[llvm] 96d3c82 - Revert "[SROA] `isVectorPromotionViable()`: memory intrinsics operate on vectors of bytes (take 3)"

[Roman Lebedev via llvm-commits]

Author: Roman Lebedev
Date: 2022-12-16T22:58:41+03:00
New Revision: 96d3c82645cf41a38543c5128cc15cda5761a76a

URL: https://github.com/llvm/llvm-project/commit/96d3c82645cf41a38543c5128cc15cda5761a76a
DIFF: https://github.com/llvm/llvm-project/commit/96d3c82645cf41a38543c5128cc15cda5761a76a.diff

LOG: Revert "[SROA] `isVectorPromotionViable()`: memory intrinsics operate on vectors of bytes (take 3)"

While the PPC litte-endian miscompile did get addressed
by https://reviews.llvm.org/D140046
the PPV big-endian bots are still unhappy.
https://lab.llvm.org/buildbot/#/builders/93/builds/12560

This reverts commit 7bd358bcb4e358b4351c69e02ef76939e08acdc7.

Added: 
    

Modified: 
    clang/test/CodeGenOpenCL/amdgpu-nullptr.cl
    llvm/lib/Transforms/Scalar/SROA.cpp
    llvm/test/CodeGen/AMDGPU/v1024.ll
    llvm/test/DebugInfo/Generic/assignment-tracking/sroa/memcpy.ll
    llvm/test/DebugInfo/Generic/assignment-tracking/sroa/memmove-to-from-same-alloca.ll
    llvm/test/DebugInfo/Generic/assignment-tracking/sroa/store.ll
    llvm/test/DebugInfo/Generic/assignment-tracking/sroa/user-memcpy.ll
    llvm/test/DebugInfo/Generic/assignment-tracking/sroa/vec-2.ll
    llvm/test/DebugInfo/X86/sroasplit-1.ll
    llvm/test/DebugInfo/X86/sroasplit-4.ll
    llvm/test/Transforms/PhaseOrdering/instcombine-sroa-inttoptr.ll
    llvm/test/Transforms/SROA/address-spaces.ll
    llvm/test/Transforms/SROA/alignment.ll
    llvm/test/Transforms/SROA/alloca-address-space.ll
    llvm/test/Transforms/SROA/basictest.ll
    llvm/test/Transforms/SROA/pointer-offset-size.ll
    llvm/test/Transforms/SROA/scalable-vectors.ll
    llvm/test/Transforms/SROA/slice-width.ll
    llvm/test/Transforms/SROA/tbaa-struct.ll
    llvm/test/Transforms/SROA/tbaa-struct2.ll
    llvm/test/Transforms/SROA/vector-promotion.ll

Removed: 
    


################################################################################
diff  --git a/clang/test/CodeGenOpenCL/amdgpu-nullptr.cl b/clang/test/CodeGenOpenCL/amdgpu-nullptr.cl
index 859e81f08d6bd..65f6f2e7d8c24 100644
--- a/clang/test/CodeGenOpenCL/amdgpu-nullptr.cl
+++ b/clang/test/CodeGenOpenCL/amdgpu-nullptr.cl
@@ -515,17 +515,13 @@ typedef struct {
   private char *p;
 } StructTy3;
 
-// CHECK-LABEL: @test_memset_private(
-// CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[TMP0:%.*]] = bitcast [[STRUCT_STRUCTTY3:%.*]] addrspace(5)* [[PTR:%.*]] to i8 addrspace(5)*
-// CHECK-NEXT:    [[S3_SROA_0_SROA_0_0_S3_SROA_0_0__SROA_CAST2_SROA_CAST:%.*]] = bitcast [[STRUCT_STRUCTTY3]] addrspace(5)* [[PTR]] to <32 x i8> addrspace(5)*
-// CHECK-NEXT:    store <32 x i8> zeroinitializer, <32 x i8> addrspace(5)* [[S3_SROA_0_SROA_0_0_S3_SROA_0_0__SROA_CAST2_SROA_CAST]], align 8, !tbaa.struct !9
-// CHECK-NEXT:    [[S3_SROA_4_0__SROA_IDX6:%.*]] = getelementptr inbounds [[STRUCT_STRUCTTY3]], [[STRUCT_STRUCTTY3]] addrspace(5)* [[PTR]], i32 0, i32 4
-// CHECK-NEXT:    store i8 addrspace(5)* addrspacecast (i8* null to i8 addrspace(5)*), i8 addrspace(5)* addrspace(5)* [[S3_SROA_4_0__SROA_IDX6]], align 8, !tbaa.struct !12
-// CHECK-NEXT:    [[S3_SROA_5_0__SROA_IDX:%.*]] = getelementptr inbounds i8, i8 addrspace(5)* [[TMP0]], i32 36
-// CHECK-NEXT:    [[S3_SROA_5_0__SROA_CAST8:%.*]] = bitcast i8 addrspace(5)* [[S3_SROA_5_0__SROA_IDX]] to i32 addrspace(5)*
-// CHECK-NEXT:    store i32 0, i32 addrspace(5)* [[S3_SROA_5_0__SROA_CAST8]], align 4, !tbaa.struct !13
-// CHECK-NEXT:    ret void
+// CHECK-LABEL: test_memset_private
+// CHECK: call void @llvm.memset.p5i8.i64(i8 addrspace(5)* noundef align 8 {{.*}}, i8 0, i64 32, i1 false)
+// CHECK: [[GEP:%.*]] = getelementptr inbounds %struct.StructTy3, %struct.StructTy3 addrspace(5)* %ptr, i32 0, i32 4
+// CHECK: store i8 addrspace(5)* addrspacecast (i8* null to i8 addrspace(5)*), i8 addrspace(5)* addrspace(5)* [[GEP]]
+// CHECK: [[GEP1:%.*]] = getelementptr inbounds i8, i8 addrspace(5)* {{.*}}, i32 36
+// CHECK: [[GEP1_CAST:%.*]] = bitcast i8 addrspace(5)* [[GEP1]] to i32 addrspace(5)*
+// CHECK: store i32 0, i32 addrspace(5)* [[GEP1_CAST]], align 4
 void test_memset_private(private StructTy3 *ptr) {
   StructTy3 S3 = {0, 0, 0, 0, 0};
   *ptr = S3;

diff  --git a/llvm/lib/Transforms/Scalar/SROA.cpp b/llvm/lib/Transforms/Scalar/SROA.cpp
index a9a0f444aec03..748b8627be7ed 100644
--- a/llvm/lib/Transforms/Scalar/SROA.cpp
+++ b/llvm/lib/Transforms/Scalar/SROA.cpp
@@ -2029,10 +2029,8 @@ static bool isVectorPromotionViableForSlice(Partition &P, const Slice &S,
                       ? Ty->getElementType()
                       : FixedVectorType::get(Ty->getElementType(), NumElements);
 
-  Type *SplitIntTy = nullptr;
-  if (uint64_t Bitwidth = NumElements * ElementSize * 8;
-      Bitwidth <= IntegerType::MAX_INT_BITS)
-    SplitIntTy = Type::getIntNTy(Ty->getContext(), Bitwidth);
+  Type *SplitIntTy =
+      Type::getIntNTy(Ty->getContext(), NumElements * ElementSize * 8);
 
   Use *U = S.getUse();
 
@@ -2051,8 +2049,7 @@ static bool isVectorPromotionViableForSlice(Partition &P, const Slice &S,
     // Disable vector promotion when there are loads or stores of an FCA.
     if (LTy->isStructTy())
       return false;
-    if (SplitIntTy &&
-        (P.beginOffset() > S.beginOffset() || P.endOffset() < S.endOffset())) {
+    if (P.beginOffset() > S.beginOffset() || P.endOffset() < S.endOffset()) {
       assert(LTy->isIntegerTy());
       LTy = SplitIntTy;
     }
@@ -2065,8 +2062,7 @@ static bool isVectorPromotionViableForSlice(Partition &P, const Slice &S,
     // Disable vector promotion when there are loads or stores of an FCA.
     if (STy->isStructTy())
       return false;
-    if (SplitIntTy &&
-        (P.beginOffset() > S.beginOffset() || P.endOffset() < S.endOffset())) {
+    if (P.beginOffset() > S.beginOffset() || P.endOffset() < S.endOffset()) {
       assert(STy->isIntegerTy());
       STy = SplitIntTy;
     }
@@ -2116,8 +2112,7 @@ static bool checkVectorTypeForPromotion(Partition &P, VectorType *VTy,
 /// SSA value. We only can ensure this for a limited set of operations, and we
 /// don't want to do the rewrites unless we are confident that the result will
 /// be promotable, so we have an early test here.
-static VectorType *isVectorPromotionViable(Partition &P, LLVMContext &Ctx,
-                                           const DataLayout &DL) {
+static VectorType *isVectorPromotionViable(Partition &P, const DataLayout &DL) {
   // Collect the candidate types for vector-based promotion. Also track whether
   // we have 
diff erent element types.
   SmallVector<VectorType *, 4> CandidateTys;
@@ -2154,7 +2149,6 @@ static VectorType *isVectorPromotionViable(Partition &P, LLVMContext &Ctx,
       }
     }
   };
-  bool SeenMemTransferInst = false;
   // Consider any loads or stores that are the exact size of the slice.
   for (const Slice &S : P)
     if (S.beginOffset() == P.beginOffset() &&
@@ -2163,29 +2157,8 @@ static VectorType *isVectorPromotionViable(Partition &P, LLVMContext &Ctx,
         CheckCandidateType(LI->getType());
       else if (auto *SI = dyn_cast<StoreInst>(S.getUse()->getUser()))
         CheckCandidateType(SI->getValueOperand()->getType());
-      else if (isa<MemTransferInst>(S.getUse()->getUser()))
-        SeenMemTransferInst = true;
     }
 
-  // If we have seen mem transfer intrinsic,
-  // and the partition is small-enough,
-  // enqueue appropriate byte vector.
-  //
-  // The "small-enough" threshold is somewhat arbitrary,
-  // and is mostly dictated by compile-time concerns,
-  // but, at the same time, SDAG SDNode can't handle
-  // more then 65535 operands, so we should not
-  // produce vectors with more than ~32768 elements.
-  //
-  // Perhaps, we should also take into account the TTI:
-  //     `getNumberOfRegisters() * getRegisterBitWidth() / 8` ?
-  //
-  // FIXME: byte type is sticky. If we had any op with byte-typed elements,
-  //        then we should choose that type.
-  if (SeenMemTransferInst && P.size() <= 64)
-    CheckCandidateType(
-        FixedVectorType::get(IntegerType::getInt8Ty(Ctx), P.size()));
-
   // If we didn't find a vector type, nothing to do here.
   if (CandidateTys.empty())
     return nullptr;
@@ -2242,6 +2215,13 @@ static VectorType *isVectorPromotionViable(Partition &P, LLVMContext &Ctx,
     CandidateTys.resize(1);
   }
 
+  // FIXME: hack. Do we have a named constant for this?
+  // SDAG SDNode can't have more than 65535 operands.
+  llvm::erase_if(CandidateTys, [](VectorType *VTy) {
+    return cast<FixedVectorType>(VTy)->getNumElements() >
+           std::numeric_limits<unsigned short>::max();
+  });
+
   for (VectorType *VTy : CandidateTys)
     if (checkVectorTypeForPromotion(P, VTy, DL))
       return VTy;
@@ -4654,9 +4634,8 @@ AllocaInst *SROAPass::rewritePartition(AllocaInst &AI, AllocaSlices &AS,
 
   bool IsIntegerPromotable = isIntegerWideningViable(P, SliceTy, DL);
 
-  VectorType *VecTy = IsIntegerPromotable
-                          ? nullptr
-                          : isVectorPromotionViable(P, AI.getContext(), DL);
+  VectorType *VecTy =
+      IsIntegerPromotable ? nullptr : isVectorPromotionViable(P, DL);
   if (VecTy)
     SliceTy = VecTy;
 

diff  --git a/llvm/test/CodeGen/AMDGPU/v1024.ll b/llvm/test/CodeGen/AMDGPU/v1024.ll
index 6dbb9443fd5cf..47e2904f19afd 100644
--- a/llvm/test/CodeGen/AMDGPU/v1024.ll
+++ b/llvm/test/CodeGen/AMDGPU/v1024.ll
@@ -4,28 +4,27 @@
 
 ; GCN-LABEL: {{^}}test_v1024:
 ; GCN-NOT: v_accvgpr
-; GCN-COUNT-10: v_mov_b32_e32
+; GCN-COUNT-32: v_mov_b32_e32
 ; GCN-NOT: v_accvgpr
 define amdgpu_kernel void @test_v1024() {
 entry:
   %alloca = alloca <32 x i32>, align 16, addrspace(5)
-  %cast = bitcast <32 x i32> addrspace(5)* %alloca to i8 addrspace(5)*
-  call void @llvm.memset.p5i8.i32(i8 addrspace(5)* %cast, i8 0, i32 128, i1 false)
+  call void @llvm.memset.p5.i32(ptr addrspace(5) %alloca, i8 0, i32 128, i1 false)
   br i1 undef, label %if.then.i.i, label %if.else.i
 
 if.then.i.i:                                      ; preds = %entry
-  call void @llvm.memcpy.p5i8.p5i8.i64(i8 addrspace(5)* align 16 %cast, i8 addrspace(5)* align 4 undef, i64 128, i1 false)
+  call void @llvm.memcpy.p5.p5.i64(ptr addrspace(5) align 16 %alloca, ptr addrspace(5) align 4 undef, i64 128, i1 false)
   br label %if.then.i62.i
 
 if.else.i:                                        ; preds = %entry
   br label %if.then.i62.i
 
 if.then.i62.i:                                    ; preds = %if.else.i, %if.then.i.i
-  call void @llvm.memcpy.p1i8.p5i8.i64(i8 addrspace(1)* align 4 undef, i8 addrspace(5)* align 16 %cast, i64 128, i1 false)
+  call void @llvm.memcpy.p1.p5.i64(ptr addrspace(1) align 4 undef, ptr addrspace(5) align 16 %alloca, i64 128, i1 false)
   ret void
 }
 
