i64 %vl, <vscale x 1 x i1> %mask) sanitize_address {
+; CHECK-LABEL: @test_vsuxseg4_mask_nxv1i32_nxv1i16(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i64, ptr @__asan_shadow_memory_dynamic_address, align 8
+; CHECK-NEXT:    tail call void @llvm.riscv.vsoxseg4.mask.triscv.vector.tuple_nxv4i8_4t.p0.nxv1i16.nxv1i1.i64(target("riscv.vector.tuple", <vscale x 4 x i8>, 4) [[VAL:%.*]], ptr [[BASE:%.*]], <vscale x 1 x i16> [[INDEX:%.*]], <vscale x 1 x i1> [[MASK:%.*]], i64 [[VL:%.*]], i64 5)
+; CHECK-NEXT:    ret void
+;
+entry:
+  tail call void @llvm.riscv.vsoxseg4.mask.triscv.vector.tuple_nxv4i8_4t.nxv1i16.nxv1i1(target("riscv.vector.tuple", <vscale x 4 x i8>, 4) %val, ptr %base, <vscale x 1 x i16> %index, <vscale x 1 x i1> %mask, i64 %vl, i64 5)
+  ret void
+}
+
+
+declare void @llvm.riscv.vsuxseg5.triscv.vector.tuple_nxv4i8_5t.nxv4i8(target("riscv.vector.tuple", <vscale x 4 x i8>, 5), ptr, <vscale x 4 x i8>, i64, i64)
+declare void @llvm.riscv.vsuxseg5.mask.triscv.vector.tuple_nxv4i8_5t.nxv4i8.nxv4i1(target("riscv.vector.tuple", <vscale x 4 x i8>, 5), ptr, <vscale x 4 x i8>, <vscale x 4 x i1>, i64, i64)
+
+define void @test_vsuxseg5_nxv1i32_nxv1i16(target("riscv.vector.tuple", <vscale x 4 x i8>, 5) %val, ptr %base, <vscale x 1 x i16> %index, i64 %vl) sanitize_address {
+; CHECK-LABEL: @test_vsuxseg5_nxv1i32_nxv1i16(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i64, ptr @__asan_shadow_memory_dynamic_address, align 8
+; CHECK-NEXT:    tail call void @llvm.riscv.vsuxseg5.triscv.vector.tuple_nxv4i8_5t.p0.nxv1i16.i64(target("riscv.vector.tuple", <vscale x 4 x i8>, 5) [[VAL:%.*]], ptr [[BASE:%.*]], <vscale x 1 x i16> [[INDEX:%.*]], i64 [[VL:%.*]], i64 5)
+; CHECK-NEXT:    ret void
+;
+entry:
+  tail call void @llvm.riscv.vsuxseg5.triscv.vector.tuple_nxv4i8_5t.nxv1i16(target("riscv.vector.tuple", <vscale x 4 x i8>, 5) %val, ptr %base, <vscale x 1 x i16> %index, i64 %vl, i64 5)
+  ret void
+}
+
+define void @test_vsuxseg5_mask_nxv1i32_nxv1i16(target("riscv.vector.tuple", <vscale x 4 x i8>, 5) %val, ptr %base, <vscale x 1 x i16> %index, i64 %vl, <vscale x 1 x i1> %mask) sanitize_address {
+; CHECK-LABEL: @test_vsuxseg5_mask_nxv1i32_nxv1i16(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i64, ptr @__asan_shadow_memory_dynamic_address, align 8
+; CHECK-NEXT:    tail call void @llvm.riscv.vsoxseg5.mask.triscv.vector.tuple_nxv4i8_5t.p0.nxv1i16.nxv1i1.i64(target("riscv.vector.tuple", <vscale x 4 x i8>, 5) [[VAL:%.*]], ptr [[BASE:%.*]], <vscale x 1 x i16> [[INDEX:%.*]], <vscale x 1 x i1> [[MASK:%.*]], i64 [[VL:%.*]], i64 5)
+; CHECK-NEXT:    ret void
+;
+entry:
+  tail call void @llvm.riscv.vsoxseg5.mask.triscv.vector.tuple_nxv4i8_5t.nxv1i16.nxv1i1(target("riscv.vector.tuple", <vscale x 4 x i8>, 5) %val, ptr %base, <vscale x 1 x i16> %index, <vscale x 1 x i1> %mask, i64 %vl, i64 5)
+  ret void
+}
+
+
+declare void @llvm.riscv.vsuxseg6.triscv.vector.tuple_nxv4i8_6t.nxv4i8(target("riscv.vector.tuple", <vscale x 4 x i8>, 6), ptr, <vscale x 4 x i8>, i64, i64)
+declare void @llvm.riscv.vsuxseg6.mask.triscv.vector.tuple_nxv4i8_6t.nxv4i8.nxv4i1(target("riscv.vector.tuple", <vscale x 4 x i8>, 6), ptr, <vscale x 4 x i8>, <vscale x 4 x i1>, i64, i64)
+
+define void @test_vsuxseg6_nxv1i32_nxv1i16(target("riscv.vector.tuple", <vscale x 4 x i8>, 6) %val, ptr %base, <vscale x 1 x i16> %index, i64 %vl) sanitize_address {
+; CHECK-LABEL: @test_vsuxseg6_nxv1i32_nxv1i16(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i64, ptr @__asan_shadow_memory_dynamic_address, align 8
+; CHECK-NEXT:    tail call void @llvm.riscv.vsuxseg6.triscv.vector.tuple_nxv4i8_6t.p0.nxv1i16.i64(target("riscv.vector.tuple", <vscale x 4 x i8>, 6) [[VAL:%.*]], ptr [[BASE:%.*]], <vscale x 1 x i16> [[INDEX:%.*]], i64 [[VL:%.*]], i64 5)
+; CHECK-NEXT:    ret void
+;
+entry:
+  tail call void @llvm.riscv.vsuxseg6.triscv.vector.tuple_nxv4i8_6t.nxv1i16(target("riscv.vector.tuple", <vscale x 4 x i8>, 6) %val, ptr %base, <vscale x 1 x i16> %index, i64 %vl, i64 5)
+  ret void
+}
+
+define void @test_vsuxseg6_mask_nxv1i32_nxv1i16(target("riscv.vector.tuple", <vscale x 4 x i8>, 6) %val, ptr %base, <vscale x 1 x i16> %index, i64 %vl, <vscale x 1 x i1> %mask) sanitize_address {
+; CHECK-LABEL: @test_vsuxseg6_mask_nxv1i32_nxv1i16(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i64, ptr @__asan_shadow_memory_dynamic_address, align 8
+; CHECK-NEXT:    tail call void @llvm.riscv.vsoxseg6.mask.triscv.vector.tuple_nxv4i8_6t.p0.nxv1i16.nxv1i1.i64(target("riscv.vector.tuple", <vscale x 4 x i8>, 6) [[VAL:%.*]], ptr [[BASE:%.*]], <vscale x 1 x i16> [[INDEX:%.*]], <vscale x 1 x i1> [[MASK:%.*]], i64 [[VL:%.*]], i64 5)
+; CHECK-NEXT:    ret void
+;
+entry:
+  tail call void @llvm.riscv.vsoxseg6.mask.triscv.vector.tuple_nxv4i8_6t.nxv1i16.nxv1i1(target("riscv.vector.tuple", <vscale x 4 x i8>, 6) %val, ptr %base, <vscale x 1 x i16> %index, <vscale x 1 x i1> %mask, i64 %vl, i64 5)
+  ret void
+}
+
+
+declare void @llvm.riscv.vsuxseg7.triscv.vector.tuple_nxv4i8_7t.nxv4i8(target("riscv.vector.tuple", <vscale x 4 x i8>, 7), ptr, <vscale x 4 x i8>, i64, i64)
+declare void @llvm.riscv.vsuxseg7.mask.triscv.vector.tuple_nxv4i8_7t.nxv4i8.nxv4i1(target("riscv.vector.tuple", <vscale x 4 x i8>, 7), ptr, <vscale x 4 x i8>, <vscale x 4 x i1>, i64, i64)
+
+define void @test_vsuxseg7_nxv1i32_nxv1i16(target("riscv.vector.tuple", <vscale x 4 x i8>, 7) %val, ptr %base, <vscale x 1 x i16> %index, i64 %vl) sanitize_address {
+; CHECK-LABEL: @test_vsuxseg7_nxv1i32_nxv1i16(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i64, ptr @__asan_shadow_memory_dynamic_address, align 8
+; CHECK-NEXT:    tail call void @llvm.riscv.vsuxseg7.triscv.vector.tuple_nxv4i8_7t.p0.nxv1i16.i64(target("riscv.vector.tuple", <vscale x 4 x i8>, 7) [[VAL:%.*]], ptr [[BASE:%.*]], <vscale x 1 x i16> [[INDEX:%.*]], i64 [[VL:%.*]], i64 5)
+; CHECK-NEXT:    ret void
+;
+entry:
+  tail call void @llvm.riscv.vsuxseg7.triscv.vector.tuple_nxv4i8_7t.nxv1i16(target("riscv.vector.tuple", <vscale x 4 x i8>, 7) %val, ptr %base, <vscale x 1 x i16> %index, i64 %vl, i64 5)
+  ret void
+}
+
+define void @test_vsuxseg7_mask_nxv1i32_nxv1i16(target("riscv.vector.tuple", <vscale x 4 x i8>, 7) %val, ptr %base, <vscale x 1 x i16> %index, i64 %vl, <vscale x 1 x i1> %mask) sanitize_address {
+; CHECK-LABEL: @test_vsuxseg7_mask_nxv1i32_nxv1i16(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i64, ptr @__asan_shadow_memory_dynamic_address, align 8
+; CHECK-NEXT:    tail call void @llvm.riscv.vsoxseg7.mask.triscv.vector.tuple_nxv4i8_7t.p0.nxv1i16.nxv1i1.i64(target("riscv.vector.tuple", <vscale x 4 x i8>, 7) [[VAL:%.*]], ptr [[BASE:%.*]], <vscale x 1 x i16> [[INDEX:%.*]], <vscale x 1 x i1> [[MASK:%.*]], i64 [[VL:%.*]], i64 5)
+; CHECK-NEXT:    ret void
+;
+entry:
+  tail call void @llvm.riscv.vsoxseg7.mask.triscv.vector.tuple_nxv4i8_7t.nxv1i16.nxv1i1(target("riscv.vector.tuple", <vscale x 4 x i8>, 7) %val, ptr %base, <vscale x 1 x i16> %index, <vscale x 1 x i1> %mask, i64 %vl, i64 5)
+  ret void
+}
+
+
+declare void @llvm.riscv.vsuxseg8.triscv.vector.tuple_nxv4i8_8t.nxv4i8(target("riscv.vector.tuple", <vscale x 4 x i8>, 8), ptr, <vscale x 4 x i8>, i64, i64)
+declare void @llvm.riscv.vsuxseg8.mask.triscv.vector.tuple_nxv4i8_8t.nxv4i8.nxv4i1(target("riscv.vector.tuple", <vscale x 4 x i8>, 8), ptr, <vscale x 4 x i8>, <vscale x 4 x i1>, i64, i64)
+
+define void @test_vsuxseg8_nxv1i32_nxv1i16(target("riscv.vector.tuple", <vscale x 4 x i8>, 8) %val, ptr %base, <vscale x 1 x i16> %index, i64 %vl) sanitize_address {
+; CHECK-LABEL: @test_vsuxseg8_nxv1i32_nxv1i16(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i64, ptr @__asan_shadow_memory_dynamic_address, align 8
+; CHECK-NEXT:    tail call void @llvm.riscv.vsuxseg8.triscv.vector.tuple_nxv4i8_8t.p0.nxv1i16.i64(target("riscv.vector.tuple", <vscale x 4 x i8>, 8) [[VAL:%.*]], ptr [[BASE:%.*]], <vscale x 1 x i16> [[INDEX:%.*]], i64 [[VL:%.*]], i64 5)
+; CHECK-NEXT:    ret void
+;
+entry:
+  tail call void @llvm.riscv.vsuxseg8.triscv.vector.tuple_nxv4i8_8t.nxv1i16(target("riscv.vector.tuple", <vscale x 4 x i8>, 8) %val, ptr %base, <vscale x 1 x i16> %index, i64 %vl, i64 5)
+  ret void
+}
+
+define void @test_vsuxseg8_mask_nxv1i32_nxv1i16(target("riscv.vector.tuple", <vscale x 4 x i8>, 8) %val, ptr %base, <vscale x 1 x i16> %index, i64 %vl, <vscale x 1 x i1> %mask) sanitize_address {
+; CHECK-LABEL: @test_vsuxseg8_mask_nxv1i32_nxv1i16(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i64, ptr @__asan_shadow_memory_dynamic_address, align 8
+; CHECK-NEXT:    tail call void @llvm.riscv.vsoxseg8.mask.triscv.vector.tuple_nxv4i8_8t.p0.nxv1i16.nxv1i1.i64(target("riscv.vector.tuple", <vscale x 4 x i8>, 8) [[VAL:%.*]], ptr [[BASE:%.*]], <vscale x 1 x i16> [[INDEX:%.*]], <vscale x 1 x i1> [[MASK:%.*]], i64 [[VL:%.*]], i64 5)
+; CHECK-NEXT:    ret void
+;
+entry:
+  tail call void @llvm.riscv.vsoxseg8.mask.triscv.vector.tuple_nxv4i8_8t.nxv1i16.nxv1i1(target("riscv.vector.tuple", <vscale x 4 x i8>, 8) %val, ptr %base, <vscale x 1 x i16> %index, <vscale x 1 x i1> %mask, i64 %vl, i64 5)
+  ret void
+}
+