-declare void @llvm.memset.p5i8.i32(i8 addrspace(5)* nocapture readonly, i8, i32, i1 immarg)
-declare void @llvm.memcpy.p5i8.p5i8.i64(i8 addrspace(5)* nocapture writeonly, i8 addrspace(5)* nocapture readonly, i64, i1 immarg)
+declare void @llvm.memset.p5.i32(ptr addrspace(5) nocapture readonly, i8, i32, i1 immarg)
+declare void @llvm.memcpy.p5.p5.i64(ptr addrspace(5) nocapture writeonly, ptr addrspace(5) nocapture readonly, i64, i1 immarg)
 
-declare void @llvm.memcpy.p1i8.p5i8.i64(i8 addrspace(1)* nocapture writeonly, i8 addrspace(5)* nocapture readonly, i64, i1 immarg)
+declare void @llvm.memcpy.p1.p5.i64(ptr addrspace(1) nocapture writeonly, ptr addrspace(5) nocapture readonly, i64, i1 immarg)

diff  --git a/llvm/test/DebugInfo/Generic/assignment-tracking/sroa/memcpy.ll b/llvm/test/DebugInfo/Generic/assignment-tracking/sroa/memcpy.ll
index 8d17505cba203..f310456cc60a8 100644
--- a/llvm/test/DebugInfo/Generic/assignment-tracking/sroa/memcpy.ll
+++ b/llvm/test/DebugInfo/Generic/assignment-tracking/sroa/memcpy.ll
@@ -25,12 +25,22 @@
 
 ;; Split alloca.
 ; CHECK: entry:
-; CHECK-NEXT: %To.sroa.0.sroa.0.0.copyload = load <12 x i8>, ptr @From, align 4, !dbg
-; CHECK-NEXT: %To.sroa.3.0.copyload = load i32, ptr getelementptr inbounds (i8, ptr @From, i64 12), align 4, !dbg
-; CHECK-NEXT: %To.sroa.4.sroa.0.0.copyload = load <12 x i8>, ptr getelementptr inbounds (i8, ptr @From, i64 16), align 4, !dbg
-; CHECK-NEXT: call void @llvm.dbg.assign(metadata <12 x i8> %To.sroa.0.sroa.0.0.copyload, metadata ![[TO:[0-9]+]], metadata !DIExpression(DW_OP_LLVM_fragment, 0, 96), metadata ![[ID_1:[0-9]+]], metadata ptr undef, metadata !DIExpression()), !dbg
-; CHECK-NEXT: call void @llvm.dbg.assign(metadata i32 %To.sroa.3.0.copyload, metadata ![[TO]], metadata !DIExpression(DW_OP_LLVM_fragment, 96, 32), metadata ![[ID_3:[0-9]+]], metadata ptr undef, metadata !DIExpression()), !dbg
-; CHECK-NEXT: call void @llvm.dbg.assign(metadata <12 x i8> %To.sroa.4.sroa.0.0.copyload, metadata ![[TO]], metadata !DIExpression(DW_OP_LLVM_fragment, 128, 96), metadata ![[ID_6:[0-9]+]], metadata ptr undef, metadata !DIExpression()), !dbg
+; CHECK-NEXT: %To.sroa.0 = alloca { i32, i32, i32 }, align 8, !DIAssignID ![[ID_1:[0-9]+]]
+; CHECK-NEXT: call void @llvm.dbg.assign(metadata {{.+}} undef, metadata ![[TO:[0-9]+]], metadata !DIExpression(DW_OP_LLVM_fragment, 0, 96), metadata ![[ID_1]], metadata ptr %To.sroa.0, metadata !DIExpression()), !dbg
+
+; CHECK-NEXT: %To.sroa.4 = alloca { i32, i32, i32 }, align 8, !DIAssignID ![[ID_3:[0-9]+]]
+; CHECK-NEXT: call void @llvm.dbg.assign(metadata {{.+}} undef, metadata ![[TO]], metadata !DIExpression(DW_OP_LLVM_fragment, 128, 96), metadata ![[ID_3]], metadata ptr %To.sroa.4, metadata !DIExpression()), !dbg
+
+;; Split memcpy.
+; CHECK: call void @llvm.memcpy{{.*}}(ptr align 8 %To.sroa.0, ptr align 4 @From, i64 12, i1 false),{{.*}}!DIAssignID ![[ID_4:[0-9]+]]
+;; This slice has been split and is promoted.
+; CHECK: %To.sroa.3.0.copyload = load i32, ptr getelementptr inbounds (i8, ptr @From, i64 12)
+; CHECK: call void @llvm.memcpy{{.*}}(ptr align 8 %To.sroa.4, ptr align 4 getelementptr inbounds (i8, ptr @From, i64 16), i64 12, i1 false){{.*}}!DIAssignID ![[ID_6:[0-9]+]]
+
+;; Intrinsics for the splits above.
+; CHECK-NEXT: call void @llvm.dbg.assign(metadata {{.+}} undef, metadata ![[TO]], metadata !DIExpression(DW_OP_LLVM_fragment, 0, 96), metadata ![[ID_4]], metadata ptr %To.sroa.0, metadata !DIExpression()), !dbg
+; CHECK-NEXT: call void @llvm.dbg.assign(metadata i32 %To.sroa.3.0.copyload, metadata ![[TO]], metadata !DIExpression(DW_OP_LLVM_fragment, 96, 32), metadata !{{.+}}, metadata ptr undef, metadata !DIExpression()), !dbg
+; CHECK-NEXT: call void @llvm.dbg.assign(metadata {{.+}} undef, metadata ![[TO]], metadata !DIExpression(DW_OP_LLVM_fragment, 128, 96), metadata ![[ID_6]], metadata ptr %To.sroa.4, metadata !DIExpression()), !dbg
 
 target datalayout = "e-m:e-p270:32:32-p271:32:32-p272:64:64-i64:64-f80:128-n8:16:32:64-S128"
 

diff  --git a/llvm/test/DebugInfo/Generic/assignment-tracking/sroa/memmove-to-from-same-alloca.ll b/llvm/test/DebugInfo/Generic/assignment-tracking/sroa/memmove-to-from-same-alloca.ll
index 7e0f87fbec5ff..766e338aab5d7 100644
--- a/llvm/test/DebugInfo/Generic/assignment-tracking/sroa/memmove-to-from-same-alloca.ll
+++ b/llvm/test/DebugInfo/Generic/assignment-tracking/sroa/memmove-to-from-same-alloca.ll
@@ -6,7 +6,7 @@
 ;; __attribute__((nodebug)) int Cond;
 ;; __attribute__((nodebug)) Blob *C;
 ;; __attribute__((nodebug)) void call(int);
-;;
+;; 
 ;; void f() {
 ;;   int A[16];
 ;;   __attribute__ ((nodebug)) int B[16];
@@ -14,7 +14,7 @@
 ;;   __builtin_memmove(&A[0], &Glob, sizeof(Blob));
 ;;   call(0);
 ;;   // B[8:14) <- Glob
-;;   __builtin_memmove(&B[8], &Glob, sizeof(Blob));
+;;   __builtin_memmove(&B[8], &Glob, sizeof(Blob));  
 ;;   call(A[0]);
 ;;   // A[8:14) <- A[0:6)
 ;;   __builtin_memmove(&A[8], &A[0], sizeof(Blob));
@@ -24,9 +24,9 @@
 ;;     __builtin_memmove(C, &A[8], sizeof(Blob));
 ;;   else
 ;;     // C <- B[8:14)
-;;     __builtin_memmove(C, &B[8], sizeof(Blob));
+;;     __builtin_memmove(C, &B[8], sizeof(Blob));    
 ;; }
-;;
+;; 
 ;; using:
 ;;   clang test.cpp -emit-llvm -S -g -O2 -Xclang -disable-llvm-passes -o - \
 ;;   | opt -passes=declare-to-assign -o test.ll - -S
@@ -38,12 +38,12 @@
 ;; memcpy. Check that the dbg.assign address and fragment are correct and
 ;; ensure the DIAssignID still links it to the memmove(/memcpy).
 
-; CHECK: %B = alloca [16 x i32]
-; CHECK: %A.sroa.0.sroa.5.0.copyload = load <20 x i8>, ptr getelementptr inbounds (i8, ptr @Glob, i64 4), align 4, !dbg
+; CHECK: %A.sroa.0.sroa.5 = alloca [5 x i32]
+; CHECK: llvm.memcpy{{.*}}(ptr align 4 %A.sroa.0.sroa.5, ptr align 4 getelementptr inbounds (i8, ptr @Glob, i64 4), i64 20, i1 false){{.*}}!DIAssignID ![[ID:[0-9]+]]
 ;; Here's the dbg.assign for element 0 - it's not important for the test.
 ; CHECK-NEXT: llvm.dbg.assign({{.*}}!DIExpression(DW_OP_LLVM_fragment, 0, 32){{.*}})
 ;; This is the dbg.assign we care about:
-; CHECK-NEXT: llvm.dbg.assign(metadata <20 x i8> %A.sroa.0.sroa.5.0.copyload, metadata ![[VAR:[0-9]+]], metadata !DIExpression(DW_OP_LLVM_fragment, 32, 160), metadata ![[ID:[0-9]+]], metadata ptr undef, metadata !DIExpression())
+; CHECK-NEXT: llvm.dbg.assign(metadata i1 undef, metadata ![[VAR:[0-9]+]], metadata !DIExpression(DW_OP_LLVM_fragment, 32, 160), metadata ![[ID]], metadata ptr %A.sroa.0.sroa.5, metadata !DIExpression())
 
 ; CHECK: ![[VAR]] = !DILocalVariable(name: "A"
 

diff  --git a/llvm/test/DebugInfo/Generic/assignment-tracking/sroa/store.ll b/llvm/test/DebugInfo/Generic/assignment-tracking/sroa/store.ll
index a1c98708afbe8..c29cf488ce8b9 100644
--- a/llvm/test/DebugInfo/Generic/assignment-tracking/sroa/store.ll
+++ b/llvm/test/DebugInfo/Generic/assignment-tracking/sroa/store.ll
@@ -26,13 +26,22 @@
 ;;   | opt -passes=declare-to-assign -S -o -
 
 ; CHECK: entry:
-; CHECK-NEXT:   %agg.tmp = alloca %struct.LargeStruct
-; CHECK-NEXT:   call void @llvm.dbg.assign(metadata <12 x i8> zeroinitializer, metadata ![[VAR:[0-9]+]], metadata !DIExpression(DW_OP_LLVM_fragment, 0, 96), metadata ![[ID_1:[0-9]+]], metadata ptr undef, metadata !DIExpression()), !dbg
+; CHECK-NEXT:   %S.sroa.0 = alloca { i32, i32, i32 }, align 8, !DIAssignID ![[ID_1:[0-9]+]]
+; CHECK-NEXT:   call void @llvm.dbg.assign(metadata i1 undef, metadata ![[VAR:[0-9]+]], metadata !DIExpression(DW_OP_LLVM_fragment, 0, 96), metadata ![[ID_1]], metadata ptr %S.sroa.0, metadata !DIExpression()), !dbg
 
+; CHECK-NEXT:   %S.sroa.6 = alloca { i32, i32, i32 }, align 8, !DIAssignID ![[ID_3:[0-9]+]]
+; CHECK-NEXT:   call void @llvm.dbg.assign(metadata i1 undef, metadata ![[VAR]], metadata !DIExpression(DW_OP_LLVM_fragment, 128, 96), metadata ![[ID_3]], metadata ptr %S.sroa.6, metadata !DIExpression()), !dbg
+
+;; The memset has been split into [0, 96)[96, 128)[128, 224) bit slices. The
+;; memset for the middle slice has been removed.
+; CHECK: call void @llvm.memset{{.*}}(ptr align 8 %S.sroa.0, i8 0, i64 12, i1 false), !dbg !{{.+}}, !DIAssignID ![[ID_4:[0-9]+]]
+; CHECK-NEXT: call void @llvm.memset{{.*}}(ptr align 8 %S.sroa.6, i8 0, i64 12, i1 false), !dbg !{{.+}}, !DIAssignID ![[ID_5:[0-9]+]]
+
+; CHECK-NEXT: call void @llvm.dbg.assign(metadata i8 0, metadata ![[VAR]], metadata !DIExpression(DW_OP_LLVM_fragment, 0, 96), metadata ![[ID_4]], metadata ptr %S.sroa.0, metadata !DIExpression()), !dbg
 ;; This is the one we care about most in this test: check that a memset->store
 ;; gets a correct dbg.assign.
 ; CHECK-NEXT: call void @llvm.dbg.assign(metadata i32 0, metadata ![[VAR]], metadata !DIExpression(DW_OP_LLVM_fragment, 96, 32), metadata !{{.+}}, metadata ptr undef, metadata !DIExpression()), !dbg
-; CHECK-NEXT: call void @llvm.dbg.assign(metadata <12 x i8> zeroinitializer, metadata ![[VAR]], metadata !DIExpression(DW_OP_LLVM_fragment, 128, 96), metadata ![[ID_5:[0-9]+]], metadata ptr undef, metadata !DIExpression()), !dbg
+; CHECK-NEXT: call void @llvm.dbg.assign(metadata i8 0, metadata ![[VAR]], metadata !DIExpression(DW_OP_LLVM_fragment, 128, 96), metadata ![[ID_5]], metadata ptr %S.sroa.6, metadata !DIExpression()), !dbg
 
 ;; The load from global+store becomes a load.
 ;; FIXME: In reality it is actually stored again later on.

diff  --git a/llvm/test/DebugInfo/Generic/assignment-tracking/sroa/user-memcpy.ll b/llvm/test/DebugInfo/Generic/assignment-tracking/sroa/user-memcpy.ll
index 8d246ef932dd7..d8b6d76456d80 100644
--- a/llvm/test/DebugInfo/Generic/assignment-tracking/sroa/user-memcpy.ll
+++ b/llvm/test/DebugInfo/Generic/assignment-tracking/sroa/user-memcpy.ll
@@ -24,8 +24,9 @@
 
 ;; | V3i point = {0, 0, 0};
 ; CHECK-NEXT: call void @llvm.dbg.assign(metadata i64 0, metadata ![[point:[0-9]+]], metadata !DIExpression(DW_OP_LLVM_fragment, 0, 64), metadata !{{.+}}, metadata ptr undef, metadata !DIExpression()), !dbg
-; CHECK-NEXT: call void @llvm.dbg.assign(metadata <16 x i8> zeroinitializer, metadata ![[point]], metadata !DIExpression(DW_OP_LLVM_fragment, 64, 128), metadata !{{.+}}, metadata ptr undef, metadata !DIExpression()), !dbg
-; CHECK: call void @llvm.dbg.value(metadata <16 x i8> %point.sroa.3.16.vecblend, metadata ![[point:[0-9]+]], metadata !DIExpression(DW_OP_LLVM_fragment, 64, 128)), !dbg
+; CHECK-NEXT: call void @llvm.dbg.assign(metadata i64 0, metadata ![[point]], metadata !DIExpression(DW_OP_LLVM_fragment, 64, 64), metadata !{{.+}}, metadata ptr undef, metadata !DIExpression()), !dbg
+; CHECK-NEXT: call void @llvm.dbg.assign(metadata i64 0, metadata ![[point]], metadata !DIExpression(DW_OP_LLVM_fragment, 128, 64), metadata !{{.+}}, metadata ptr undef, metadata !DIExpression()), !dbg
+
 ;; point.z = 5000;
 ; CHECK-NEXT: call void @llvm.dbg.assign(metadata i64 5000, metadata ![[point]], metadata !DIExpression(DW_OP_LLVM_fragment, 128, 64), metadata !{{.+}}, metadata ptr undef, metadata !DIExpression()), !dbg
 
@@ -34,15 +35,20 @@
 ;;     local.other.x = global.other.x
 ;;     local.other.y = global.other.y
 ;;     local.other.z = global.other.z
-; CHECK-NEXT: %other.sroa.0.0.copyload = load <24 x i8>, ptr @__const._Z3funv.other, align 8, !dbg
-; CHECK-NEXT: call void @llvm.dbg.assign(metadata <24 x i8> %other.sroa.0.0.copyload, metadata ![[other:[0-9]+]], metadata !DIExpression(), metadata !{{.+}}, metadata ptr undef, metadata !DIExpression()), !dbg
+; CHECK-NEXT: %other.sroa.0.0.copyload = load i64, ptr @__const._Z3funv.other
+; CHECK-NEXT: %other.sroa.4.0.copyload = load i64, ptr getelementptr inbounds (i8, ptr @__const._Z3funv.other, i64 8)
+; CHECK-NEXT: %other.sroa.5.0.copyload = load i64, ptr getelementptr inbounds (i8, ptr @__const._Z3funv.other, i64 16)
+; CHECK-NEXT: call void @llvm.dbg.assign(metadata i64 %other.sroa.0.0.copyload, metadata ![[other:[0-9]+]], metadata !DIExpression(DW_OP_LLVM_fragment, 0, 64), metadata !{{.+}}, metadata ptr undef, metadata !DIExpression()), !dbg
+; CHECK-NEXT: call void @llvm.dbg.assign(metadata i64 %other.sroa.4.0.copyload, metadata ![[other]], metadata !DIExpression(DW_OP_LLVM_fragment, 64, 64), metadata !{{.+}}, metadata ptr undef, metadata !DIExpression()), !dbg
+; CHECK-NEXT: call void @llvm.dbg.assign(metadata i64 %other.sroa.5.0.copyload, metadata ![[other]], metadata !DIExpression(DW_OP_LLVM_fragment, 128, 64), metadata !{{.+}}, metadata ptr undef, metadata !DIExpression()), !dbg
 
 ;; | std::memcpy(&point.y, &other.x, sizeof(long) * 2);
 ;;   other is now 3 scalars:
 ;;     point.y = other.x
+; CHECK-NEXT: call void @llvm.dbg.assign(metadata i64 %other.sroa.0.0.copyload, metadata ![[point]], metadata !DIExpression(DW_OP_LLVM_fragment, 64, 64), metadata !{{.+}}, metadata ptr undef, metadata !DIExpression()), !dbg
+;;
 ;;     point.z = other.y
-; CHECK-NEXT: %other.sroa.0.0.vec.extract = shufflevector
-; CHECK-NEXT: call void @llvm.dbg.assign(metadata <16 x i8> %other.sroa.0.0.vec.extract, metadata ![[point]], metadata !DIExpression(DW_OP_LLVM_fragment, 64, 128), metadata !{{.+}}, metadata ptr undef, metadata !DIExpression()), !dbg
+; CHECK-NEXT: call void @llvm.dbg.assign(metadata i64 %other.sroa.4.0.copyload, metadata ![[point]], metadata !DIExpression(DW_OP_LLVM_fragment, 128, 64), metadata !{{.+}}, metadata ptr undef, metadata !DIExpression()), !dbg
 
 ; CHECK: ![[point]] = !DILocalVariable(name: "point",
 ; CHECK: ![[other]] = !DILocalVariable(name: "other",

diff  --git a/llvm/test/DebugInfo/Generic/assignment-tracking/sroa/vec-2.ll b/llvm/test/DebugInfo/Generic/assignment-tracking/sroa/vec-2.ll
index 49e08e02cdeb3..e04a5a9b58e0c 100644
--- a/llvm/test/DebugInfo/Generic/assignment-tracking/sroa/vec-2.ll
+++ b/llvm/test/DebugInfo/Generic/assignment-tracking/sroa/vec-2.ll
@@ -26,11 +26,12 @@
 ;; dbg.assign/DIExpression. Ensure that only the value-expression gets fragment
 ;; info; that the address-expression remains untouched.
 
-; CHECK:  [[ADDR:%.*]] = getelementptr inbounds i8, ptr %i.sroa.0, i64 12, !dbg !64
-; CHECK-NEXT:  store float %2, ptr [[ADDR]], align 4, !dbg !64, !DIAssignID !66
-;; There's a dbg intrinsics we're not interested in testing wedged in here.
+; CHECK: %i.sroa.4.12.vec.insert = insertelement <2 x float> %i.sroa.4.0.vec.insert, float %2, i32 1, !dbg
+;; There's a few dbg intrinsics we're not interested in testing wedged in here.
+; CHECK-NEXT: dbg.value
 ; CHECK-NEXT: dbg.assign
-; CHECK-NEXT: call void @llvm.dbg.assign(metadata float %2,{{.+}}, metadata !DIExpression(DW_OP_LLVM_fragment, 96, 32),{{.+}}, metadata ptr %i.sroa.0.12..sroa_idx2, metadata !DIExpression()), !dbg
+; CHECK-NEXT: dbg.assign
+; CHECK-NEXT: call void @llvm.dbg.assign(metadata float %2,{{.+}}, metadata !DIExpression(DW_OP_LLVM_fragment, 96, 32),{{.+}}, metadata ptr undef, metadata !DIExpression()), !dbg
 
 %class.d = type { %class.a }
 %class.a = type { [4 x float] }

diff  --git a/llvm/test/DebugInfo/X86/sroasplit-1.ll b/llvm/test/DebugInfo/X86/sroasplit-1.ll
index 44e3d735251b7..5ac5c828b11ed 100644
--- a/llvm/test/DebugInfo/X86/sroasplit-1.ll
+++ b/llvm/test/DebugInfo/X86/sroasplit-1.ll
@@ -20,8 +20,10 @@
 ;
 
 ; Verify that SROA creates a variable piece when splitting i1.
-; CHECK: %[[I1:.*]] = load <12 x i8>,
-; CHECK: call void @llvm.dbg.value(metadata <12 x i8> %[[I1]], metadata ![[VAR:.*]], metadata !DIExpression(DW_OP_LLVM_fragment, 32, 96))
+; CHECK: %[[I1:.*]] = alloca [12 x i8], align 4
+; CHECK: call void @llvm.dbg.declare(metadata ptr %[[I1]], metadata ![[VAR:[0-9]+]], metadata !DIExpression(DW_OP_LLVM_fragment, 32, 96))
+; CHECK: call void @llvm.dbg.value(metadata i32 %[[A:.*]], metadata ![[VAR]], metadata !DIExpression(DW_OP_LLVM_fragment, 0, 32))
+; CHECK: ret i32 %[[A]]
 ; Read Var and Piece:
 ; CHECK: ![[VAR]] = !DILocalVariable(name: "i1",{{.*}} line: 11,
 

diff  --git a/llvm/test/DebugInfo/X86/sroasplit-4.ll b/llvm/test/DebugInfo/X86/sroasplit-4.ll
index 6830944ed12bc..7f6cd3dde73a4 100644
--- a/llvm/test/DebugInfo/X86/sroasplit-4.ll
+++ b/llvm/test/DebugInfo/X86/sroasplit-4.ll
@@ -1,12 +1,12 @@
 ; RUN: opt -passes='sroa' < %s -S -o - | FileCheck %s
 ;
 ; Test that recursively splitting an alloca updates the debug info correctly.
-; CHECK: call void @llvm.dbg.value(metadata <16 x i8> %[[Y_VEC:.*]], metadata ![[Y:.*]], metadata !DIExpression())
-; CHECK: call void @llvm.dbg.value(metadata <16 x i8> %[[Y_VEC1:.*]], metadata ![[Y]], metadata !DIExpression())
-; CHECK: call void @llvm.dbg.value(metadata i32 0, metadata ![[R:.*]], metadata !DIExpression(DW_OP_LLVM_fragment, 0, 32))
-; CHECK: call void @llvm.dbg.value(metadata i64 0, metadata ![[R]], metadata !DIExpression(DW_OP_LLVM_fragment, 64, 64))
-; CHECK: call void @llvm.dbg.value(metadata i64 0, metadata ![[R]], metadata !DIExpression(DW_OP_LLVM_fragment, 128, 64))
-; CHECK: call void @llvm.dbg.value(metadata <16 x i8> %[[Y_VEC1]], metadata ![[R]], metadata !DIExpression(DW_OP_LLVM_fragment, 192, 128))
+; CHECK: %[[T:.*]] = load i64, ptr @t, align 8
+; CHECK: call void @llvm.dbg.value(metadata i64 %[[T]], metadata ![[Y:.*]], metadata !DIExpression(DW_OP_LLVM_fragment, 0, 64))
+; CHECK: %[[T1:.*]] = load i64, ptr @t, align 8
+; CHECK: call void @llvm.dbg.value(metadata i64 %[[T1]], metadata ![[Y]], metadata !DIExpression(DW_OP_LLVM_fragment, 64, 64))
+; CHECK: call void @llvm.dbg.value(metadata i64 %[[T]], metadata ![[R:.*]], metadata !DIExpression(DW_OP_LLVM_fragment, 192, 64))
+; CHECK: call void @llvm.dbg.value(metadata i64 %[[T1]], metadata ![[R]], metadata !DIExpression(DW_OP_LLVM_fragment, 256, 64))
 ;
 ; struct p {
 ;   __SIZE_TYPE__ s;

diff  --git a/llvm/test/Transforms/PhaseOrdering/instcombine-sroa-inttoptr.ll b/llvm/test/Transforms/PhaseOrdering/instcombine-sroa-inttoptr.ll
index 3361ad11c244b..ba6c36abd24a1 100644
--- a/llvm/test/Transforms/PhaseOrdering/instcombine-sroa-inttoptr.ll
+++ b/llvm/test/Transforms/PhaseOrdering/instcombine-sroa-inttoptr.ll
@@ -45,123 +45,100 @@
 
 target datalayout = "e-m:e-p270:32:32-p271:32:32-p272:64:64-i64:64-f80:128-n8:16:32:64-S128"
 
-%0 = type { i32*, i32, i32, i32 }
+%0 = type { ptr, i32, i32, i32 }
 
-define dso_local void @_Z3gen1S(%0* noalias sret(%0) align 8 %arg, %0* byval(%0) align 8 %arg1) {
+define dso_local void @_Z3gen1S(ptr noalias sret(%0) align 8 %arg, ptr byval(%0) align 8 %arg1) {
 ; CHECK-LABEL: @_Z3gen1S(
 ; CHECK-NEXT:  bb:
-; CHECK-NEXT:    [[I:%.*]] = getelementptr inbounds [[TMP0:%.*]], %0* [[ARG1:%.*]], i64 0, i32 0
-; CHECK-NEXT:    [[I2:%.*]] = load i32*, i32** [[I]], align 8
-; CHECK-NEXT:    [[I3:%.*]] = getelementptr inbounds [[TMP0]], %0* [[ARG:%.*]], i64 0, i32 0
-; CHECK-NEXT:    store i32* [[I2]], i32** [[I3]], align 8
+; CHECK-NEXT:    [[I2:%.*]] = load ptr, ptr [[ARG1:%.*]], align 8
+; CHECK-NEXT:    store ptr [[I2]], ptr [[ARG:%.*]], align 8
 ; CHECK-NEXT:    ret void
 ;
 bb:
-  %i = getelementptr inbounds %0, %0* %arg1, i32 0, i32 0
-  %i2 = load i32*, i32** %i, align 8
-  %i3 = getelementptr inbounds %0, %0* %arg, i32 0, i32 0
-  store i32* %i2, i32** %i3, align 8
+  %i2 = load ptr, ptr %arg1, align 8
+  store ptr %i2, ptr %arg, align 8
   ret void
 }
 