From d88808bb0dfe4cf15439f8d0b968fe01d7976b01 Mon Sep 17 00:00:00 2001
From: Hank Chang <hank.chang@sifive.com>
Date: Mon, 18 Aug 2025 15:05:26 +0800
Subject: [PATCH 2/9] Address Kito's comment

---
 llvm/lib/Target/AMDGPU/AMDGPUTargetTransformInfo.cpp | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/llvm/lib/Target/AMDGPU/AMDGPUTargetTransformInfo.cpp b/llvm/lib/Target/AMDGPU/AMDGPUTargetTransformInfo.cpp
index e99466cf81dde..763d301239b51 100644
--- a/llvm/lib/Target/AMDGPU/AMDGPUTargetTransformInfo.cpp
+++ b/llvm/lib/Target/AMDGPU/AMDGPUTargetTransformInfo.cpp
@@ -509,7 +509,7 @@ GCNTTIImpl::getTgtMemIntrinsic(IntrinsicInst *Inst) const {
     auto *Ordering = dyn_cast<ConstantInt>(Inst->getArgOperand(2));
     auto *Volatile = dyn_cast<ConstantInt>(Inst->getArgOperand(4));
     if (!Ordering || !Volatile)
-      return std::make_pair(std::nullopt, std::nullopt);
+      return std::make_pair(std::nullopt, std::nullopt); // Invalid
 
     unsigned OrderingVal = Ordering->getZExtValue();
     if (OrderingVal >

From 867b0d72d0115109a90721d97d9f19617bfed7c2 Mon Sep 17 00:00:00 2001
From: Hank Chang <hank.chang@sifive.com>
Date: Mon, 18 Aug 2025 15:52:28 +0800
Subject: [PATCH 3/9] Address Kito's 2nd comment

---
 llvm/lib/Transforms/Scalar/EarlyCSE.cpp | 12 +++++++-----
 1 file changed, 7 insertions(+), 5 deletions(-)

diff --git a/llvm/lib/Transforms/Scalar/EarlyCSE.cpp b/llvm/lib/Transforms/Scalar/EarlyCSE.cpp
index 5fb07026672bc..5e8d4b2faee32 100644
--- a/llvm/lib/Transforms/Scalar/EarlyCSE.cpp
+++ b/llvm/lib/Transforms/Scalar/EarlyCSE.cpp
@@ -808,6 +808,13 @@ class EarlyCSE {
       : Inst(Inst) {
       if (IntrinsicInst *II = dyn_cast<IntrinsicInst>(Inst)) {
         IntrID = II->getIntrinsicID();
+        std::pair<std::optional<MemIntrinsicInfo>,
+                  std::optional<SmallVector<InterestingMemoryOperand, 1>>>
+            MemInfo = TTI.getTgtMemIntrinsic(II);
+        if (MemInfo.first != std::nullopt) {
+          Info = *MemInfo.first;
+          return;
+        }
         if (isHandledNonTargetIntrinsic(IntrID)) {
           switch (IntrID) {
           case Intrinsic::masked_load:
@@ -832,11 +839,6 @@ class EarlyCSE {
             break;
           }
         }
-        std::pair<std::optional<MemIntrinsicInfo>,
-                  std::optional<SmallVector<InterestingMemoryOperand, 1>>>
-            MemInfo = TTI.getTgtMemIntrinsic(II);
-        if (MemInfo.first != std::nullopt)
-          Info = *MemInfo.first;
       }
     }
 

From 7e574c7fbf3da0046928738ee3f228740d332fcf Mon Sep 17 00:00:00 2001
From: Hank Chang <hank.chang@sifive.com>
Date: Wed, 20 Aug 2025 15:57:38 +0800
Subject: [PATCH 4/9] Move InterestingMemoryOperand from Transforms header to
 Analysis

---
 .../llvm/Analysis/InterestingMemoryOperand.h  | 55 +++++++++++++++++++
 .../llvm/Analysis/TargetTransformInfo.h       |  2 +-
 .../Instrumentation/AddressSanitizerCommon.h  | 33 +----------
 3 files changed, 57 insertions(+), 33 deletions(-)
 create mode 100644 llvm/include/llvm/Analysis/InterestingMemoryOperand.h

diff --git a/llvm/include/llvm/Analysis/InterestingMemoryOperand.h b/llvm/include/llvm/Analysis/InterestingMemoryOperand.h
new file mode 100644
index 0000000000000..1585c09228f26
--- /dev/null
+++ b/llvm/include/llvm/Analysis/InterestingMemoryOperand.h
@@ -0,0 +1,55 @@
+//===--------- Definition of the InterestingMemoryOperand class -*- C++
+//-*------------===//
+//
+// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
+// See https://llvm.org/LICENSE.txt for license information.
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+//
+//===----------------------------------------------------------------------===//
+//
+// This file defines InterestingMemoryOperand class that is used when getting
+// the information of a memory reference instruction.
+//
+//===----------------------------------------------------------------------===//
+
+#ifndef LLVM_ANALYSIS_INTERESTINGMEMORYOPERAND_H
+#define LLVM_ANALYSIS_INTERESTINGMEMORYOPERAND_H
+
+#include "llvm/IR/DataLayout.h"
+#include "llvm/IR/Instruction.h"
+#include "llvm/Support/TypeSize.h"
+
+namespace llvm {
+class InterestingMemoryOperand {
+public:
+  Use *PtrUse;
+  bool IsWrite;
+  Type *OpType;
+  TypeSize TypeStoreSize = TypeSize::getFixed(0);
+  MaybeAlign Alignment;
+  // The mask Value, if we're looking at a masked load/store.
+  Value *MaybeMask;
+  // The EVL Value, if we're looking at a vp intrinsic.
+  Value *MaybeEVL;
+  // The Stride Value, if we're looking at a strided load/store.
+  Value *MaybeStride;
+
+  InterestingMemoryOperand(Instruction *I, unsigned OperandNo, bool IsWrite,
+                           class Type *OpType, MaybeAlign Alignment,
+                           Value *MaybeMask = nullptr,
+                           Value *MaybeEVL = nullptr,
+                           Value *MaybeStride = nullptr)
+      : IsWrite(IsWrite), OpType(OpType), Alignment(Alignment),
+        MaybeMask(MaybeMask), MaybeEVL(MaybeEVL), MaybeStride(MaybeStride) {
+    const DataLayout &DL = I->getDataLayout();
+    TypeStoreSize = DL.getTypeStoreSizeInBits(OpType);
+    PtrUse = &I->getOperandUse(OperandNo);
+  }
+
+  Instruction *getInsn() { return cast<Instruction>(PtrUse->getUser()); }
+
+  Value *getPtr() { return PtrUse->get(); }
+};
+
+} // namespace llvm
+#endif
diff --git a/llvm/include/llvm/Analysis/TargetTransformInfo.h b/llvm/include/llvm/Analysis/TargetTransformInfo.h
index a5952b6d0d720..5dbffd70090a1 100644
--- a/llvm/include/llvm/Analysis/TargetTransformInfo.h
+++ b/llvm/include/llvm/Analysis/TargetTransformInfo.h
@@ -24,6 +24,7 @@
 #include "llvm/ADT/APInt.h"
 #include "llvm/ADT/ArrayRef.h"
 #include "llvm/Analysis/IVDescriptors.h"
+#include "llvm/Analysis/InterestingMemoryOperand.h"
 #include "llvm/IR/FMF.h"
 #include "llvm/IR/InstrTypes.h"
 #include "llvm/IR/PassManager.h"
@@ -32,7 +33,6 @@
 #include "llvm/Support/BranchProbability.h"
 #include "llvm/Support/Compiler.h"
 #include "llvm/Support/InstructionCost.h"
-#include "llvm/Transforms/Instrumentation/AddressSanitizerCommon.h"
 #include <functional>
 #include <optional>
 #include <utility>
diff --git a/llvm/include/llvm/Transforms/Instrumentation/AddressSanitizerCommon.h b/llvm/include/llvm/Transforms/Instrumentation/AddressSanitizerCommon.h
index 4e0e9010b42f0..6d2def3d2b72d 100644
--- a/llvm/include/llvm/Transforms/Instrumentation/AddressSanitizerCommon.h
+++ b/llvm/include/llvm/Transforms/Instrumentation/AddressSanitizerCommon.h
@@ -14,6 +14,7 @@
 #define LLVM_TRANSFORMS_INSTRUMENTATION_ADDRESSSANITIZERCOMMON_H
 