-define dso_local i32* @_Z3foo1S(%0* byval(%0) align 8 %arg) {
+define dso_local ptr @_Z3foo1S(ptr byval(%0) align 8 %arg) {
 ; CHECK-LABEL: @_Z3foo1S(
 ; CHECK-NEXT:  bb:
 ; CHECK-NEXT:    [[I2:%.*]] = alloca [[TMP0:%.*]], align 8
-; CHECK-NEXT:    [[TMP0]] = bitcast %0* [[ARG:%.*]] to i64*
-; CHECK-NEXT:    [[I11_SROA_0_0_VEC_EXTRACT_EXTRACT:%.*]] = load i64, i64* [[TMP0]], align 8
-; CHECK-NEXT:    [[TMP1:%.*]] = inttoptr i64 [[I11_SROA_0_0_VEC_EXTRACT_EXTRACT]] to i32*
-; CHECK-NEXT:    [[I_SROA_0_0_I6_SROA_IDX:%.*]] = getelementptr inbounds [[TMP0]], %0* [[I2]], i64 0, i32 0
-; CHECK-NEXT:    store i32* [[TMP1]], i32** [[I_SROA_0_0_I6_SROA_IDX]], align 8
-; CHECK-NEXT:    tail call void @_Z7escape01S(%0* nonnull byval([[TMP0]]) align 8 [[I2]])
-; CHECK-NEXT:    ret i32* [[TMP1]]
+; CHECK-NEXT:    [[I1_SROA_0_0_COPYLOAD:%.*]] = load ptr, ptr [[ARG:%.*]], align 8
+; CHECK-NEXT:    store ptr [[I1_SROA_0_0_COPYLOAD]], ptr [[I2]], align 8
+; CHECK-NEXT:    tail call void @_Z7escape01S(ptr nonnull byval([[TMP0]]) align 8 [[I2]])
+; CHECK-NEXT:    ret ptr [[I1_SROA_0_0_COPYLOAD]]
 ;
 bb:
   %i = alloca %0, align 8
   %i1 = alloca %0, align 8
   %i2 = alloca %0, align 8
-  %i3 = bitcast %0* %i to i8*
-  call void @llvm.lifetime.start.p0i8(i64 24, i8* %i3)
-  %i4 = bitcast %0* %i1 to i8*
-  %i5 = bitcast %0* %arg to i8*
-  call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 8 %i4, i8* align 8 %i5, i64 24, i1 false)
-  call void @_Z3gen1S(%0* sret(%0) align 8 %i, %0* byval(%0) align 8 %i1)
-  %i6 = bitcast %0* %i2 to i8*
-  %i7 = bitcast %0* %i to i8*
-  call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 8 %i6, i8* align 8 %i7, i64 24, i1 false)
-  call void @_Z7escape01S(%0* byval(%0) align 8 %i2)
-  %i8 = getelementptr inbounds %0, %0* %i, i32 0, i32 0
-  %i9 = load i32*, i32** %i8, align 8
-  %i10 = bitcast %0* %i to i8*
-  call void @llvm.lifetime.end.p0i8(i64 24, i8* %i10)
-  ret i32* %i9
+  call void @llvm.lifetime.start.p0(i64 24, ptr %i)
+  call void @llvm.memcpy.p0.p0.i64(ptr align 8 %i1, ptr align 8 %arg, i64 24, i1 false)
+  call void @_Z3gen1S(ptr sret(%0) align 8 %i, ptr byval(%0) align 8 %i1)
+  call void @llvm.memcpy.p0.p0.i64(ptr align 8 %i2, ptr align 8 %i, i64 24, i1 false)
+  call void @_Z7escape01S(ptr byval(%0) align 8 %i2)
+  %i9 = load ptr, ptr %i, align 8
+  call void @llvm.lifetime.end.p0(i64 24, ptr %i)
+  ret ptr %i9
 }
 
-declare void @llvm.lifetime.start.p0i8(i64 immarg, i8* nocapture)
+declare void @llvm.lifetime.start.p0(i64 immarg, ptr nocapture)
 
-declare void @llvm.memcpy.p0i8.p0i8.i64(i8* noalias nocapture writeonly, i8* noalias nocapture readonly, i64, i1 immarg)
+declare void @llvm.memcpy.p0.p0.i64(ptr noalias nocapture writeonly, ptr noalias nocapture readonly, i64, i1 immarg)
 
-declare dso_local void @_Z7escape01S(%0* byval(%0) align 8)
+declare dso_local void @_Z7escape01S(ptr byval(%0) align 8)
 
-declare void @llvm.lifetime.end.p0i8(i64 immarg, i8* nocapture)
+declare void @llvm.lifetime.end.p0(i64 immarg, ptr nocapture)
 
-define dso_local i32* @_Z3bar1S(%0* byval(%0) align 8 %arg) {
+define dso_local ptr @_Z3bar1S(ptr byval(%0) align 8 %arg) {
 ; CHECK-LABEL: @_Z3bar1S(
 ; CHECK-NEXT:  bb:
-; CHECK-NEXT:    [[TMP0:%.*]] = bitcast %0* [[ARG:%.*]] to i64*
-; CHECK-NEXT:    [[I13_SROA_0_0_VEC_EXTRACT_EXTRACT:%.*]] = load i64, i64* [[TMP0]], align 8
-; CHECK-NEXT:    [[TMP1:%.*]] = inttoptr i64 [[I13_SROA_0_0_VEC_EXTRACT_EXTRACT]] to i32*
+; CHECK-NEXT:    [[I1_SROA_0_0_COPYLOAD:%.*]] = load ptr, ptr [[ARG:%.*]], align 8
 ; CHECK-NEXT:    [[I5:%.*]] = tail call i32 @_Z4condv()
 ; CHECK-NEXT:    [[I6_NOT:%.*]] = icmp eq i32 [[I5]], 0
 ; CHECK-NEXT:    br i1 [[I6_NOT]], label [[BB10:%.*]], label [[BB7:%.*]]
 ; CHECK:       bb7:
 ; CHECK-NEXT:    tail call void @_Z5sync0v()
-; CHECK-NEXT:    tail call void @_Z7escape0Pi(i32* [[TMP1]])
+; CHECK-NEXT:    tail call void @_Z7escape0Pi(ptr [[I1_SROA_0_0_COPYLOAD]])
 ; CHECK-NEXT:    br label [[BB13:%.*]]
 ; CHECK:       bb10:
 ; CHECK-NEXT:    tail call void @_Z5sync1v()
-; CHECK-NEXT:    tail call void @_Z7escape1Pi(i32* [[TMP1]])
+; CHECK-NEXT:    tail call void @_Z7escape1Pi(ptr [[I1_SROA_0_0_COPYLOAD]])
 ; CHECK-NEXT:    br label [[BB13]]
 ; CHECK:       bb13:
-; CHECK-NEXT:    ret i32* [[TMP1]]
+; CHECK-NEXT:    ret ptr [[I1_SROA_0_0_COPYLOAD]]
 ;
 bb:
   %i = alloca %0, align 8
   %i1 = alloca %0, align 8
-  %i2 = bitcast %0* %i to i8*
-  call void @llvm.lifetime.start.p0i8(i64 24, i8* %i2)
-  %i3 = bitcast %0* %i1 to i8*
-  %i4 = bitcast %0* %arg to i8*
-  call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 8 %i3, i8* align 8 %i4, i64 24, i1 false)
-  call void @_Z3gen1S(%0* sret(%0) align 8 %i, %0* byval(%0) align 8 %i1)
+  call void @llvm.lifetime.start.p0(i64 24, ptr %i)
+  call void @llvm.memcpy.p0.p0.i64(ptr align 8 %i1, ptr align 8 %arg, i64 24, i1 false)
+  call void @_Z3gen1S(ptr sret(%0) align 8 %i, ptr byval(%0) align 8 %i1)
   %i5 = call i32 @_Z4condv()
   %i6 = icmp ne i32 %i5, 0
   br i1 %i6, label %bb7, label %bb10
 
 bb7:
   call void @_Z5sync0v()
-  %i8 = getelementptr inbounds %0, %0* %i, i32 0, i32 0
-  %i9 = load i32*, i32** %i8, align 8
-  call void @_Z7escape0Pi(i32* %i9)
+  %i9 = load ptr, ptr %i, align 8
+  call void @_Z7escape0Pi(ptr %i9)
   br label %bb13
 
 bb10:
   call void @_Z5sync1v()
-  %i11 = getelementptr inbounds %0, %0* %i, i32 0, i32 0
-  %i12 = load i32*, i32** %i11, align 8
-  call void @_Z7escape1Pi(i32* %i12)
+  %i12 = load ptr, ptr %i, align 8
+  call void @_Z7escape1Pi(ptr %i12)
   br label %bb13
 
 bb13:
-  %i14 = getelementptr inbounds %0, %0* %i, i32 0, i32 0
-  %i15 = load i32*, i32** %i14, align 8
-  %i16 = bitcast %0* %i to i8*
-  call void @llvm.lifetime.end.p0i8(i64 24, i8* %i16)
-  ret i32* %i15
+  %i15 = load ptr, ptr %i, align 8
+  call void @llvm.lifetime.end.p0(i64 24, ptr %i)
+  ret ptr %i15
 }
 
 declare dso_local i32 @_Z4condv()
 declare dso_local void @_Z5sync0v()
-declare dso_local void @_Z7escape0Pi(i32*)
+declare dso_local void @_Z7escape0Pi(ptr)
 declare dso_local void @_Z5sync1v()
-declare dso_local void @_Z7escape1Pi(i32*)
+declare dso_local void @_Z7escape1Pi(ptr)

diff  --git a/llvm/test/Transforms/SROA/address-spaces.ll b/llvm/test/Transforms/SROA/address-spaces.ll
index 8d8f49adcef95..f90dbdf0cd1ee 100644
--- a/llvm/test/Transforms/SROA/address-spaces.ll
+++ b/llvm/test/Transforms/SROA/address-spaces.ll
@@ -12,8 +12,8 @@ declare void @llvm.memcpy.p1.p1.i32(ptr addrspace(1) nocapture, ptr addrspace(1)
 ; Make sure an illegal bitcast isn't introduced
 define void @test_address_space_1_1(ptr addrspace(1) %a, ptr addrspace(1) %b) {
 ; CHECK-LABEL: @test_address_space_1_1(
-; CHECK-NEXT:    [[AA_SROA_0_0_COPYLOAD:%.*]] = load <16 x i8>, ptr addrspace(1) [[A:%.*]], align 2
-; CHECK-NEXT:    store <16 x i8> [[AA_SROA_0_0_COPYLOAD]], ptr addrspace(1) [[B:%.*]], align 2
+; CHECK-NEXT:    [[AA_0_COPYLOAD:%.*]] = load <2 x i64>, ptr addrspace(1) [[A:%.*]], align 2
+; CHECK-NEXT:    store <2 x i64> [[AA_0_COPYLOAD]], ptr addrspace(1) [[B:%.*]], align 2
 ; CHECK-NEXT:    ret void
 ;
   %aa = alloca <2 x i64>, align 16
@@ -24,8 +24,8 @@ define void @test_address_space_1_1(ptr addrspace(1) %a, ptr addrspace(1) %b) {
 
 define void @test_address_space_1_0(ptr addrspace(1) %a, ptr %b) {
 ; CHECK-LABEL: @test_address_space_1_0(
-; CHECK-NEXT:    [[AA_SROA_0_0_COPYLOAD:%.*]] = load <16 x i8>, ptr addrspace(1) [[A:%.*]], align 2
-; CHECK-NEXT:    store <16 x i8> [[AA_SROA_0_0_COPYLOAD]], ptr [[B:%.*]], align 2
+; CHECK-NEXT:    [[AA_0_COPYLOAD:%.*]] = load <2 x i64>, ptr addrspace(1) [[A:%.*]], align 2
+; CHECK-NEXT:    store <2 x i64> [[AA_0_COPYLOAD]], ptr [[B:%.*]], align 2
 ; CHECK-NEXT:    ret void
 ;
   %aa = alloca <2 x i64>, align 16
@@ -36,8 +36,8 @@ define void @test_address_space_1_0(ptr addrspace(1) %a, ptr %b) {
 
 define void @test_address_space_0_1(ptr %a, ptr addrspace(1) %b) {
 ; CHECK-LABEL: @test_address_space_0_1(
-; CHECK-NEXT:    [[AA_SROA_0_0_COPYLOAD:%.*]] = load <16 x i8>, ptr [[A:%.*]], align 2
-; CHECK-NEXT:    store <16 x i8> [[AA_SROA_0_0_COPYLOAD]], ptr addrspace(1) [[B:%.*]], align 2
+; CHECK-NEXT:    [[AA_0_COPYLOAD:%.*]] = load <2 x i64>, ptr [[A:%.*]], align 2
+; CHECK-NEXT:    store <2 x i64> [[AA_0_COPYLOAD]], ptr addrspace(1) [[B:%.*]], align 2
 ; CHECK-NEXT:    ret void
 ;
   %aa = alloca <2 x i64>, align 16

diff  --git a/llvm/test/Transforms/SROA/alignment.ll b/llvm/test/Transforms/SROA/alignment.ll
index 72a8c3c47b491..b870a15b312ca 100644
--- a/llvm/test/Transforms/SROA/alignment.ll
+++ b/llvm/test/Transforms/SROA/alignment.ll
@@ -92,15 +92,15 @@ define void @PR13920(ptr %a, ptr %b) {
 ;
 ; CHECK-LABEL: @PR13920(
 ; CHECK-NEXT:  entry:
-; CHECK-NEXT:    [[AA_SROA_0_0_COPYLOAD:%.*]] = load <16 x i8>, ptr [[A:%.*]], align 2
-; CHECK-NEXT:    store <16 x i8> [[AA_SROA_0_0_COPYLOAD]], ptr [[B:%.*]], align 2
+; CHECK-NEXT:    [[AA_0_COPYLOAD:%.*]] = load <2 x i64>, ptr [[A:%.*]], align 2
+; CHECK-NEXT:    store <2 x i64> [[AA_0_COPYLOAD]], ptr [[B:%.*]], align 2
 ; CHECK-NEXT:    ret void
 ;
 ; CHECK-DEBUGLOC-LABEL: @PR13920(
 ; CHECK-DEBUGLOC-NEXT:  entry:
 ; CHECK-DEBUGLOC-NEXT:    call void @llvm.dbg.value(metadata ptr undef, metadata [[META37:![0-9]+]], metadata !DIExpression()), !dbg [[DBG38:![0-9]+]]
-; CHECK-DEBUGLOC-NEXT:    [[AA_SROA_0_0_COPYLOAD:%.*]] = load <16 x i8>, ptr [[A:%.*]], align 2, !dbg [[DBG39:![0-9]+]]
-; CHECK-DEBUGLOC-NEXT:    store <16 x i8> [[AA_SROA_0_0_COPYLOAD]], ptr [[B:%.*]], align 2, !dbg [[DBG40:![0-9]+]]
+; CHECK-DEBUGLOC-NEXT:    [[AA_0_COPYLOAD:%.*]] = load <2 x i64>, ptr [[A:%.*]], align 2, !dbg [[DBG39:![0-9]+]]
+; CHECK-DEBUGLOC-NEXT:    store <2 x i64> [[AA_0_COPYLOAD]], ptr [[B:%.*]], align 2, !dbg [[DBG40:![0-9]+]]
 ; CHECK-DEBUGLOC-NEXT:    ret void, !dbg [[DBG41:![0-9]+]]
 ;
 entry:
@@ -119,17 +119,21 @@ define void @test3(ptr %x) {
 ;
 ; CHECK-LABEL: @test3(
 ; CHECK-NEXT:  entry:
-; CHECK-NEXT:    [[A_SROA_0_0_COPYLOAD:%.*]] = load <22 x i8>, ptr [[X:%.*]], align 8
-; CHECK-NEXT:    [[B_SROA_0_6_COPYLOAD:%.*]] = load <18 x i8>, ptr [[X]], align 2
+; CHECK-NEXT:    [[A_SROA_0:%.*]] = alloca [22 x i8], align 8
+; CHECK-NEXT:    [[B_SROA_0:%.*]] = alloca [18 x i8], align 2
+; CHECK-NEXT:    call void @llvm.memcpy.p0.p0.i32(ptr align 8 [[A_SROA_0]], ptr align 8 [[X:%.*]], i32 22, i1 false)
+; CHECK-NEXT:    call void @llvm.memcpy.p0.p0.i32(ptr align 2 [[B_SROA_0]], ptr align 2 [[X]], i32 18, i1 false)
 ; CHECK-NEXT:    ret void
 ;
 ; CHECK-DEBUGLOC-LABEL: @test3(
 ; CHECK-DEBUGLOC-NEXT:  entry:
-; CHECK-DEBUGLOC-NEXT:    call void @llvm.dbg.value(metadata ptr undef, metadata [[META44:![0-9]+]], metadata !DIExpression()), !dbg [[DBG47:![0-9]+]]
-; CHECK-DEBUGLOC-NEXT:    call void @llvm.dbg.value(metadata ptr undef, metadata [[META45:![0-9]+]], metadata !DIExpression()), !dbg [[DBG48:![0-9]+]]
-; CHECK-DEBUGLOC-NEXT:    [[A_SROA_0_0_COPYLOAD:%.*]] = load <22 x i8>, ptr [[X:%.*]], align 8, !dbg [[DBG49:![0-9]+]]
+; CHECK-DEBUGLOC-NEXT:    [[A_SROA_0:%.*]] = alloca [22 x i8], align 8, !dbg [[DBG47:![0-9]+]]
+; CHECK-DEBUGLOC-NEXT:    call void @llvm.dbg.value(metadata ptr undef, metadata [[META44:![0-9]+]], metadata !DIExpression()), !dbg [[DBG47]]
+; CHECK-DEBUGLOC-NEXT:    [[B_SROA_0:%.*]] = alloca [18 x i8], align 2, !dbg [[DBG48:![0-9]+]]
+; CHECK-DEBUGLOC-NEXT:    call void @llvm.dbg.value(metadata ptr undef, metadata [[META45:![0-9]+]], metadata !DIExpression()), !dbg [[DBG48]]
+; CHECK-DEBUGLOC-NEXT:    call void @llvm.memcpy.p0.p0.i32(ptr align 8 [[A_SROA_0]], ptr align 8 [[X:%.*]], i32 22, i1 false), !dbg [[DBG49:![0-9]+]]
 ; CHECK-DEBUGLOC-NEXT:    call void @llvm.dbg.value(metadata ptr undef, metadata [[META46:![0-9]+]], metadata !DIExpression()), !dbg [[DBG50:![0-9]+]]
-; CHECK-DEBUGLOC-NEXT:    [[B_SROA_0_6_COPYLOAD:%.*]] = load <18 x i8>, ptr [[X]], align 2, !dbg [[DBG51:![0-9]+]]
+; CHECK-DEBUGLOC-NEXT:    call void @llvm.memcpy.p0.p0.i32(ptr align 2 [[B_SROA_0]], ptr align 2 [[X]], i32 18, i1 false), !dbg [[DBG51:![0-9]+]]
 ; CHECK-DEBUGLOC-NEXT:    ret void, !dbg [[DBG52:![0-9]+]]
 ;
 entry:

diff  --git a/llvm/test/Transforms/SROA/alloca-address-space.ll b/llvm/test/Transforms/SROA/alloca-address-space.ll
index 877b894971f33..4c638a9235bd3 100644
--- a/llvm/test/Transforms/SROA/alloca-address-space.ll
+++ b/llvm/test/Transforms/SROA/alloca-address-space.ll
@@ -11,8 +11,8 @@ declare void @llvm.memcpy.p1.p1.i32(ptr addrspace(1) nocapture, ptr addrspace(1)
 
 define void @test_address_space_1_1(ptr addrspace(1) %a, ptr addrspace(1) %b) {
 ; CHECK-LABEL: @test_address_space_1_1(
-; CHECK-NEXT:    [[AA_SROA_0_0_COPYLOAD:%.*]] = load <16 x i8>, ptr addrspace(1) [[A:%.*]], align 2
-; CHECK-NEXT:    store <16 x i8> [[AA_SROA_0_0_COPYLOAD]], ptr addrspace(1) [[B:%.*]], align 2
+; CHECK-NEXT:    [[AA_0_COPYLOAD:%.*]] = load <2 x i64>, ptr addrspace(1) [[A:%.*]], align 2
+; CHECK-NEXT:    store <2 x i64> [[AA_0_COPYLOAD]], ptr addrspace(1) [[B:%.*]], align 2
 ; CHECK-NEXT:    ret void
 ;
   %aa = alloca <2 x i64>, align 16, addrspace(2)
@@ -23,8 +23,8 @@ define void @test_address_space_1_1(ptr addrspace(1) %a, ptr addrspace(1) %b) {
 
 define void @test_address_space_1_0(ptr addrspace(1) %a, ptr addrspace(2) %b) {
 ; CHECK-LABEL: @test_address_space_1_0(
-; CHECK-NEXT:    [[AA_SROA_0_0_COPYLOAD:%.*]] = load <16 x i8>, ptr addrspace(1) [[A:%.*]], align 2
-; CHECK-NEXT:    store <16 x i8> [[AA_SROA_0_0_COPYLOAD]], ptr addrspace(2) [[B:%.*]], align 2
+; CHECK-NEXT:    [[AA_0_COPYLOAD:%.*]] = load <2 x i64>, ptr addrspace(1) [[A:%.*]], align 2
+; CHECK-NEXT:    store <2 x i64> [[AA_0_COPYLOAD]], ptr addrspace(2) [[B:%.*]], align 2
 ; CHECK-NEXT:    ret void
 ;
   %aa = alloca <2 x i64>, align 16, addrspace(2)
@@ -35,8 +35,8 @@ define void @test_address_space_1_0(ptr addrspace(1) %a, ptr addrspace(2) %b) {
 
 define void @test_address_space_0_1(ptr addrspace(2) %a, ptr addrspace(1) %b) {
 ; CHECK-LABEL: @test_address_space_0_1(
-; CHECK-NEXT:    [[AA_SROA_0_0_COPYLOAD:%.*]] = load <16 x i8>, ptr addrspace(2) [[A:%.*]], align 2
-; CHECK-NEXT:    store <16 x i8> [[AA_SROA_0_0_COPYLOAD]], ptr addrspace(1) [[B:%.*]], align 2
+; CHECK-NEXT:    [[AA_0_COPYLOAD:%.*]] = load <2 x i64>, ptr addrspace(2) [[A:%.*]], align 2
+; CHECK-NEXT:    store <2 x i64> [[AA_0_COPYLOAD]], ptr addrspace(1) [[B:%.*]], align 2
 ; CHECK-NEXT:    ret void
 ;
   %aa = alloca <2 x i64>, align 16, addrspace(2)