 #include "llvm/Analysis/CFG.h"
+#include "llvm/Analysis/InterestingMemoryOperand.h"
 #include "llvm/Analysis/PostDominators.h"
 #include "llvm/IR/Dominators.h"
 #include "llvm/IR/Instruction.h"
@@ -21,38 +22,6 @@
 #include "llvm/IR/Module.h"
 
 namespace llvm {
-
-class InterestingMemoryOperand {
-public:
-  Use *PtrUse;
-  bool IsWrite;
-  Type *OpType;
-  TypeSize TypeStoreSize = TypeSize::getFixed(0);
-  MaybeAlign Alignment;
-  // The mask Value, if we're looking at a masked load/store.
-  Value *MaybeMask;
-  // The EVL Value, if we're looking at a vp intrinsic.
-  Value *MaybeEVL;
-  // The Stride Value, if we're looking at a strided load/store.
-  Value *MaybeStride;
-
-  InterestingMemoryOperand(Instruction *I, unsigned OperandNo, bool IsWrite,
-                           class Type *OpType, MaybeAlign Alignment,
-                           Value *MaybeMask = nullptr,
-                           Value *MaybeEVL = nullptr,
-                           Value *MaybeStride = nullptr)
-      : IsWrite(IsWrite), OpType(OpType), Alignment(Alignment),
-        MaybeMask(MaybeMask), MaybeEVL(MaybeEVL), MaybeStride(MaybeStride) {
-    const DataLayout &DL = I->getDataLayout();
-    TypeStoreSize = DL.getTypeStoreSizeInBits(OpType);
-    PtrUse = &I->getOperandUse(OperandNo);
-  }
-
-  Instruction *getInsn() { return cast<Instruction>(PtrUse->getUser()); }
-
-  Value *getPtr() { return PtrUse->get(); }
-};
-
 // Get AddressSanitizer parameters.
 void getAddressSanitizerParams(const Triple &TargetTriple, int LongSize,
                                bool IsKasan, uint64_t *ShadowBase,

From f2abced960e7be54d8eb3aa6557dbc472c4f40f2 Mon Sep 17 00:00:00 2001
From: Hank Chang <hank.chang@sifive.com>
Date: Thu, 21 Aug 2025 16:22:08 +0800
Subject: [PATCH 5/9] Drop std::optional in getTgtMemIntrinsic return type

This commit drops std::optional in getTgtMemIntrinsic return type,
use nullptr in PtrVal to identify empty MemIntrinsicInfo and empty()
to identify empty SmallVector<InterestingMemoryOperand, 1>.
---
 llvm/include/llvm/Analysis/TargetTransformInfo.h  |  3 +--
 .../llvm/Analysis/TargetTransformInfoImpl.h       |  7 ++++---
 llvm/lib/Analysis/TargetTransformInfo.cpp         |  3 +--
 .../Target/AArch64/AArch64TargetTransformInfo.cpp |  8 ++++----
 .../Target/AArch64/AArch64TargetTransformInfo.h   |  3 +--
 .../Target/AMDGPU/AMDGPUTargetTransformInfo.cpp   | 15 ++++++++-------
 .../lib/Target/AMDGPU/AMDGPUTargetTransformInfo.h |  3 +--
 .../lib/Target/PowerPC/PPCTargetTransformInfo.cpp | 12 ++++++------
 llvm/lib/Target/PowerPC/PPCTargetTransformInfo.h  |  5 +++--
 .../lib/Target/RISCV/RISCVTargetTransformInfo.cpp | 13 ++++++-------
 llvm/lib/Target/RISCV/RISCVTargetTransformInfo.h  |  3 +--
 .../Instrumentation/AddressSanitizer.cpp          | 10 ++++------
 llvm/lib/Transforms/Scalar/EarlyCSE.cpp           |  7 +++----
 llvm/lib/Transforms/Scalar/LoopStrengthReduce.cpp | 14 ++++++--------
 14 files changed, 49 insertions(+), 57 deletions(-)

diff --git a/llvm/include/llvm/Analysis/TargetTransformInfo.h b/llvm/include/llvm/Analysis/TargetTransformInfo.h
index 5dbffd70090a1..3d50ed8af0661 100644
--- a/llvm/include/llvm/Analysis/TargetTransformInfo.h
+++ b/llvm/include/llvm/Analysis/TargetTransformInfo.h
@@ -1697,8 +1697,7 @@ class TargetTransformInfo {
   /// will contain additional information - whether the intrinsic may write
   /// or read to memory, volatility and the pointer.  Info is undefined
   /// if false is returned.
-  LLVM_ABI std::pair<std::optional<MemIntrinsicInfo>,
-                     std::optional<SmallVector<InterestingMemoryOperand, 1>>>
+  LLVM_ABI std::pair<MemIntrinsicInfo, SmallVector<InterestingMemoryOperand, 1>>
   getTgtMemIntrinsic(IntrinsicInst *Inst) const;
 
   /// \returns The maximum element size, in bytes, for an element
diff --git a/llvm/include/llvm/Analysis/TargetTransformInfoImpl.h b/llvm/include/llvm/Analysis/TargetTransformInfoImpl.h
index b6ed3097f7133..08a3edcf35a1e 100644
--- a/llvm/include/llvm/Analysis/TargetTransformInfoImpl.h
+++ b/llvm/include/llvm/Analysis/TargetTransformInfoImpl.h
@@ -975,10 +975,11 @@ class TargetTransformInfoImplBase {
     return 0;
   }
 
-  virtual std::pair<std::optional<MemIntrinsicInfo>,
-                    std::optional<SmallVector<InterestingMemoryOperand, 1>>>
+  virtual std::pair<MemIntrinsicInfo, SmallVector<InterestingMemoryOperand, 1>>
   getTgtMemIntrinsic(IntrinsicInst *Inst) const {
-    return std::make_pair(std::nullopt, std::nullopt);
+    MemIntrinsicInfo Info;
+    SmallVector<InterestingMemoryOperand, 1> Interesting;
+    return std::make_pair(Info, Interesting);
   }
 
   virtual unsigned getAtomicMemIntrinsicMaxElementSize() const {
diff --git a/llvm/lib/Analysis/TargetTransformInfo.cpp b/llvm/lib/Analysis/TargetTransformInfo.cpp
index cb810601ee1f8..88e5437185efa 100644
--- a/llvm/lib/Analysis/TargetTransformInfo.cpp
+++ b/llvm/lib/Analysis/TargetTransformInfo.cpp
@@ -1286,8 +1286,7 @@ TargetTransformInfo::getCostOfKeepingLiveOverCall(ArrayRef<Type *> Tys) const {
   return TTIImpl->getCostOfKeepingLiveOverCall(Tys);
 }
 
-std::pair<std::optional<MemIntrinsicInfo>,
-          std::optional<SmallVector<InterestingMemoryOperand, 1>>>
+std::pair<MemIntrinsicInfo, SmallVector<InterestingMemoryOperand, 1>>
 TargetTransformInfo::getTgtMemIntrinsic(IntrinsicInst *Inst) const {
   return TTIImpl->getTgtMemIntrinsic(Inst);
 }
diff --git a/llvm/lib/Target/AArch64/AArch64TargetTransformInfo.cpp b/llvm/lib/Target/AArch64/AArch64TargetTransformInfo.cpp
index e9eba8311ebd5..c6411ec6f72a2 100644
--- a/llvm/lib/Target/AArch64/AArch64TargetTransformInfo.cpp
+++ b/llvm/lib/Target/AArch64/AArch64TargetTransformInfo.cpp
@@ -5125,10 +5125,10 @@ Value *AArch64TTIImpl::getOrCreateResultFromMemIntrinsic(IntrinsicInst *Inst,
   }
 }
 
-std::pair<std::optional<MemIntrinsicInfo>,
-          std::optional<SmallVector<InterestingMemoryOperand, 1>>>
+std::pair<MemIntrinsicInfo, SmallVector<InterestingMemoryOperand, 1>>
 AArch64TTIImpl::getTgtMemIntrinsic(IntrinsicInst *Inst) const {
   MemIntrinsicInfo Info;
+  SmallVector<InterestingMemoryOperand, 1> Interesting;
   switch (Inst->getIntrinsicID()) {
   default:
     break;
@@ -5150,7 +5150,7 @@ AArch64TTIImpl::getTgtMemIntrinsic(IntrinsicInst *Inst) const {
 
   switch (Inst->getIntrinsicID()) {
   default:
-    return std::make_pair(std::nullopt, std::nullopt);
+    break;
   case Intrinsic::aarch64_neon_ld2:
   case Intrinsic::aarch64_neon_st2:
     Info.MatchingId = VECTOR_LDST_TWO_ELEMENTS;
@@ -5164,7 +5164,7 @@ AArch64TTIImpl::getTgtMemIntrinsic(IntrinsicInst *Inst) const {
     Info.MatchingId = VECTOR_LDST_FOUR_ELEMENTS;
     break;
   }
-  return std::make_pair(Info, std::nullopt);
+  return std::make_pair(Info, Interesting);
 }
 
 /// See if \p I should be considered for address type promotion. We check if \p
diff --git a/llvm/lib/Target/AArch64/AArch64TargetTransformInfo.h b/llvm/lib/Target/AArch64/AArch64TargetTransformInfo.h
index 6ade1b3d08405..2dd3fcfcd20dd 100644
--- a/llvm/lib/Target/AArch64/AArch64TargetTransformInfo.h
+++ b/llvm/lib/Target/AArch64/AArch64TargetTransformInfo.h
@@ -273,8 +273,7 @@ class AArch64TTIImpl final : public BasicTTIImplBase<AArch64TTIImpl> {
   getOrCreateResultFromMemIntrinsic(IntrinsicInst *Inst, Type *ExpectedType,
                                     bool CanCreate = true) const override;
 
-  std::pair<std::optional<MemIntrinsicInfo>,
-            std::optional<SmallVector<InterestingMemoryOperand, 1>>>
+  std::pair<MemIntrinsicInfo, SmallVector<InterestingMemoryOperand, 1>>
   getTgtMemIntrinsic(IntrinsicInst *Inst) const override;
 
   bool isElementTypeLegalForScalableVector(Type *Ty) const override {
diff --git a/llvm/lib/Target/AMDGPU/AMDGPUTargetTransformInfo.cpp b/llvm/lib/Target/AMDGPU/AMDGPUTargetTransformInfo.cpp
index 763d301239b51..97afe89553600 100644
--- a/llvm/lib/Target/AMDGPU/AMDGPUTargetTransformInfo.cpp
+++ b/llvm/lib/Target/AMDGPU/AMDGPUTargetTransformInfo.cpp
@@ -500,33 +500,34 @@ unsigned GCNTTIImpl::getMaxInterleaveFactor(ElementCount VF) const {
   return 8;
 }
 
-std::pair<std::optional<MemIntrinsicInfo>,
-          std::optional<SmallVector<InterestingMemoryOperand, 1>>>
+std::pair<MemIntrinsicInfo, SmallVector<InterestingMemoryOperand, 1>>
 GCNTTIImpl::getTgtMemIntrinsic(IntrinsicInst *Inst) const {
+  MemIntrinsicInfo Info;
+  SmallVector<InterestingMemoryOperand, 1> Interesting;
   switch (Inst->getIntrinsicID()) {
   case Intrinsic::amdgcn_ds_ordered_add:
   case Intrinsic::amdgcn_ds_ordered_swap: {
     auto *Ordering = dyn_cast<ConstantInt>(Inst->getArgOperand(2));
     auto *Volatile = dyn_cast<ConstantInt>(Inst->getArgOperand(4));
     if (!Ordering || !Volatile)
-      return std::make_pair(std::nullopt, std::nullopt); // Invalid
+      break; // Invalid
 
     unsigned OrderingVal = Ordering->getZExtValue();
     if (OrderingVal >
         static_cast<unsigned>(AtomicOrdering::SequentiallyConsistent))
-      return std::make_pair(std::nullopt, std::nullopt);
+      break;
 
-    MemIntrinsicInfo Info;
     Info.PtrVal = Inst->getArgOperand(0);
     Info.Ordering = static_cast<AtomicOrdering>(OrderingVal);
     Info.ReadMem = true;
     Info.WriteMem = true;
     Info.IsVolatile = !Volatile->isZero();
-    return std::make_pair(Info, std::nullopt);
+    break;
   }
   default:
-    return std::make_pair(std::nullopt, std::nullopt);
+    break;
   }
+  return std::make_pair(Info, Interesting);
 }
 
 InstructionCost GCNTTIImpl::getArithmeticInstrCost(
diff --git a/llvm/lib/Target/AMDGPU/AMDGPUTargetTransformInfo.h b/llvm/lib/Target/AMDGPU/AMDGPUTargetTransformInfo.h
index 9195915758dba..29c8f2528cebf 100644
--- a/llvm/lib/Target/AMDGPU/AMDGPUTargetTransformInfo.h
+++ b/llvm/lib/Target/AMDGPU/AMDGPUTargetTransformInfo.h
@@ -151,8 +151,7 @@ class GCNTTIImpl final : public BasicTTIImplBase<GCNTTIImpl> {
       std::optional<uint32_t> AtomicCpySize) const override;
   unsigned getMaxInterleaveFactor(ElementCount VF) const override;
 
-  std::pair<std::optional<MemIntrinsicInfo>,
-            std::optional<SmallVector<InterestingMemoryOperand, 1>>>
+  std::pair<MemIntrinsicInfo, SmallVector<InterestingMemoryOperand, 1>>
   getTgtMemIntrinsic(IntrinsicInst *Inst) const override;
 
   InstructionCost getArithmeticInstrCost(
diff --git a/llvm/lib/Target/PowerPC/PPCTargetTransformInfo.cpp b/llvm/lib/Target/PowerPC/PPCTargetTransformInfo.cpp
index 28a8ae31d708a..c50a741cee858 100644
--- a/llvm/lib/Target/PowerPC/PPCTargetTransformInfo.cpp
+++ b/llvm/lib/Target/PowerPC/PPCTargetTransformInfo.cpp
@@ -975,10 +975,10 @@ bool PPCTTIImpl::shouldBuildRelLookupTables() const {
   return BaseT::shouldBuildRelLookupTables();
 }
 
-std::pair<std::optional<MemIntrinsicInfo>,
-          std::optional<SmallVector<InterestingMemoryOperand, 1>>>
+std::pair<MemIntrinsicInfo, SmallVector<InterestingMemoryOperand, 1>>
 PPCTTIImpl::getTgtMemIntrinsic(IntrinsicInst *Inst) const {
   MemIntrinsicInfo Info;
+  SmallVector<InterestingMemoryOperand, 1> Interesting;
   switch (Inst->getIntrinsicID()) {
   case Intrinsic::ppc_altivec_lvx:
   case Intrinsic::ppc_altivec_lvxl:
@@ -995,7 +995,7 @@ PPCTTIImpl::getTgtMemIntrinsic(IntrinsicInst *Inst) const {
     Info.PtrVal = Inst->getArgOperand(0);
     Info.ReadMem = true;
     Info.WriteMem = false;
-    return std::make_pair(Info, std::nullopt);
+    break;
   }
   case Intrinsic::ppc_altivec_stvx:
   case Intrinsic::ppc_altivec_stvxl:
@@ -1012,7 +1012,7 @@ PPCTTIImpl::getTgtMemIntrinsic(IntrinsicInst *Inst) const {
     Info.PtrVal = Inst->getArgOperand(1);
     Info.ReadMem = false;
     Info.WriteMem = true;
-    return std::make_pair(Info, std::nullopt);
+    break;
   }
   case Intrinsic::ppc_stbcx:
   case Intrinsic::ppc_sthcx:
@@ -1021,12 +1021,12 @@ PPCTTIImpl::getTgtMemIntrinsic(IntrinsicInst *Inst) const {
     Info.PtrVal = Inst->getArgOperand(0);
     Info.ReadMem = false;
     Info.WriteMem = true;
-    return std::make_pair(Info, std::nullopt);
+    break;
   }
   default:
     break;
   }
-  return std::make_pair(std::nullopt, std::nullopt);
+  return std::make_pair(Info, Interesting);
 }
 
 bool PPCTTIImpl::supportsTailCallFor(const CallBase *CB) const {
diff --git a/llvm/lib/Target/PowerPC/PPCTargetTransformInfo.h b/llvm/lib/Target/PowerPC/PPCTargetTransformInfo.h
index 923931f14bc77..410beaea83ebe 100644
--- a/llvm/lib/Target/PowerPC/PPCTargetTransformInfo.h
+++ b/llvm/lib/Target/PowerPC/PPCTargetTransformInfo.h
@@ -69,9 +69,10 @@ class PPCTTIImpl final : public BasicTTIImplBase<PPCTTIImpl> {
   bool canSaveCmp(Loop *L, BranchInst **BI, ScalarEvolution *SE, LoopInfo *LI,
                   DominatorTree *DT, AssumptionCache *AC,
                   TargetLibraryInfo *LibInfo) const override;
-  std::pair<std::optional<MemIntrinsicInfo>,
-            std::optional<SmallVector<InterestingMemoryOperand, 1>>>
+
+  std::pair<MemIntrinsicInfo, SmallVector<InterestingMemoryOperand, 1>>
   getTgtMemIntrinsic(IntrinsicInst *Inst) const override;
+
   void getUnrollingPreferences(Loop *L, ScalarEvolution &SE,
                                TTI::UnrollingPreferences &UP,
                                OptimizationRemarkEmitter *ORE) const override;
diff --git a/llvm/lib/Target/RISCV/RISCVTargetTransformInfo.cpp b/llvm/lib/Target/RISCV/RISCVTargetTransformInfo.cpp
index 7ee153278208d..19688553b9b8d 100644
--- a/llvm/lib/Target/RISCV/RISCVTargetTransformInfo.cpp
+++ b/llvm/lib/Target/RISCV/RISCVTargetTransformInfo.cpp
@@ -2674,13 +2674,14 @@ void RISCVTTIImpl::getPeelingPreferences(Loop *L, ScalarEvolution &SE,
   BaseT::getPeelingPreferences(L, SE, PP);
 }
 