diff  --git a/llvm/test/Transforms/SROA/basictest.ll b/llvm/test/Transforms/SROA/basictest.ll
index ebac0d42e9fbd..de7e11d1e491c 100644
--- a/llvm/test/Transforms/SROA/basictest.ll
+++ b/llvm/test/Transforms/SROA/basictest.ll
@@ -134,83 +134,100 @@ L2:
 define void @test3(ptr %dst, ptr align 8 %src) {
 ; CHECK-LABEL: @test3(
 ; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[A_SROA_0:%.*]] = alloca [42 x i8], align 1
 ; CHECK-NEXT:    [[A_SROA_3:%.*]] = alloca [99 x i8], align 1
+; CHECK-NEXT:    [[A_SROA_32:%.*]] = alloca [16 x i8], align 1
+; CHECK-NEXT:    [[A_SROA_15:%.*]] = alloca [42 x i8], align 1
+; CHECK-NEXT:    [[A_SROA_16:%.*]] = alloca [7 x i8], align 1
+; CHECK-NEXT:    [[A_SROA_235:%.*]] = alloca [7 x i8], align 1
 ; CHECK-NEXT:    [[A_SROA_31:%.*]] = alloca [85 x i8], align 1
-; CHECK-NEXT:    [[A_SROA_0_SROA_0_0_COPYLOAD:%.*]] = load <42 x i8>, ptr [[SRC:%.*]], align 8, !tbaa [[TBAA0:![0-9]+]]
+; CHECK-NEXT:    call void @llvm.memcpy.p0.p0.i32(ptr align 1 [[A_SROA_0]], ptr align 8 [[SRC:%.*]], i32 42, i1 false), !tbaa [[TBAA0:![0-9]+]]
 ; CHECK-NEXT:    [[A_SROA_2_0_SRC_SROA_IDX:%.*]] = getelementptr inbounds i8, ptr [[SRC]], i64 42
 ; CHECK-NEXT:    [[A_SROA_2_0_COPYLOAD:%.*]] = load i8, ptr [[A_SROA_2_0_SRC_SROA_IDX]], align 2, !tbaa [[TBAA0]]
 ; CHECK-NEXT:    [[A_SROA_3_0_SRC_SROA_IDX:%.*]] = getelementptr inbounds i8, ptr [[SRC]], i64 43
 ; CHECK-NEXT:    call void @llvm.memcpy.p0.p0.i32(ptr align 1 [[A_SROA_3]], ptr align 1 [[A_SROA_3_0_SRC_SROA_IDX]], i32 99, i1 false), !tbaa [[TBAA0]]
 ; CHECK-NEXT:    [[A_SROA_32_0_SRC_SROA_IDX:%.*]] = getelementptr inbounds i8, ptr [[SRC]], i64 142
-; CHECK-NEXT:    [[A_SROA_32_SROA_0_0_COPYLOAD:%.*]] = load <16 x i8>, ptr [[A_SROA_32_0_SRC_SROA_IDX]], align 2, !tbaa [[TBAA0]]
+; CHECK-NEXT:    call void @llvm.memcpy.p0.p0.i32(ptr align 1 [[A_SROA_32]], ptr align 2 [[A_SROA_32_0_SRC_SROA_IDX]], i32 16, i1 false), !tbaa [[TBAA0]]
 ; CHECK-NEXT:    [[A_SROA_15_0_SRC_SROA_IDX:%.*]] = getelementptr inbounds i8, ptr [[SRC]], i64 158
-; CHECK-NEXT:    [[A_SROA_15_SROA_0_0_COPYLOAD:%.*]] = load <42 x i8>, ptr [[A_SROA_15_0_SRC_SROA_IDX]], align 2, !tbaa [[TBAA0]]
+; CHECK-NEXT:    call void @llvm.memcpy.p0.p0.i32(ptr align 1 [[A_SROA_15]], ptr align 2 [[A_SROA_15_0_SRC_SROA_IDX]], i32 42, i1 false), !tbaa [[TBAA0]]
 ; CHECK-NEXT:    [[A_SROA_16_0_SRC_SROA_IDX:%.*]] = getelementptr inbounds i8, ptr [[SRC]], i64 200
-; CHECK-NEXT:    [[A_SROA_16_SROA_0_0_COPYLOAD:%.*]] = load <7 x i8>, ptr [[A_SROA_16_0_SRC_SROA_IDX]], align 8, !tbaa [[TBAA0]]
+; CHECK-NEXT:    call void @llvm.memcpy.p0.p0.i32(ptr align 1 [[A_SROA_16]], ptr align 8 [[A_SROA_16_0_SRC_SROA_IDX]], i32 7, i1 false), !tbaa [[TBAA0]]
 ; CHECK-NEXT:    [[A_SROA_23_0_SRC_SROA_IDX:%.*]] = getelementptr inbounds i8, ptr [[SRC]], i64 207
 ; CHECK-NEXT:    [[A_SROA_23_0_COPYLOAD:%.*]] = load i8, ptr [[A_SROA_23_0_SRC_SROA_IDX]], align 1, !tbaa [[TBAA0]]
 ; CHECK-NEXT:    [[A_SROA_235_0_SRC_SROA_IDX:%.*]] = getelementptr inbounds i8, ptr [[SRC]], i64 208
-; CHECK-NEXT:    [[A_SROA_235_SROA_0_0_COPYLOAD:%.*]] = load <7 x i8>, ptr [[A_SROA_235_0_SRC_SROA_IDX]], align 8, !tbaa [[TBAA0]]
+; CHECK-NEXT:    call void @llvm.memcpy.p0.p0.i32(ptr align 1 [[A_SROA_235]], ptr align 8 [[A_SROA_235_0_SRC_SROA_IDX]], i32 7, i1 false), !tbaa [[TBAA0]]
 ; CHECK-NEXT:    [[A_SROA_31_0_SRC_SROA_IDX:%.*]] = getelementptr inbounds i8, ptr [[SRC]], i64 215
 ; CHECK-NEXT:    call void @llvm.memcpy.p0.p0.i32(ptr align 1 [[A_SROA_31]], ptr align 1 [[A_SROA_31_0_SRC_SROA_IDX]], i32 85, i1 false), !tbaa [[TBAA0]]
-; CHECK-NEXT:    [[A_SROA_32_SROA_0_0_VEC_INSERT:%.*]] = insertelement <16 x i8> [[A_SROA_32_SROA_0_0_COPYLOAD]], i8 1, i32 0
-; CHECK-NEXT:    [[A_SROA_32_SROA_0_0_VECBLEND28:%.*]] = select <16 x i1> <i1 true, i1 true, i1 false, i1 false, i1 false, i1 false, i1 false, i1 false, i1 false, i1 false, i1 false, i1 false, i1 false, i1 false, i1 false, i1 false>, <16 x i8> <i8 extractelement (<2 x i8> bitcast (<1 x i16> <i16 1> to <2 x i8>), i32 0), i8 extractelement (<2 x i8> bitcast (<1 x i16> <i16 1> to <2 x i8>), i32 1), i8 undef, i8 undef, i8 undef, i8 undef, i8 undef, i8 undef, i8 undef, i8 undef, i8 undef, i8 undef, i8 undef, i8 undef, i8 undef, i8 undef>, <16 x i8> [[A_SROA_32_SROA_0_0_VEC_INSERT]]
-; CHECK-NEXT:    [[A_SROA_32_SROA_0_0_VECBLEND26:%.*]] = select <16 x i1> <i1 true, i1 true, i1 true, i1 true, i1 false, i1 false, i1 false, i1 false, i1 false, i1 false, i1 false, i1 false, i1 false, i1 false, i1 false, i1 false>, <16 x i8> <i8 extractelement (<4 x i8> bitcast (<1 x i32> <i32 1> to <4 x i8>), i32 0), i8 extractelement (<4 x i8> bitcast (<1 x i32> <i32 1> to <4 x i8>), i32 1), i8 extractelement (<4 x i8> bitcast (<1 x i32> <i32 1> to <4 x i8>), i32 2), i8 extractelement (<4 x i8> bitcast (<1 x i32> <i32 1> to <4 x i8>), i32 3), i8 undef, i8 undef, i8 undef, i8 undef, i8 undef, i8 undef, i8 undef, i8 undef, i8 undef, i8 undef, i8 undef, i8 undef>, <16 x i8> [[A_SROA_32_SROA_0_0_VECBLEND28]]
-; CHECK-NEXT:    [[A_SROA_32_SROA_0_0_VECBLEND:%.*]] = select <16 x i1> <i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 false, i1 false, i1 false, i1 false, i1 false, i1 false, i1 false, i1 false>, <16 x i8> <i8 extractelement (<8 x i8> bitcast (<1 x i64> <i64 1> to <8 x i8>), i32 0), i8 extractelement (<8 x i8> bitcast (<1 x i64> <i64 1> to <8 x i8>), i32 1), i8 extractelement (<8 x i8> bitcast (<1 x i64> <i64 1> to <8 x i8>), i32 2), i8 extractelement (<8 x i8> bitcast (<1 x i64> <i64 1> to <8 x i8>), i32 3), i8 extractelement (<8 x i8> bitcast (<1 x i64> <i64 1> to <8 x i8>), i32 4), i8 extractelement (<8 x i8> bitcast (<1 x i64> <i64 1> to <8 x i8>), i32 5), i8 extractelement (<8 x i8> bitcast (<1 x i64> <i64 1> to <8 x i8>), i32 6), i8 extractelement (<8 x i8> bitcast (<1 x i64> <i64 1> to <8 x i8>), i32 7), i8 undef, i8 undef, i8 undef, i8 undef, i8 undef, i8 undef, i8 undef, i8 undef>, <16 x i8> [[A_SROA_32_SROA_0_0_VECBLEND26]]
-; CHECK-NEXT:    [[A_SROA_32_SROA_0_1_VECBLEND:%.*]] = select <16 x i1> <i1 false, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 false, i1 false, i1 false, i1 false, i1 false, i1 false, i1 false>, <16 x i8> <i8 undef, i8 extractelement (<8 x i8> bitcast (<1 x i64> <i64 2> to <8 x i8>), i32 0), i8 extractelement (<8 x i8> bitcast (<1 x i64> <i64 2> to <8 x i8>), i32 1), i8 extractelement (<8 x i8> bitcast (<1 x i64> <i64 2> to <8 x i8>), i32 2), i8 extractelement (<8 x i8> bitcast (<1 x i64> <i64 2> to <8 x i8>), i32 3), i8 extractelement (<8 x i8> bitcast (<1 x i64> <i64 2> to <8 x i8>), i32 4), i8 extractelement (<8 x i8> bitcast (<1 x i64> <i64 2> to <8 x i8>), i32 5), i8 extractelement (<8 x i8> bitcast (<1 x i64> <i64 2> to <8 x i8>), i32 6), i8 extractelement (<8 x i8> bitcast (<1 x i64> <i64 2> to <8 x i8>), i32 7), i8 undef, i8 undef, i8 undef, i8 undef, i8 undef, i8 undef, i8 undef>, <16 x i8> [[A_SROA_32_SROA_0_0_VECBLEND]]
-; CHECK-NEXT:    [[A_SROA_32_SROA_0_2_VECBLEND:%.*]] = select <16 x i1> <i1 false, i1 false, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 false, i1 false, i1 false, i1 false, i1 false, i1 false>, <16 x i8> <i8 undef, i8 undef, i8 extractelement (<8 x i8> bitcast (<1 x i64> <i64 3> to <8 x i8>), i32 0), i8 extractelement (<8 x i8> bitcast (<1 x i64> <i64 3> to <8 x i8>), i32 1), i8 extractelement (<8 x i8> bitcast (<1 x i64> <i64 3> to <8 x i8>), i32 2), i8 extractelement (<8 x i8> bitcast (<1 x i64> <i64 3> to <8 x i8>), i32 3), i8 extractelement (<8 x i8> bitcast (<1 x i64> <i64 3> to <8 x i8>), i32 4), i8 extractelement (<8 x i8> bitcast (<1 x i64> <i64 3> to <8 x i8>), i32 5), i8 extractelement (<8 x i8> bitcast (<1 x i64> <i64 3> to <8 x i8>), i32 6), i8 extractelement (<8 x i8> bitcast (<1 x i64> <i64 3> to <8 x i8>), i32 7), i8 undef, i8 undef, i8 undef, i8 undef, i8 undef, i8 undef>, <16 x i8> [[A_SROA_32_SROA_0_1_VECBLEND]]
-; CHECK-NEXT:    [[A_SROA_32_SROA_0_3_VECBLEND:%.*]] = select <16 x i1> <i1 false, i1 false, i1 false, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 false, i1 false, i1 false, i1 false, i1 false>, <16 x i8> <i8 undef, i8 undef, i8 undef, i8 extractelement (<8 x i8> bitcast (<1 x i64> <i64 4> to <8 x i8>), i32 0), i8 extractelement (<8 x i8> bitcast (<1 x i64> <i64 4> to <8 x i8>), i32 1), i8 extractelement (<8 x i8> bitcast (<1 x i64> <i64 4> to <8 x i8>), i32 2), i8 extractelement (<8 x i8> bitcast (<1 x i64> <i64 4> to <8 x i8>), i32 3), i8 extractelement (<8 x i8> bitcast (<1 x i64> <i64 4> to <8 x i8>), i32 4), i8 extractelement (<8 x i8> bitcast (<1 x i64> <i64 4> to <8 x i8>), i32 5), i8 extractelement (<8 x i8> bitcast (<1 x i64> <i64 4> to <8 x i8>), i32 6), i8 extractelement (<8 x i8> bitcast (<1 x i64> <i64 4> to <8 x i8>), i32 7), i8 undef, i8 undef, i8 undef, i8 undef, i8 undef>, <16 x i8> [[A_SROA_32_SROA_0_2_VECBLEND]]
-; CHECK-NEXT:    [[A_SROA_32_SROA_0_4_VECBLEND:%.*]] = select <16 x i1> <i1 false, i1 false, i1 false, i1 false, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 false, i1 false, i1 false, i1 false>, <16 x i8> <i8 undef, i8 undef, i8 undef, i8 undef, i8 extractelement (<8 x i8> bitcast (<1 x i64> <i64 5> to <8 x i8>), i32 0), i8 extractelement (<8 x i8> bitcast (<1 x i64> <i64 5> to <8 x i8>), i32 1), i8 extractelement (<8 x i8> bitcast (<1 x i64> <i64 5> to <8 x i8>), i32 2), i8 extractelement (<8 x i8> bitcast (<1 x i64> <i64 5> to <8 x i8>), i32 3), i8 extractelement (<8 x i8> bitcast (<1 x i64> <i64 5> to <8 x i8>), i32 4), i8 extractelement (<8 x i8> bitcast (<1 x i64> <i64 5> to <8 x i8>), i32 5), i8 extractelement (<8 x i8> bitcast (<1 x i64> <i64 5> to <8 x i8>), i32 6), i8 extractelement (<8 x i8> bitcast (<1 x i64> <i64 5> to <8 x i8>), i32 7), i8 undef, i8 undef, i8 undef, i8 undef>, <16 x i8> [[A_SROA_32_SROA_0_3_VECBLEND]]