-std::pair<std::optional<MemIntrinsicInfo>,
-          std::optional<SmallVector<InterestingMemoryOperand, 1>>>
+std::pair<MemIntrinsicInfo, SmallVector<InterestingMemoryOperand, 1>>
 RISCVTTIImpl::getTgtMemIntrinsic(IntrinsicInst *Inst) const {
   const DataLayout &DL = getDataLayout();
   Intrinsic::ID IID = Inst->getIntrinsicID();
   LLVMContext &C = Inst->getContext();
   bool HasMask = false;
+  MemIntrinsicInfo Info;
+  SmallVector<InterestingMemoryOperand, 1> Interesting;
   switch (IID) {
   case Intrinsic::riscv_vle_mask:
   case Intrinsic::riscv_vse_mask:
@@ -2693,7 +2694,6 @@ RISCVTTIImpl::getTgtMemIntrinsic(IntrinsicInst *Inst) const {
     // riscv_vle_mask(merge, ptr, mask, vl, policy)
     // riscv_vse(val, ptr, vl)
     // riscv_vse_mask(val, ptr, mask, vl, policy)
-    SmallVector<InterestingMemoryOperand, 1> Interesting;
     bool IsWrite = Inst->getType()->isVoidTy();
     Type *Ty = IsWrite ? Inst->getArgOperand(0)->getType() : Inst->getType();
     const auto *RVVIInfo = RISCVVIntrinsicsTable::getRISCVVIntrinsicInfo(IID);
@@ -2708,7 +2708,7 @@ RISCVTTIImpl::getTgtMemIntrinsic(IntrinsicInst *Inst) const {
     Value *EVL = Inst->getArgOperand(VLIndex);
     Interesting.emplace_back(Inst, PtrOperandNo, IsWrite, Ty, Alignment, Mask,
                              EVL);
-    return std::make_pair(std::nullopt, Interesting);
+    break;
   }
   case Intrinsic::riscv_vlse_mask:
   case Intrinsic::riscv_vsse_mask:
@@ -2721,7 +2721,6 @@ RISCVTTIImpl::getTgtMemIntrinsic(IntrinsicInst *Inst) const {
     // riscv_vlse_mask(merge, ptr, stride, mask, vl, policy)
     // riscv_vsse(val, ptr, stride, vl)
     // riscv_vsse_mask(val, ptr, stride, mask, vl, policy)
-    SmallVector<InterestingMemoryOperand, 1> Interesting;
     bool IsWrite = Inst->getType()->isVoidTy();
     Type *Ty = IsWrite ? Inst->getArgOperand(0)->getType() : Inst->getType();
     const auto *RVVIInfo = RISCVVIntrinsicsTable::getRISCVVIntrinsicInfo(IID);
@@ -2747,10 +2746,10 @@ RISCVTTIImpl::getTgtMemIntrinsic(IntrinsicInst *Inst) const {
     Value *EVL = Inst->getArgOperand(VLIndex);
     Interesting.emplace_back(Inst, PtrOperandNo, IsWrite, Ty, Alignment, Mask,
                              EVL, Stride);
-    return std::make_pair(std::nullopt, Interesting);
+    break;
   }
   }
-  return std::make_pair(std::nullopt, std::nullopt);
+  return std::make_pair(Info, Interesting);
 }
 
 unsigned RISCVTTIImpl::getRegUsageForType(Type *Ty) const {
diff --git a/llvm/lib/Target/RISCV/RISCVTargetTransformInfo.h b/llvm/lib/Target/RISCV/RISCVTargetTransformInfo.h
index d8a829ac21b64..6a784b14349a3 100644
--- a/llvm/lib/Target/RISCV/RISCVTargetTransformInfo.h
+++ b/llvm/lib/Target/RISCV/RISCVTargetTransformInfo.h
@@ -158,8 +158,7 @@ class RISCVTTIImpl final : public BasicTTIImplBase<RISCVTTIImpl> {
   void getPeelingPreferences(Loop *L, ScalarEvolution &SE,
                              TTI::PeelingPreferences &PP) const override;
 
-  std::pair<std::optional<MemIntrinsicInfo>,
-            std::optional<SmallVector<InterestingMemoryOperand, 1>>>
+  std::pair<MemIntrinsicInfo, SmallVector<InterestingMemoryOperand, 1>>
   getTgtMemIntrinsic(IntrinsicInst *Inst) const override;
 
   unsigned getMinVectorRegisterBitWidth() const override {
diff --git a/llvm/lib/Transforms/Instrumentation/AddressSanitizer.cpp b/llvm/lib/Transforms/Instrumentation/AddressSanitizer.cpp
index f413c5eb7a2bb..a91c76e3ed7ba 100644
--- a/llvm/lib/Transforms/Instrumentation/AddressSanitizer.cpp
+++ b/llvm/lib/Transforms/Instrumentation/AddressSanitizer.cpp
@@ -1573,12 +1573,10 @@ void AddressSanitizer::getInterestingMemoryOperands(
     }
     default:
       if (auto *II = dyn_cast<IntrinsicInst>(I)) {
-        std::pair<std::optional<MemIntrinsicInfo>,
-                  std::optional<SmallVector<InterestingMemoryOperand, 1>>>
-            MemInfo;
-        MemInfo = TTI->getTgtMemIntrinsic(II);
-        if (MemInfo.second != std::nullopt)
-          Interesting = *MemInfo.second;
+        std::pair<MemIntrinsicInfo, SmallVector<InterestingMemoryOperand, 1>>
+            MemInfo = TTI->getTgtMemIntrinsic(II);
+        if (!MemInfo.second.empty())
+          Interesting = MemInfo.second;
         return;
       }
       for (unsigned ArgNo = 0; ArgNo < CI->arg_size(); ArgNo++) {
diff --git a/llvm/lib/Transforms/Scalar/EarlyCSE.cpp b/llvm/lib/Transforms/Scalar/EarlyCSE.cpp
index 5e8d4b2faee32..975c97999bdf4 100644
--- a/llvm/lib/Transforms/Scalar/EarlyCSE.cpp
+++ b/llvm/lib/Transforms/Scalar/EarlyCSE.cpp
@@ -808,11 +808,10 @@ class EarlyCSE {
       : Inst(Inst) {
       if (IntrinsicInst *II = dyn_cast<IntrinsicInst>(Inst)) {
         IntrID = II->getIntrinsicID();
-        std::pair<std::optional<MemIntrinsicInfo>,
-                  std::optional<SmallVector<InterestingMemoryOperand, 1>>>
+        std::pair<MemIntrinsicInfo, SmallVector<InterestingMemoryOperand, 1>>
             MemInfo = TTI.getTgtMemIntrinsic(II);
-        if (MemInfo.first != std::nullopt) {
-          Info = *MemInfo.first;
+        if (MemInfo.first.PtrVal != nullptr) {
+          Info = MemInfo.first;
           return;
         }
         if (isHandledNonTargetIntrinsic(IntrID)) {
diff --git a/llvm/lib/Transforms/Scalar/LoopStrengthReduce.cpp b/llvm/lib/Transforms/Scalar/LoopStrengthReduce.cpp
index 049087a59abe6..e3ad2965917f6 100644
--- a/llvm/lib/Transforms/Scalar/LoopStrengthReduce.cpp
+++ b/llvm/lib/Transforms/Scalar/LoopStrengthReduce.cpp
@@ -1018,11 +1018,10 @@ static bool isAddressUse(const TargetTransformInfo &TTI,
         isAddress = true;
       break;
     default: {
-      std::pair<std::optional<MemIntrinsicInfo>,
-                std::optional<SmallVector<InterestingMemoryOperand, 1>>>
+      std::pair<MemIntrinsicInfo, SmallVector<InterestingMemoryOperand, 1>>
           MemInfo = TTI.getTgtMemIntrinsic(II);
-      if (MemInfo.first != std::nullopt) {
-        if (MemInfo.first->PtrVal == OperandVal)
+      if (MemInfo.first.PtrVal != nullptr) {
+        if (MemInfo.first.PtrVal == OperandVal)
           isAddress = true;
       }
     }
@@ -1076,12 +1075,11 @@ static MemAccessTy getAccessType(const TargetTransformInfo &TTI,
           II->getArgOperand(1)->getType()->getPointerAddressSpace();
       break;
     default: {
-      std::pair<std::optional<MemIntrinsicInfo>,
-                std::optional<SmallVector<InterestingMemoryOperand, 1>>>
+      std::pair<MemIntrinsicInfo, SmallVector<InterestingMemoryOperand, 1>>
           MemInfo = TTI.getTgtMemIntrinsic(II);
-      if (MemInfo.first != std::nullopt && MemInfo.first->PtrVal) {
+      if (MemInfo.first.PtrVal) {
         AccessTy.AddrSpace =
-            MemInfo.first->PtrVal->getType()->getPointerAddressSpace();
+            MemInfo.first.PtrVal->getType()->getPointerAddressSpace();
       }
 
       break;

From e89d6ba9f1f61cc2d568d8d563d5cc70deba3d63 Mon Sep 17 00:00:00 2001
From: Hank Chang <hank.chang@sifive.com>
Date: Thu, 28 Aug 2025 22:00:57 +0800
Subject: [PATCH 6/9] [TTI] Make SmallVector<InterestingMemoryOperand> a member
 of MemIntrinsicInfo

This commit make SmallVector<InterestingMemoryOperand> a member of MemIntrinsicInfo so that
we can keep the original interface of getTgtMemIntrinsic.
---
 .../llvm/Analysis/InterestingMemoryOperand.h  |  3 +--
 .../llvm/Analysis/TargetTransformInfo.h       |  6 ++++--
 .../llvm/Analysis/TargetTransformInfoImpl.h   |  8 +++-----
 llvm/lib/Analysis/TargetTransformInfo.cpp     |  6 +++---
 .../AArch64/AArch64TargetTransformInfo.cpp    | 10 ++++------
 .../AArch64/AArch64TargetTransformInfo.h      |  4 ++--
 .../AMDGPU/AMDGPUTargetTransformInfo.cpp      | 15 ++++++--------
 .../Target/AMDGPU/AMDGPUTargetTransformInfo.h |  4 ++--
 .../Target/PowerPC/PPCTargetTransformInfo.cpp | 14 ++++++-------
 .../Target/PowerPC/PPCTargetTransformInfo.h   |  4 ++--
 .../Target/RISCV/RISCVTargetTransformInfo.cpp | 20 +++++++++----------
 .../Target/RISCV/RISCVTargetTransformInfo.h   |  4 ++--
 .../Instrumentation/AddressSanitizer.cpp      |  9 ++++-----
 llvm/lib/Transforms/Scalar/EarlyCSE.cpp       |  5 +----
 .../Transforms/Scalar/LoopStrengthReduce.cpp  | 15 ++++++--------
 15 files changed, 55 insertions(+), 72 deletions(-)

diff --git a/llvm/include/llvm/Analysis/InterestingMemoryOperand.h b/llvm/include/llvm/Analysis/InterestingMemoryOperand.h
index 1585c09228f26..69960e6f012d4 100644
--- a/llvm/include/llvm/Analysis/InterestingMemoryOperand.h
+++ b/llvm/include/llvm/Analysis/InterestingMemoryOperand.h
@@ -1,5 +1,4 @@
-//===--------- Definition of the InterestingMemoryOperand class -*- C++
-//-*------------===//
+//===- InterestingMemoryOperand.h -------------------------------*- C++ -*-===//
 //
 // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
 // See https://llvm.org/LICENSE.txt for license information.
diff --git a/llvm/include/llvm/Analysis/TargetTransformInfo.h b/llvm/include/llvm/Analysis/TargetTransformInfo.h
index 3d50ed8af0661..8cab4bcaa9056 100644
--- a/llvm/include/llvm/Analysis/TargetTransformInfo.h
+++ b/llvm/include/llvm/Analysis/TargetTransformInfo.h
@@ -88,6 +88,8 @@ struct MemIntrinsicInfo {
   bool WriteMem = false;
   bool IsVolatile = false;
 
+  SmallVector<InterestingMemoryOperand, 1> Interesting;
+
   bool isUnordered() const {
     return (Ordering == AtomicOrdering::NotAtomic ||
             Ordering == AtomicOrdering::Unordered) &&
@@ -1697,8 +1699,8 @@ class TargetTransformInfo {
   /// will contain additional information - whether the intrinsic may write
   /// or read to memory, volatility and the pointer.  Info is undefined
   /// if false is returned.
-  LLVM_ABI std::pair<MemIntrinsicInfo, SmallVector<InterestingMemoryOperand, 1>>
-  getTgtMemIntrinsic(IntrinsicInst *Inst) const;
+  LLVM_ABI bool getTgtMemIntrinsic(IntrinsicInst *Inst,
+                                   MemIntrinsicInfo &Info) const;
 
   /// \returns The maximum element size, in bytes, for an element
   /// unordered-atomic memory intrinsic.
diff --git a/llvm/include/llvm/Analysis/TargetTransformInfoImpl.h b/llvm/include/llvm/Analysis/TargetTransformInfoImpl.h
index 08a3edcf35a1e..200cbafbaa6e2 100644
--- a/llvm/include/llvm/Analysis/TargetTransformInfoImpl.h
+++ b/llvm/include/llvm/Analysis/TargetTransformInfoImpl.h
@@ -975,11 +975,9 @@ class TargetTransformInfoImplBase {
     return 0;
   }
 
-  virtual std::pair<MemIntrinsicInfo, SmallVector<InterestingMemoryOperand, 1>>
-  getTgtMemIntrinsic(IntrinsicInst *Inst) const {
-    MemIntrinsicInfo Info;
-    SmallVector<InterestingMemoryOperand, 1> Interesting;
-    return std::make_pair(Info, Interesting);
+  virtual bool getTgtMemIntrinsic(IntrinsicInst *Inst,
+                                  MemIntrinsicInfo &Info) const {
+    return false;
   }
 
   virtual unsigned getAtomicMemIntrinsicMaxElementSize() const {
diff --git a/llvm/lib/Analysis/TargetTransformInfo.cpp b/llvm/lib/Analysis/TargetTransformInfo.cpp
index 88e5437185efa..d21fb2dd238f7 100644
--- a/llvm/lib/Analysis/TargetTransformInfo.cpp
+++ b/llvm/lib/Analysis/TargetTransformInfo.cpp
@@ -1286,9 +1286,9 @@ TargetTransformInfo::getCostOfKeepingLiveOverCall(ArrayRef<Type *> Tys) const {
   return TTIImpl->getCostOfKeepingLiveOverCall(Tys);
 }
 
-std::pair<MemIntrinsicInfo, SmallVector<InterestingMemoryOperand, 1>>
-TargetTransformInfo::getTgtMemIntrinsic(IntrinsicInst *Inst) const {
-  return TTIImpl->getTgtMemIntrinsic(Inst);
+bool TargetTransformInfo::getTgtMemIntrinsic(IntrinsicInst *Inst,
+                                             MemIntrinsicInfo &Info) const {
+  return TTIImpl->getTgtMemIntrinsic(Inst, Info);
 }
 
 unsigned TargetTransformInfo::getAtomicMemIntrinsicMaxElementSize() const {
diff --git a/llvm/lib/Target/AArch64/AArch64TargetTransformInfo.cpp b/llvm/lib/Target/AArch64/AArch64TargetTransformInfo.cpp
index c6411ec6f72a2..3042251cf754d 100644
--- a/llvm/lib/Target/AArch64/AArch64TargetTransformInfo.cpp
+++ b/llvm/lib/Target/AArch64/AArch64TargetTransformInfo.cpp
@@ -5125,10 +5125,8 @@ Value *AArch64TTIImpl::getOrCreateResultFromMemIntrinsic(IntrinsicInst *Inst,
   }
 }
 
-std::pair<MemIntrinsicInfo, SmallVector<InterestingMemoryOperand, 1>>
-AArch64TTIImpl::getTgtMemIntrinsic(IntrinsicInst *Inst) const {
-  MemIntrinsicInfo Info;
-  SmallVector<InterestingMemoryOperand, 1> Interesting;
+bool AArch64TTIImpl::getTgtMemIntrinsic(IntrinsicInst *Inst,
+                                        MemIntrinsicInfo &Info) const {
   switch (Inst->getIntrinsicID()) {
   default:
     break;
@@ -5150,7 +5148,7 @@ AArch64TTIImpl::getTgtMemIntrinsic(IntrinsicInst *Inst) const {
 
   switch (Inst->getIntrinsicID()) {
   default:
-    break;
+    return false;
   case Intrinsic::aarch64_neon_ld2:
   case Intrinsic::aarch64_neon_st2:
     Info.MatchingId = VECTOR_LDST_TWO_ELEMENTS;
@@ -5164,7 +5162,7 @@ AArch64TTIImpl::getTgtMemIntrinsic(IntrinsicInst *Inst) const {
     Info.MatchingId = VECTOR_LDST_FOUR_ELEMENTS;
     break;
   }
-  return std::make_pair(Info, Interesting);
+  return true;
 }
 
 /// See if \p I should be considered for address type promotion. We check if \p
diff --git a/llvm/lib/Target/AArch64/AArch64TargetTransformInfo.h b/llvm/lib/Target/AArch64/AArch64TargetTransformInfo.h
index 2dd3fcfcd20dd..9c96fdd427814 100644
--- a/llvm/lib/Target/AArch64/AArch64TargetTransformInfo.h
+++ b/llvm/lib/Target/AArch64/AArch64TargetTransformInfo.h
@@ -273,8 +273,8 @@ class AArch64TTIImpl final : public BasicTTIImplBase<AArch64TTIImpl> {
   getOrCreateResultFromMemIntrinsic(IntrinsicInst *Inst, Type *ExpectedType,
                                     bool CanCreate = true) const override;
 
-  std::pair<MemIntrinsicInfo, SmallVector<InterestingMemoryOperand, 1>>
-  getTgtMemIntrinsic(IntrinsicInst *Inst) const override;
+  bool getTgtMemIntrinsic(IntrinsicInst *Inst,
+                          MemIntrinsicInfo &Info) const override;
 
   bool isElementTypeLegalForScalableVector(Type *Ty) const override {
     if (Ty->isPointerTy())
diff --git a/llvm/lib/Target/AMDGPU/AMDGPUTargetTransformInfo.cpp b/llvm/lib/Target/AMDGPU/AMDGPUTargetTransformInfo.cpp
index 97afe89553600..dadf10c4296b3 100644
--- a/llvm/lib/Target/AMDGPU/AMDGPUTargetTransformInfo.cpp
+++ b/llvm/lib/Target/AMDGPU/AMDGPUTargetTransformInfo.cpp
@@ -500,34 +500,31 @@ unsigned GCNTTIImpl::getMaxInterleaveFactor(ElementCount VF) const {
   return 8;
 }
 
-std::pair<MemIntrinsicInfo, SmallVector<InterestingMemoryOperand, 1>>
-GCNTTIImpl::getTgtMemIntrinsic(IntrinsicInst *Inst) const {
-  MemIntrinsicInfo Info;
-  SmallVector<InterestingMemoryOperand, 1> Interesting;
+bool GCNTTIImpl::getTgtMemIntrinsic(IntrinsicInst *Inst,
+                                    MemIntrinsicInfo &Info) const {
   switch (Inst->getIntrinsicID()) {
   case Intrinsic::amdgcn_ds_ordered_add:
   case Intrinsic::amdgcn_ds_ordered_swap: {
     auto *Ordering = dyn_cast<ConstantInt>(Inst->getArgOperand(2));
     auto *Volatile = dyn_cast<ConstantInt>(Inst->getArgOperand(4));
     if (!Ordering || !Volatile)
-      break; // Invalid
+      return false; // Invalid
 
     unsigned OrderingVal = Ordering->getZExtValue();
     if (OrderingVal >
         static_cast<unsigned>(AtomicOrdering::SequentiallyConsistent))
-      break;
+      return false;
 
     Info.PtrVal = Inst->getArgOperand(0);
     Info.Ordering = static_cast<AtomicOrdering>(OrderingVal);
     Info.ReadMem = true;
     Info.WriteMem = true;
     Info.IsVolatile = !Volatile->isZero();
-    break;
+    return true;
   }
   default:
-    break;
+    return false;
   }
-  return std::make_pair(Info, Interesting);
 }
 
 InstructionCost GCNTTIImpl::getArithmeticInstrCost(
diff --git a/llvm/lib/Target/AMDGPU/AMDGPUTargetTransformInfo.h b/llvm/lib/Target/AMDGPU/AMDGPUTargetTransformInfo.h
index 29c8f2528cebf..20da8344c9d37 100644
--- a/llvm/lib/Target/AMDGPU/AMDGPUTargetTransformInfo.h
+++ b/llvm/lib/Target/AMDGPU/AMDGPUTargetTransformInfo.h
@@ -151,8 +151,8 @@ class GCNTTIImpl final : public BasicTTIImplBase<GCNTTIImpl> {
       std::optional<uint32_t> AtomicCpySize) const override;
   unsigned getMaxInterleaveFactor(ElementCount VF) const override;
 
-  std::pair<MemIntrinsicInfo, SmallVector<InterestingMemoryOperand, 1>>
-  getTgtMemIntrinsic(IntrinsicInst *Inst) const override;
+  bool getTgtMemIntrinsic(IntrinsicInst *Inst,
+                          MemIntrinsicInfo &Info) const override;
 
   InstructionCost getArithmeticInstrCost(
       unsigned Opcode, Type *Ty, TTI::TargetCostKind CostKind,
diff --git a/llvm/lib/Target/PowerPC/PPCTargetTransformInfo.cpp b/llvm/lib/Target/PowerPC/PPCTargetTransformInfo.cpp
index c50a741cee858..2566c84b39302 100644
--- a/llvm/lib/Target/PowerPC/PPCTargetTransformInfo.cpp
+++ b/llvm/lib/Target/PowerPC/PPCTargetTransformInfo.cpp
@@ -975,10 +975,8 @@ bool PPCTTIImpl::shouldBuildRelLookupTables() const {
   return BaseT::shouldBuildRelLookupTables();
 }
 