-; CHECK-NEXT:    [[A_SROA_32_SROA_0_5_VECBLEND:%.*]] = select <16 x i1> <i1 false, i1 false, i1 false, i1 false, i1 false, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 false, i1 false, i1 false>, <16 x i8> <i8 undef, i8 undef, i8 undef, i8 undef, i8 undef, i8 extractelement (<8 x i8> bitcast (<1 x i64> <i64 6> to <8 x i8>), i32 0), i8 extractelement (<8 x i8> bitcast (<1 x i64> <i64 6> to <8 x i8>), i32 1), i8 extractelement (<8 x i8> bitcast (<1 x i64> <i64 6> to <8 x i8>), i32 2), i8 extractelement (<8 x i8> bitcast (<1 x i64> <i64 6> to <8 x i8>), i32 3), i8 extractelement (<8 x i8> bitcast (<1 x i64> <i64 6> to <8 x i8>), i32 4), i8 extractelement (<8 x i8> bitcast (<1 x i64> <i64 6> to <8 x i8>), i32 5), i8 extractelement (<8 x i8> bitcast (<1 x i64> <i64 6> to <8 x i8>), i32 6), i8 extractelement (<8 x i8> bitcast (<1 x i64> <i64 6> to <8 x i8>), i32 7), i8 undef, i8 undef, i8 undef>, <16 x i8> [[A_SROA_32_SROA_0_4_VECBLEND]]
-; CHECK-NEXT:    [[A_SROA_32_SROA_0_6_VECBLEND:%.*]] = select <16 x i1> <i1 false, i1 false, i1 false, i1 false, i1 false, i1 false, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 false, i1 false>, <16 x i8> <i8 undef, i8 undef, i8 undef, i8 undef, i8 undef, i8 undef, i8 extractelement (<8 x i8> bitcast (<1 x i64> <i64 7> to <8 x i8>), i32 0), i8 extractelement (<8 x i8> bitcast (<1 x i64> <i64 7> to <8 x i8>), i32 1), i8 extractelement (<8 x i8> bitcast (<1 x i64> <i64 7> to <8 x i8>), i32 2), i8 extractelement (<8 x i8> bitcast (<1 x i64> <i64 7> to <8 x i8>), i32 3), i8 extractelement (<8 x i8> bitcast (<1 x i64> <i64 7> to <8 x i8>), i32 4), i8 extractelement (<8 x i8> bitcast (<1 x i64> <i64 7> to <8 x i8>), i32 5), i8 extractelement (<8 x i8> bitcast (<1 x i64> <i64 7> to <8 x i8>), i32 6), i8 extractelement (<8 x i8> bitcast (<1 x i64> <i64 7> to <8 x i8>), i32 7), i8 undef, i8 undef>, <16 x i8> [[A_SROA_32_SROA_0_5_VECBLEND]]
-; CHECK-NEXT:    [[A_SROA_32_SROA_0_7_VECBLEND:%.*]] = select <16 x i1> <i1 false, i1 false, i1 false, i1 false, i1 false, i1 false, i1 false, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 false>, <16 x i8> <i8 undef, i8 undef, i8 undef, i8 undef, i8 undef, i8 undef, i8 undef, i8 extractelement (<8 x i8> bitcast (<1 x i64> <i64 8> to <8 x i8>), i32 0), i8 extractelement (<8 x i8> bitcast (<1 x i64> <i64 8> to <8 x i8>), i32 1), i8 extractelement (<8 x i8> bitcast (<1 x i64> <i64 8> to <8 x i8>), i32 2), i8 extractelement (<8 x i8> bitcast (<1 x i64> <i64 8> to <8 x i8>), i32 3), i8 extractelement (<8 x i8> bitcast (<1 x i64> <i64 8> to <8 x i8>), i32 4), i8 extractelement (<8 x i8> bitcast (<1 x i64> <i64 8> to <8 x i8>), i32 5), i8 extractelement (<8 x i8> bitcast (<1 x i64> <i64 8> to <8 x i8>), i32 6), i8 extractelement (<8 x i8> bitcast (<1 x i64> <i64 8> to <8 x i8>), i32 7), i8 undef>, <16 x i8> [[A_SROA_32_SROA_0_6_VECBLEND]]
-; CHECK-NEXT:    [[A_SROA_32_SROA_0_8_VECBLEND:%.*]] = select <16 x i1> <i1 false, i1 false, i1 false, i1 false, i1 false, i1 false, i1 false, i1 false, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true>, <16 x i8> <i8 undef, i8 undef, i8 undef, i8 undef, i8 undef, i8 undef, i8 undef, i8 undef, i8 extractelement (<8 x i8> bitcast (<1 x i64> <i64 9> to <8 x i8>), i32 0), i8 extractelement (<8 x i8> bitcast (<1 x i64> <i64 9> to <8 x i8>), i32 1), i8 extractelement (<8 x i8> bitcast (<1 x i64> <i64 9> to <8 x i8>), i32 2), i8 extractelement (<8 x i8> bitcast (<1 x i64> <i64 9> to <8 x i8>), i32 3), i8 extractelement (<8 x i8> bitcast (<1 x i64> <i64 9> to <8 x i8>), i32 4), i8 extractelement (<8 x i8> bitcast (<1 x i64> <i64 9> to <8 x i8>), i32 5), i8 extractelement (<8 x i8> bitcast (<1 x i64> <i64 9> to <8 x i8>), i32 6), i8 extractelement (<8 x i8> bitcast (<1 x i64> <i64 9> to <8 x i8>), i32 7)>, <16 x i8> [[A_SROA_32_SROA_0_7_VECBLEND]]
-; CHECK-NEXT:    [[A_SROA_16_SROA_0_0_VEC_INSERT:%.*]] = insertelement <7 x i8> [[A_SROA_16_SROA_0_0_COPYLOAD]], i8 1, i32 0
-; CHECK-NEXT:    [[A_SROA_16_SROA_0_0_VECBLEND18:%.*]] = select <7 x i1> <i1 true, i1 true, i1 false, i1 false, i1 false, i1 false, i1 false>, <7 x i8> <i8 extractelement (<2 x i8> bitcast (<1 x i16> <i16 1> to <2 x i8>), i32 0), i8 extractelement (<2 x i8> bitcast (<1 x i16> <i16 1> to <2 x i8>), i32 1), i8 undef, i8 undef, i8 undef, i8 undef, i8 undef>, <7 x i8> [[A_SROA_16_SROA_0_0_VEC_INSERT]]
-; CHECK-NEXT:    [[A_SROA_16_SROA_0_0_VECBLEND:%.*]] = select <7 x i1> <i1 true, i1 true, i1 true, i1 true, i1 false, i1 false, i1 false>, <7 x i8> <i8 extractelement (<4 x i8> bitcast (<1 x i32> <i32 1> to <4 x i8>), i32 0), i8 extractelement (<4 x i8> bitcast (<1 x i32> <i32 1> to <4 x i8>), i32 1), i8 extractelement (<4 x i8> bitcast (<1 x i32> <i32 1> to <4 x i8>), i32 2), i8 extractelement (<4 x i8> bitcast (<1 x i32> <i32 1> to <4 x i8>), i32 3), i8 undef, i8 undef, i8 undef>, <7 x i8> [[A_SROA_16_SROA_0_0_VECBLEND18]]
-; CHECK-NEXT:    [[A_SROA_16_SROA_0_1_VECBLEND:%.*]] = select <7 x i1> <i1 false, i1 true, i1 true, i1 true, i1 true, i1 false, i1 false>, <7 x i8> <i8 undef, i8 extractelement (<4 x i8> bitcast (<1 x i32> <i32 2> to <4 x i8>), i32 0), i8 extractelement (<4 x i8> bitcast (<1 x i32> <i32 2> to <4 x i8>), i32 1), i8 extractelement (<4 x i8> bitcast (<1 x i32> <i32 2> to <4 x i8>), i32 2), i8 extractelement (<4 x i8> bitcast (<1 x i32> <i32 2> to <4 x i8>), i32 3), i8 undef, i8 undef>, <7 x i8> [[A_SROA_16_SROA_0_0_VECBLEND]]
-; CHECK-NEXT:    [[A_SROA_16_SROA_0_2_VECBLEND:%.*]] = select <7 x i1> <i1 false, i1 false, i1 true, i1 true, i1 true, i1 true, i1 false>, <7 x i8> <i8 undef, i8 undef, i8 extractelement (<4 x i8> bitcast (<1 x i32> <i32 3> to <4 x i8>), i32 0), i8 extractelement (<4 x i8> bitcast (<1 x i32> <i32 3> to <4 x i8>), i32 1), i8 extractelement (<4 x i8> bitcast (<1 x i32> <i32 3> to <4 x i8>), i32 2), i8 extractelement (<4 x i8> bitcast (<1 x i32> <i32 3> to <4 x i8>), i32 3), i8 undef>, <7 x i8> [[A_SROA_16_SROA_0_1_VECBLEND]]
-; CHECK-NEXT:    [[A_SROA_16_SROA_0_3_VECBLEND:%.*]] = select <7 x i1> <i1 false, i1 false, i1 false, i1 true, i1 true, i1 true, i1 true>, <7 x i8> <i8 undef, i8 undef, i8 undef, i8 extractelement (<4 x i8> bitcast (<1 x i32> <i32 4> to <4 x i8>), i32 0), i8 extractelement (<4 x i8> bitcast (<1 x i32> <i32 4> to <4 x i8>), i32 1), i8 extractelement (<4 x i8> bitcast (<1 x i32> <i32 4> to <4 x i8>), i32 2), i8 extractelement (<4 x i8> bitcast (<1 x i32> <i32 4> to <4 x i8>), i32 3)>, <7 x i8> [[A_SROA_16_SROA_0_2_VECBLEND]]
-; CHECK-NEXT:    [[A_SROA_235_SROA_0_0_VECBLEND:%.*]] = select <7 x i1> <i1 true, i1 true, i1 true, i1 true, i1 false, i1 false, i1 false>, <7 x i8> <i8 extractelement (<4 x i8> bitcast (<1 x i32> <i32 1> to <4 x i8>), i32 0), i8 extractelement (<4 x i8> bitcast (<1 x i32> <i32 1> to <4 x i8>), i32 1), i8 extractelement (<4 x i8> bitcast (<1 x i32> <i32 1> to <4 x i8>), i32 2), i8 extractelement (<4 x i8> bitcast (<1 x i32> <i32 1> to <4 x i8>), i32 3), i8 undef, i8 undef, i8 undef>, <7 x i8> [[A_SROA_235_SROA_0_0_COPYLOAD]]
-; CHECK-NEXT:    [[A_SROA_235_SROA_0_1_VEC_INSERT:%.*]] = insertelement <7 x i8> [[A_SROA_235_SROA_0_0_VECBLEND]], i8 1, i32 1
-; CHECK-NEXT:    [[A_SROA_235_SROA_0_1_VECBLEND13:%.*]] = select <7 x i1> <i1 false, i1 true, i1 true, i1 false, i1 false, i1 false, i1 false>, <7 x i8> <i8 undef, i8 extractelement (<2 x i8> bitcast (<1 x i16> <i16 1> to <2 x i8>), i32 0), i8 extractelement (<2 x i8> bitcast (<1 x i16> <i16 1> to <2 x i8>), i32 1), i8 undef, i8 undef, i8 undef, i8 undef>, <7 x i8> [[A_SROA_235_SROA_0_1_VEC_INSERT]]
-; CHECK-NEXT:    [[A_SROA_235_SROA_0_1_VECBLEND:%.*]] = select <7 x i1> <i1 false, i1 true, i1 true, i1 true, i1 true, i1 false, i1 false>, <7 x i8> <i8 undef, i8 extractelement (<4 x i8> bitcast (<1 x i32> <i32 1> to <4 x i8>), i32 0), i8 extractelement (<4 x i8> bitcast (<1 x i32> <i32 1> to <4 x i8>), i32 1), i8 extractelement (<4 x i8> bitcast (<1 x i32> <i32 1> to <4 x i8>), i32 2), i8 extractelement (<4 x i8> bitcast (<1 x i32> <i32 1> to <4 x i8>), i32 3), i8 undef, i8 undef>, <7 x i8> [[A_SROA_235_SROA_0_1_VECBLEND13]]
-; CHECK-NEXT:    [[A_SROA_235_SROA_0_2_VECBLEND:%.*]] = select <7 x i1> <i1 false, i1 false, i1 true, i1 true, i1 true, i1 true, i1 false>, <7 x i8> <i8 undef, i8 undef, i8 extractelement (<4 x i8> bitcast (<1 x i32> <i32 3> to <4 x i8>), i32 0), i8 extractelement (<4 x i8> bitcast (<1 x i32> <i32 3> to <4 x i8>), i32 1), i8 extractelement (<4 x i8> bitcast (<1 x i32> <i32 3> to <4 x i8>), i32 2), i8 extractelement (<4 x i8> bitcast (<1 x i32> <i32 3> to <4 x i8>), i32 3), i8 undef>, <7 x i8> [[A_SROA_235_SROA_0_1_VECBLEND]]
-; CHECK-NEXT:    [[A_SROA_235_SROA_0_3_VECBLEND:%.*]] = select <7 x i1> <i1 false, i1 false, i1 false, i1 true, i1 true, i1 true, i1 true>, <7 x i8> <i8 undef, i8 undef, i8 undef, i8 extractelement (<4 x i8> bitcast (<1 x i32> <i32 4> to <4 x i8>), i32 0), i8 extractelement (<4 x i8> bitcast (<1 x i32> <i32 4> to <4 x i8>), i32 1), i8 extractelement (<4 x i8> bitcast (<1 x i32> <i32 4> to <4 x i8>), i32 2), i8 extractelement (<4 x i8> bitcast (<1 x i32> <i32 4> to <4 x i8>), i32 3)>, <7 x i8> [[A_SROA_235_SROA_0_2_VECBLEND]]
-; CHECK-NEXT:    [[A_SROA_15_SROA_0_39_COPYLOAD:%.*]] = load <3 x i8>, ptr [[SRC]], align 1, !tbaa [[TBAA3:![0-9]+]]