-std::pair<MemIntrinsicInfo, SmallVector<InterestingMemoryOperand, 1>>
-PPCTTIImpl::getTgtMemIntrinsic(IntrinsicInst *Inst) const {
-  MemIntrinsicInfo Info;
-  SmallVector<InterestingMemoryOperand, 1> Interesting;
+bool PPCTTIImpl::getTgtMemIntrinsic(IntrinsicInst *Inst,
+                                    MemIntrinsicInfo &Info) const {
   switch (Inst->getIntrinsicID()) {
   case Intrinsic::ppc_altivec_lvx:
   case Intrinsic::ppc_altivec_lvxl:
@@ -995,7 +993,7 @@ PPCTTIImpl::getTgtMemIntrinsic(IntrinsicInst *Inst) const {
     Info.PtrVal = Inst->getArgOperand(0);
     Info.ReadMem = true;
     Info.WriteMem = false;
-    break;
+    return true;
   }
   case Intrinsic::ppc_altivec_stvx:
   case Intrinsic::ppc_altivec_stvxl:
@@ -1012,7 +1010,7 @@ PPCTTIImpl::getTgtMemIntrinsic(IntrinsicInst *Inst) const {
     Info.PtrVal = Inst->getArgOperand(1);
     Info.ReadMem = false;
     Info.WriteMem = true;
-    break;
+    return true;
   }
   case Intrinsic::ppc_stbcx:
   case Intrinsic::ppc_sthcx:
@@ -1021,12 +1019,12 @@ PPCTTIImpl::getTgtMemIntrinsic(IntrinsicInst *Inst) const {
     Info.PtrVal = Inst->getArgOperand(0);
     Info.ReadMem = false;
     Info.WriteMem = true;
-    break;
+    return true;
   }
   default:
     break;
   }
-  return std::make_pair(Info, Interesting);
+  return false;
 }
 
 bool PPCTTIImpl::supportsTailCallFor(const CallBase *CB) const {
diff --git a/llvm/lib/Target/PowerPC/PPCTargetTransformInfo.h b/llvm/lib/Target/PowerPC/PPCTargetTransformInfo.h
index 410beaea83ebe..4d935d11dab62 100644
--- a/llvm/lib/Target/PowerPC/PPCTargetTransformInfo.h
+++ b/llvm/lib/Target/PowerPC/PPCTargetTransformInfo.h
@@ -70,8 +70,8 @@ class PPCTTIImpl final : public BasicTTIImplBase<PPCTTIImpl> {
                   DominatorTree *DT, AssumptionCache *AC,
                   TargetLibraryInfo *LibInfo) const override;
 
-  std::pair<MemIntrinsicInfo, SmallVector<InterestingMemoryOperand, 1>>
-  getTgtMemIntrinsic(IntrinsicInst *Inst) const override;
+  bool getTgtMemIntrinsic(IntrinsicInst *Inst,
+                          MemIntrinsicInfo &Info) const override;
 
   void getUnrollingPreferences(Loop *L, ScalarEvolution &SE,
                                TTI::UnrollingPreferences &UP,
diff --git a/llvm/lib/Target/RISCV/RISCVTargetTransformInfo.cpp b/llvm/lib/Target/RISCV/RISCVTargetTransformInfo.cpp
index 19688553b9b8d..8b2675c22b820 100644
--- a/llvm/lib/Target/RISCV/RISCVTargetTransformInfo.cpp
+++ b/llvm/lib/Target/RISCV/RISCVTargetTransformInfo.cpp
@@ -2674,14 +2674,12 @@ void RISCVTTIImpl::getPeelingPreferences(Loop *L, ScalarEvolution &SE,
   BaseT::getPeelingPreferences(L, SE, PP);
 }
 
-std::pair<MemIntrinsicInfo, SmallVector<InterestingMemoryOperand, 1>>
-RISCVTTIImpl::getTgtMemIntrinsic(IntrinsicInst *Inst) const {
+bool RISCVTTIImpl::getTgtMemIntrinsic(IntrinsicInst *Inst,
+                                      MemIntrinsicInfo &Info) const {
   const DataLayout &DL = getDataLayout();
   Intrinsic::ID IID = Inst->getIntrinsicID();
   LLVMContext &C = Inst->getContext();
   bool HasMask = false;
-  MemIntrinsicInfo Info;
-  SmallVector<InterestingMemoryOperand, 1> Interesting;
   switch (IID) {
   case Intrinsic::riscv_vle_mask:
   case Intrinsic::riscv_vse_mask:
@@ -2706,9 +2704,9 @@ RISCVTTIImpl::getTgtMemIntrinsic(IntrinsicInst *Inst) const {
     if (HasMask)
       Mask = Inst->getArgOperand(VLIndex - 1);
     Value *EVL = Inst->getArgOperand(VLIndex);
-    Interesting.emplace_back(Inst, PtrOperandNo, IsWrite, Ty, Alignment, Mask,
-                             EVL);
-    break;
+    Info.Interesting.emplace_back(Inst, PtrOperandNo, IsWrite, Ty, Alignment,
+                                  Mask, EVL);
+    return true;
   }
   case Intrinsic::riscv_vlse_mask:
   case Intrinsic::riscv_vsse_mask:
@@ -2744,12 +2742,12 @@ RISCVTTIImpl::getTgtMemIntrinsic(IntrinsicInst *Inst) const {
     if (HasMask)
       Mask = Inst->getArgOperand(VLIndex - 1);
     Value *EVL = Inst->getArgOperand(VLIndex);
-    Interesting.emplace_back(Inst, PtrOperandNo, IsWrite, Ty, Alignment, Mask,
-                             EVL, Stride);
-    break;
+    Info.Interesting.emplace_back(Inst, PtrOperandNo, IsWrite, Ty, Alignment,
+                                  Mask, EVL, Stride);
+    return true;
   }
   }
-  return std::make_pair(Info, Interesting);
+  return false;
 }
 
 unsigned RISCVTTIImpl::getRegUsageForType(Type *Ty) const {
diff --git a/llvm/lib/Target/RISCV/RISCVTargetTransformInfo.h b/llvm/lib/Target/RISCV/RISCVTargetTransformInfo.h
index 6a784b14349a3..0567f662c07b7 100644
--- a/llvm/lib/Target/RISCV/RISCVTargetTransformInfo.h
+++ b/llvm/lib/Target/RISCV/RISCVTargetTransformInfo.h
@@ -158,8 +158,8 @@ class RISCVTTIImpl final : public BasicTTIImplBase<RISCVTTIImpl> {
   void getPeelingPreferences(Loop *L, ScalarEvolution &SE,
                              TTI::PeelingPreferences &PP) const override;
 
-  std::pair<MemIntrinsicInfo, SmallVector<InterestingMemoryOperand, 1>>
-  getTgtMemIntrinsic(IntrinsicInst *Inst) const override;
+  bool getTgtMemIntrinsic(IntrinsicInst *Inst,
+                          MemIntrinsicInfo &Info) const override;
 
   unsigned getMinVectorRegisterBitWidth() const override {
     return ST->useRVVForFixedLengthVectors() ? 16 : 0;
diff --git a/llvm/lib/Transforms/Instrumentation/AddressSanitizer.cpp b/llvm/lib/Transforms/Instrumentation/AddressSanitizer.cpp
index a91c76e3ed7ba..ac0aac3700887 100644
--- a/llvm/lib/Transforms/Instrumentation/AddressSanitizer.cpp
+++ b/llvm/lib/Transforms/Instrumentation/AddressSanitizer.cpp
@@ -1573,10 +1573,9 @@ void AddressSanitizer::getInterestingMemoryOperands(
     }
     default:
       if (auto *II = dyn_cast<IntrinsicInst>(I)) {
-        std::pair<MemIntrinsicInfo, SmallVector<InterestingMemoryOperand, 1>>
-            MemInfo = TTI->getTgtMemIntrinsic(II);
-        if (!MemInfo.second.empty())
-          Interesting = MemInfo.second;
+        MemIntrinsicInfo IntrInfo;
+        if (TTI->getTgtMemIntrinsic(II, IntrInfo))
+          Interesting = IntrInfo.Interesting;
         return;
       }
       for (unsigned ArgNo = 0; ArgNo < CI->arg_size(); ArgNo++) {
@@ -2995,7 +2994,7 @@ bool AddressSanitizer::suppressInstrumentationSiteForDebug(int &Instrumented) {
 
 bool AddressSanitizer::instrumentFunction(Function &F,
                                           const TargetLibraryInfo *TLI,
-					  const TargetTransformInfo *TTI) {
+                                          const TargetTransformInfo *TTI) {
   bool FunctionModified = false;
 
   // Do not apply any instrumentation for naked functions.
diff --git a/llvm/lib/Transforms/Scalar/EarlyCSE.cpp b/llvm/lib/Transforms/Scalar/EarlyCSE.cpp
index 975c97999bdf4..3d52099f3bf79 100644
--- a/llvm/lib/Transforms/Scalar/EarlyCSE.cpp
+++ b/llvm/lib/Transforms/Scalar/EarlyCSE.cpp
@@ -808,10 +808,7 @@ class EarlyCSE {
       : Inst(Inst) {
       if (IntrinsicInst *II = dyn_cast<IntrinsicInst>(Inst)) {
         IntrID = II->getIntrinsicID();
-        std::pair<MemIntrinsicInfo, SmallVector<InterestingMemoryOperand, 1>>
-            MemInfo = TTI.getTgtMemIntrinsic(II);
-        if (MemInfo.first.PtrVal != nullptr) {
-          Info = MemInfo.first;
+        if (TTI.getTgtMemIntrinsic(II, Info)) {
           return;
         }
         if (isHandledNonTargetIntrinsic(IntrID)) {
diff --git a/llvm/lib/Transforms/Scalar/LoopStrengthReduce.cpp b/llvm/lib/Transforms/Scalar/LoopStrengthReduce.cpp
index e3ad2965917f6..1fce64e97e604 100644
--- a/llvm/lib/Transforms/Scalar/LoopStrengthReduce.cpp
+++ b/llvm/lib/Transforms/Scalar/LoopStrengthReduce.cpp
@@ -1018,10 +1018,9 @@ static bool isAddressUse(const TargetTransformInfo &TTI,
         isAddress = true;
       break;
     default: {
-      std::pair<MemIntrinsicInfo, SmallVector<InterestingMemoryOperand, 1>>
-          MemInfo = TTI.getTgtMemIntrinsic(II);
-      if (MemInfo.first.PtrVal != nullptr) {
-        if (MemInfo.first.PtrVal == OperandVal)
+      MemIntrinsicInfo IntrInfo;
+      if (TTI.getTgtMemIntrinsic(II, IntrInfo)) {
+        if (IntrInfo.PtrVal == OperandVal)
           isAddress = true;
       }
     }
@@ -1075,13 +1074,11 @@ static MemAccessTy getAccessType(const TargetTransformInfo &TTI,
           II->getArgOperand(1)->getType()->getPointerAddressSpace();
       break;
     default: {
-      std::pair<MemIntrinsicInfo, SmallVector<InterestingMemoryOperand, 1>>
-          MemInfo = TTI.getTgtMemIntrinsic(II);
-      if (MemInfo.first.PtrVal) {
+      MemIntrinsicInfo IntrInfo;
+      if (TTI.getTgtMemIntrinsic(II, IntrInfo) && IntrInfo.PtrVal) {
         AccessTy.AddrSpace =
-            MemInfo.first.PtrVal->getType()->getPointerAddressSpace();
+            IntrInfo.PtrVal->getType()->getPointerAddressSpace();
       }
-
       break;
     }
     }

From 9f26e22e0ecb245d8d125f9ef4bea3b6f1c520af Mon Sep 17 00:00:00 2001
From: Hank Chang <hank.chang@sifive.com>
Date: Wed, 10 Sep 2025 13:48:07 +0800
Subject: [PATCH 7/9] Fix format with Craig's comment

---
 llvm/lib/Target/AMDGPU/AMDGPUTargetTransformInfo.cpp | 4 ++--
 llvm/lib/Target/PowerPC/PPCTargetTransformInfo.cpp   | 1 +
 llvm/lib/Target/PowerPC/PPCTargetTransformInfo.h     | 1 -
 llvm/lib/Transforms/Scalar/EarlyCSE.cpp              | 3 +--
 llvm/lib/Transforms/Scalar/LoopStrengthReduce.cpp    | 4 ++--
 5 files changed, 6 insertions(+), 7 deletions(-)

diff --git a/llvm/lib/Target/AMDGPU/AMDGPUTargetTransformInfo.cpp b/llvm/lib/Target/AMDGPU/AMDGPUTargetTransformInfo.cpp
index dadf10c4296b3..8b2bed4bb56e2 100644
--- a/llvm/lib/Target/AMDGPU/AMDGPUTargetTransformInfo.cpp
+++ b/llvm/lib/Target/AMDGPU/AMDGPUTargetTransformInfo.cpp
@@ -501,14 +501,14 @@ unsigned GCNTTIImpl::getMaxInterleaveFactor(ElementCount VF) const {
 }
 
 bool GCNTTIImpl::getTgtMemIntrinsic(IntrinsicInst *Inst,
-                                    MemIntrinsicInfo &Info) const {
+                                       MemIntrinsicInfo &Info) const {
   switch (Inst->getIntrinsicID()) {
   case Intrinsic::amdgcn_ds_ordered_add:
   case Intrinsic::amdgcn_ds_ordered_swap: {
     auto *Ordering = dyn_cast<ConstantInt>(Inst->getArgOperand(2));
     auto *Volatile = dyn_cast<ConstantInt>(Inst->getArgOperand(4));
     if (!Ordering || !Volatile)
-      return false; // Invalid
+      return false; // Invalid.
 
     unsigned OrderingVal = Ordering->getZExtValue();
     if (OrderingVal >
diff --git a/llvm/lib/Target/PowerPC/PPCTargetTransformInfo.cpp b/llvm/lib/Target/PowerPC/PPCTargetTransformInfo.cpp
index 2566c84b39302..2fba090f2d501 100644
--- a/llvm/lib/Target/PowerPC/PPCTargetTransformInfo.cpp
+++ b/llvm/lib/Target/PowerPC/PPCTargetTransformInfo.cpp
@@ -1024,6 +1024,7 @@ bool PPCTTIImpl::getTgtMemIntrinsic(IntrinsicInst *Inst,
   default:
     break;
   }
+
   return false;
 }
 
diff --git a/llvm/lib/Target/PowerPC/PPCTargetTransformInfo.h b/llvm/lib/Target/PowerPC/PPCTargetTransformInfo.h
index 4d935d11dab62..062578495ad9d 100644
--- a/llvm/lib/Target/PowerPC/PPCTargetTransformInfo.h
+++ b/llvm/lib/Target/PowerPC/PPCTargetTransformInfo.h
@@ -69,7 +69,6 @@ class PPCTTIImpl final : public BasicTTIImplBase<PPCTTIImpl> {
   bool canSaveCmp(Loop *L, BranchInst **BI, ScalarEvolution *SE, LoopInfo *LI,
                   DominatorTree *DT, AssumptionCache *AC,
                   TargetLibraryInfo *LibInfo) const override;
-
   bool getTgtMemIntrinsic(IntrinsicInst *Inst,
                           MemIntrinsicInfo &Info) const override;
 
diff --git a/llvm/lib/Transforms/Scalar/EarlyCSE.cpp b/llvm/lib/Transforms/Scalar/EarlyCSE.cpp
index 3d52099f3bf79..0f8cc6ca6ed21 100644
--- a/llvm/lib/Transforms/Scalar/EarlyCSE.cpp
+++ b/llvm/lib/Transforms/Scalar/EarlyCSE.cpp
@@ -808,9 +808,8 @@ class EarlyCSE {
       : Inst(Inst) {
       if (IntrinsicInst *II = dyn_cast<IntrinsicInst>(Inst)) {
         IntrID = II->getIntrinsicID();
-        if (TTI.getTgtMemIntrinsic(II, Info)) {
+        if (TTI.getTgtMemIntrinsic(II, Info))
           return;
-        }
         if (isHandledNonTargetIntrinsic(IntrID)) {
           switch (IntrID) {
           case Intrinsic::masked_load:
diff --git a/llvm/lib/Transforms/Scalar/LoopStrengthReduce.cpp b/llvm/lib/Transforms/Scalar/LoopStrengthReduce.cpp
index 1fce64e97e604..76624876cc6d3 100644
--- a/llvm/lib/Transforms/Scalar/LoopStrengthReduce.cpp
+++ b/llvm/lib/Transforms/Scalar/LoopStrengthReduce.cpp
@@ -1076,8 +1076,8 @@ static MemAccessTy getAccessType(const TargetTransformInfo &TTI,
     default: {
       MemIntrinsicInfo IntrInfo;
       if (TTI.getTgtMemIntrinsic(II, IntrInfo) && IntrInfo.PtrVal) {
-        AccessTy.AddrSpace =
-            IntrInfo.PtrVal->getType()->getPointerAddressSpace();
+        AccessTy.AddrSpace
+          = IntrInfo.PtrVal->getType()->getPointerAddressSpace();
       }
       break;
     }

From 665ebae5e0b315577574b5b7a230eedc799699f3 Mon Sep 17 00:00:00 2001
From: Hank Chang <hank.chang@sifive.com>
Date: Wed, 10 Sep 2025 14:02:00 +0800
Subject: [PATCH 8/9] Fix format

---
 llvm/lib/Transforms/Scalar/LoopStrengthReduce.cpp | 1 +
 1 file changed, 1 insertion(+)

diff --git a/llvm/lib/Transforms/Scalar/LoopStrengthReduce.cpp b/llvm/lib/Transforms/Scalar/LoopStrengthReduce.cpp
index 76624876cc6d3..e3ef9d8680b53 100644
--- a/llvm/lib/Transforms/Scalar/LoopStrengthReduce.cpp
+++ b/llvm/lib/Transforms/Scalar/LoopStrengthReduce.cpp
@@ -1079,6 +1079,7 @@ static MemAccessTy getAccessType(const TargetTransformInfo &TTI,
         AccessTy.AddrSpace
           = IntrInfo.PtrVal->getType()->getPointerAddressSpace();
       }
+
       break;
     }
     }

From 766d8c590bb4c81b5a9352152e0634a71deec576 Mon Sep 17 00:00:00 2001
From: Hank Chang <hank.chang@sifive.com>
Date: Wed, 10 Sep 2025 14:08:52 +0800
Subject: [PATCH 9/9] Fix more format

---
 llvm/lib/Target/AMDGPU/AMDGPUTargetTransformInfo.cpp | 3 +--
 llvm/lib/Target/PowerPC/PPCTargetTransformInfo.h     | 1 -
 2 files changed, 1 insertion(+), 3 deletions(-)

diff --git a/llvm/lib/Target/AMDGPU/AMDGPUTargetTransformInfo.cpp b/llvm/lib/Target/AMDGPU/AMDGPUTargetTransformInfo.cpp
index 8b2bed4bb56e2..846a0b6280f19 100644
--- a/llvm/lib/Target/AMDGPU/AMDGPUTargetTransformInfo.cpp
+++ b/llvm/lib/Target/AMDGPU/AMDGPUTargetTransformInfo.cpp
@@ -511,8 +511,7 @@ bool GCNTTIImpl::getTgtMemIntrinsic(IntrinsicInst *Inst,
       return false; // Invalid.
 
     unsigned OrderingVal = Ordering->getZExtValue();
-    if (OrderingVal >
-        static_cast<unsigned>(AtomicOrdering::SequentiallyConsistent))
+    if (OrderingVal > static_cast<unsigned>(AtomicOrdering::SequentiallyConsistent))
       return false;
 
     Info.PtrVal = Inst->getArgOperand(0);
diff --git a/llvm/lib/Target/PowerPC/PPCTargetTransformInfo.h b/llvm/lib/Target/PowerPC/PPCTargetTransformInfo.h
index 062578495ad9d..475472ac3720f 100644
--- a/llvm/lib/Target/PowerPC/PPCTargetTransformInfo.h
+++ b/llvm/lib/Target/PowerPC/PPCTargetTransformInfo.h
@@ -71,7 +71,6 @@ class PPCTTIImpl final : public BasicTTIImplBase<PPCTTIImpl> {
                   TargetLibraryInfo *LibInfo) const override;
   bool getTgtMemIntrinsic(IntrinsicInst *Inst,
                           MemIntrinsicInfo &Info) const override;
-
   void getUnrollingPreferences(Loop *L, ScalarEvolution &SE,
                                TTI::UnrollingPreferences &UP,
                                OptimizationRemarkEmitter *ORE) const override;