-; CHECK-NEXT:    [[A_SROA_15_SROA_0_39_VEC_EXPAND:%.*]] = shufflevector <3 x i8> [[A_SROA_15_SROA_0_39_COPYLOAD]], <3 x i8> poison, <42 x i32> <i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 0, i32 1, i32 2>
-; CHECK-NEXT:    [[A_SROA_15_SROA_0_39_VECBLEND:%.*]] = select <42 x i1> <i1 false, i1 false, i1 false, i1 false, i1 false, i1 false, i1 false, i1 false, i1 false, i1 false, i1 false, i1 false, i1 false, i1 false, i1 false, i1 false, i1 false, i1 false, i1 false, i1 false, i1 false, i1 false, i1 false, i1 false, i1 false, i1 false, i1 false, i1 false, i1 false, i1 false, i1 false, i1 false, i1 false, i1 false, i1 false, i1 false, i1 false, i1 false, i1 false, i1 true, i1 true, i1 true>, <42 x i8> [[A_SROA_15_SROA_0_39_VEC_EXPAND]], <42 x i8> [[A_SROA_15_SROA_0_0_COPYLOAD]]
+; CHECK-NEXT:    store i8 1, ptr [[A_SROA_32]], align 1, !tbaa [[TBAA3:![0-9]+]]
+; CHECK-NEXT:    store i16 1, ptr [[A_SROA_32]], align 1, !tbaa [[TBAA5:![0-9]+]]
+; CHECK-NEXT:    store i32 1, ptr [[A_SROA_32]], align 1, !tbaa [[TBAA7:![0-9]+]]
+; CHECK-NEXT:    store i64 1, ptr [[A_SROA_32]], align 1, !tbaa [[TBAA9:![0-9]+]]
+; CHECK-NEXT:    [[A_SROA_32_1_OVERLAP_2_I8_SROA_IDX:%.*]] = getelementptr inbounds i8, ptr [[A_SROA_32]], i64 1
+; CHECK-NEXT:    store i64 2, ptr [[A_SROA_32_1_OVERLAP_2_I8_SROA_IDX]], align 1, !tbaa [[TBAA11:![0-9]+]]
+; CHECK-NEXT:    [[A_SROA_32_2_OVERLAP_3_I8_SROA_IDX:%.*]] = getelementptr inbounds i8, ptr [[A_SROA_32]], i64 2
+; CHECK-NEXT:    store i64 3, ptr [[A_SROA_32_2_OVERLAP_3_I8_SROA_IDX]], align 1, !tbaa [[TBAA13:![0-9]+]]
+; CHECK-NEXT:    [[A_SROA_32_3_OVERLAP_4_I8_SROA_IDX:%.*]] = getelementptr inbounds i8, ptr [[A_SROA_32]], i64 3
+; CHECK-NEXT:    store i64 4, ptr [[A_SROA_32_3_OVERLAP_4_I8_SROA_IDX]], align 1, !tbaa [[TBAA15:![0-9]+]]
+; CHECK-NEXT:    [[A_SROA_32_4_OVERLAP_5_I8_SROA_IDX:%.*]] = getelementptr inbounds i8, ptr [[A_SROA_32]], i64 4
+; CHECK-NEXT:    store i64 5, ptr [[A_SROA_32_4_OVERLAP_5_I8_SROA_IDX]], align 1, !tbaa [[TBAA17:![0-9]+]]
+; CHECK-NEXT:    [[A_SROA_32_5_OVERLAP_6_I8_SROA_IDX:%.*]] = getelementptr inbounds i8, ptr [[A_SROA_32]], i64 5
+; CHECK-NEXT:    store i64 6, ptr [[A_SROA_32_5_OVERLAP_6_I8_SROA_IDX]], align 1, !tbaa [[TBAA19:![0-9]+]]
+; CHECK-NEXT:    [[A_SROA_32_6_OVERLAP_7_I8_SROA_IDX:%.*]] = getelementptr inbounds i8, ptr [[A_SROA_32]], i64 6
+; CHECK-NEXT:    store i64 7, ptr [[A_SROA_32_6_OVERLAP_7_I8_SROA_IDX]], align 1, !tbaa [[TBAA21:![0-9]+]]
+; CHECK-NEXT:    [[A_SROA_32_7_OVERLAP_8_I8_SROA_IDX:%.*]] = getelementptr inbounds i8, ptr [[A_SROA_32]], i64 7
+; CHECK-NEXT:    store i64 8, ptr [[A_SROA_32_7_OVERLAP_8_I8_SROA_IDX]], align 1, !tbaa [[TBAA23:![0-9]+]]
+; CHECK-NEXT:    [[A_SROA_32_8_OVERLAP_9_I8_SROA_IDX:%.*]] = getelementptr inbounds i8, ptr [[A_SROA_32]], i64 8
+; CHECK-NEXT:    store i64 9, ptr [[A_SROA_32_8_OVERLAP_9_I8_SROA_IDX]], align 1, !tbaa [[TBAA25:![0-9]+]]
+; CHECK-NEXT:    store i8 1, ptr [[A_SROA_16]], align 1, !tbaa [[TBAA27:![0-9]+]]
+; CHECK-NEXT:    store i16 1, ptr [[A_SROA_16]], align 1, !tbaa [[TBAA29:![0-9]+]]
+; CHECK-NEXT:    store i32 1, ptr [[A_SROA_16]], align 1, !tbaa [[TBAA31:![0-9]+]]
+; CHECK-NEXT:    [[A_SROA_16_1_OVERLAP2_1_1_I8_SROA_IDX:%.*]] = getelementptr inbounds i8, ptr [[A_SROA_16]], i64 1
+; CHECK-NEXT:    store i32 2, ptr [[A_SROA_16_1_OVERLAP2_1_1_I8_SROA_IDX]], align 1, !tbaa [[TBAA33:![0-9]+]]
+; CHECK-NEXT:    [[A_SROA_16_2_OVERLAP2_1_2_I8_SROA_IDX:%.*]] = getelementptr inbounds i8, ptr [[A_SROA_16]], i64 2
+; CHECK-NEXT:    store i32 3, ptr [[A_SROA_16_2_OVERLAP2_1_2_I8_SROA_IDX]], align 1, !tbaa [[TBAA35:![0-9]+]]
+; CHECK-NEXT:    [[A_SROA_16_3_OVERLAP2_1_3_I8_SROA_IDX:%.*]] = getelementptr inbounds i8, ptr [[A_SROA_16]], i64 3
+; CHECK-NEXT:    store i32 4, ptr [[A_SROA_16_3_OVERLAP2_1_3_I8_SROA_IDX]], align 1, !tbaa [[TBAA37:![0-9]+]]
+; CHECK-NEXT:    store i32 1, ptr [[A_SROA_235]], align 1, !tbaa [[TBAA39:![0-9]+]]
+; CHECK-NEXT:    [[A_SROA_235_1_OVERLAP2_2_1_I8_SROA_IDX11:%.*]] = getelementptr inbounds i8, ptr [[A_SROA_235]], i64 1
+; CHECK-NEXT:    store i8 1, ptr [[A_SROA_235_1_OVERLAP2_2_1_I8_SROA_IDX11]], align 1, !tbaa [[TBAA41:![0-9]+]]
+; CHECK-NEXT:    [[A_SROA_235_1_OVERLAP2_2_1_I8_SROA_IDX10:%.*]] = getelementptr inbounds i8, ptr [[A_SROA_235]], i64 1
+; CHECK-NEXT:    store i16 1, ptr [[A_SROA_235_1_OVERLAP2_2_1_I8_SROA_IDX10]], align 1, !tbaa [[TBAA43:![0-9]+]]
+; CHECK-NEXT:    [[A_SROA_235_1_OVERLAP2_2_1_I8_SROA_IDX:%.*]] = getelementptr inbounds i8, ptr [[A_SROA_235]], i64 1
+; CHECK-NEXT:    store i32 1, ptr [[A_SROA_235_1_OVERLAP2_2_1_I8_SROA_IDX]], align 1, !tbaa [[TBAA45:![0-9]+]]
+; CHECK-NEXT:    [[A_SROA_235_2_OVERLAP2_2_2_I8_SROA_IDX:%.*]] = getelementptr inbounds i8, ptr [[A_SROA_235]], i64 2
+; CHECK-NEXT:    store i32 3, ptr [[A_SROA_235_2_OVERLAP2_2_2_I8_SROA_IDX]], align 1, !tbaa [[TBAA47:![0-9]+]]
+; CHECK-NEXT:    [[A_SROA_235_3_OVERLAP2_2_3_I8_SROA_IDX:%.*]] = getelementptr inbounds i8, ptr [[A_SROA_235]], i64 3
+; CHECK-NEXT:    store i32 4, ptr [[A_SROA_235_3_OVERLAP2_2_3_I8_SROA_IDX]], align 1, !tbaa [[TBAA49:![0-9]+]]
+; CHECK-NEXT:    [[A_SROA_15_197_OVERLAP2_PREFIX_SROA_IDX:%.*]] = getelementptr inbounds i8, ptr [[A_SROA_15]], i64 39
+; CHECK-NEXT:    call void @llvm.memcpy.p0.p0.i32(ptr align 1 [[A_SROA_15_197_OVERLAP2_PREFIX_SROA_IDX]], ptr align 1 [[SRC]], i32 3, i1 false), !tbaa [[TBAA51:![0-9]+]]
 ; CHECK-NEXT:    [[A_SROA_16_197_SRC_SROA_IDX:%.*]] = getelementptr inbounds i8, ptr [[SRC]], i64 3
-; CHECK-NEXT:    [[A_SROA_16_SROA_0_0_COPYLOAD21:%.*]] = load <5 x i8>, ptr [[A_SROA_16_197_SRC_SROA_IDX]], align 1, !tbaa [[TBAA3]]
-; CHECK-NEXT:    [[A_SROA_16_SROA_0_0_VEC_EXPAND:%.*]] = shufflevector <5 x i8> [[A_SROA_16_SROA_0_0_COPYLOAD21]], <5 x i8> poison, <7 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 undef, i32 undef>
-; CHECK-NEXT:    [[A_SROA_16_SROA_0_0_VECBLEND22:%.*]] = select <7 x i1> <i1 true, i1 true, i1 true, i1 true, i1 true, i1 false, i1 false>, <7 x i8> [[A_SROA_16_SROA_0_0_VEC_EXPAND]], <7 x i8> [[A_SROA_16_SROA_0_3_VECBLEND]]
-; CHECK-NEXT:    [[A_SROA_16_SROA_0_2_VECBLEND23:%.*]] = select <7 x i1> <i1 false, i1 false, i1 true, i1 true, i1 true, i1 true, i1 true>, <7 x i8> <i8 undef, i8 undef, i8 42, i8 42, i8 42, i8 42, i8 42>, <7 x i8> [[A_SROA_16_SROA_0_0_VECBLEND22]]
-; CHECK-NEXT:    [[A_SROA_235_SROA_0_0_VECBLEND11:%.*]] = select <7 x i1> <i1 true, i1 true, i1 false, i1 false, i1 false, i1 false, i1 false>, <7 x i8> <i8 42, i8 42, i8 undef, i8 undef, i8 undef, i8 undef, i8 undef>, <7 x i8> [[A_SROA_235_SROA_0_3_VECBLEND]]
-; CHECK-NEXT:    [[A_SROA_235_SROA_0_1_COPYLOAD:%.*]] = load <5 x i8>, ptr [[SRC]], align 1, !tbaa [[TBAA5:![0-9]+]]
-; CHECK-NEXT:    [[A_SROA_235_SROA_0_1_VEC_EXPAND:%.*]] = shufflevector <5 x i8> [[A_SROA_235_SROA_0_1_COPYLOAD]], <5 x i8> poison, <7 x i32> <i32 undef, i32 0, i32 1, i32 2, i32 3, i32 4, i32 undef>
-; CHECK-NEXT:    [[A_SROA_235_SROA_0_1_VECBLEND15:%.*]] = select <7 x i1> <i1 false, i1 true, i1 true, i1 true, i1 true, i1 true, i1 false>, <7 x i8> [[A_SROA_235_SROA_0_1_VEC_EXPAND]], <7 x i8> [[A_SROA_235_SROA_0_0_VECBLEND11]]
-; CHECK-NEXT:    [[A_SROA_235_SROA_0_2_COPYLOAD:%.*]] = load <5 x i8>, ptr [[SRC]], align 1, !tbaa [[TBAA7:![0-9]+]]
-; CHECK-NEXT:    [[A_SROA_235_SROA_0_2_VEC_EXPAND:%.*]] = shufflevector <5 x i8> [[A_SROA_235_SROA_0_2_COPYLOAD]], <5 x i8> poison, <7 x i32> <i32 undef, i32 undef, i32 0, i32 1, i32 2, i32 3, i32 4>
-; CHECK-NEXT:    [[A_SROA_235_SROA_0_2_VECBLEND16:%.*]] = select <7 x i1> <i1 false, i1 false, i1 true, i1 true, i1 true, i1 true, i1 true>, <7 x i8> [[A_SROA_235_SROA_0_2_VEC_EXPAND]], <7 x i8> [[A_SROA_235_SROA_0_1_VECBLEND15]]
+; CHECK-NEXT:    call void @llvm.memcpy.p0.p0.i32(ptr align 1 [[A_SROA_16]], ptr align 1 [[A_SROA_16_197_SRC_SROA_IDX]], i32 5, i1 false), !tbaa [[TBAA51]]
+; CHECK-NEXT:    [[A_SROA_16_2_OVERLAP2_1_2_I8_SROA_IDX12:%.*]] = getelementptr inbounds i8, ptr [[A_SROA_16]], i64 2
+; CHECK-NEXT:    call void @llvm.memset.p0.i32(ptr align 1 [[A_SROA_16_2_OVERLAP2_1_2_I8_SROA_IDX12]], i8 42, i32 5, i1 false), !tbaa [[TBAA53:![0-9]+]]
+; CHECK-NEXT:    call void @llvm.memset.p0.i32(ptr align 1 [[A_SROA_235]], i8 42, i32 2, i1 false), !tbaa [[TBAA53]]
+; CHECK-NEXT:    [[A_SROA_235_209_OVERLAP2_2_1_I8_SROA_IDX8:%.*]] = getelementptr inbounds i8, ptr [[A_SROA_235]], i64 1
+; CHECK-NEXT:    call void @llvm.memcpy.p0.p0.i32(ptr align 1 [[A_SROA_235_209_OVERLAP2_2_1_I8_SROA_IDX8]], ptr align 1 [[SRC]], i32 5, i1 false), !tbaa [[TBAA55:![0-9]+]]
+; CHECK-NEXT:    [[A_SROA_235_210_OVERLAP2_2_2_I8_SROA_IDX9:%.*]] = getelementptr inbounds i8, ptr [[A_SROA_235]], i64 2
+; CHECK-NEXT:    call void @llvm.memcpy.p0.p0.i32(ptr align 1 [[A_SROA_235_210_OVERLAP2_2_2_I8_SROA_IDX9]], ptr align 1 [[SRC]], i32 5, i1 false), !tbaa [[TBAA57:![0-9]+]]
 ; CHECK-NEXT:    [[A_SROA_31_210_SRC_SROA_IDX:%.*]] = getelementptr inbounds i8, ptr [[SRC]], i64 5
-; CHECK-NEXT:    call void @llvm.memcpy.p0.p0.i32(ptr align 1 [[A_SROA_31]], ptr align 1 [[A_SROA_31_210_SRC_SROA_IDX]], i32 3, i1 false), !tbaa [[TBAA7]]
-; CHECK-NEXT:    store <42 x i8> [[A_SROA_0_SROA_0_0_COPYLOAD]], ptr [[DST:%.*]], align 1, !tbaa [[TBAA9:![0-9]+]]
+; CHECK-NEXT:    call void @llvm.memcpy.p0.p0.i32(ptr align 1 [[A_SROA_31]], ptr align 1 [[A_SROA_31_210_SRC_SROA_IDX]], i32 3, i1 false), !tbaa [[TBAA57]]
+; CHECK-NEXT:    call void @llvm.memcpy.p0.p0.i32(ptr align 1 [[DST:%.*]], ptr align 1 [[A_SROA_0]], i32 42, i1 false), !tbaa [[TBAA59:![0-9]+]]
 ; CHECK-NEXT:    [[A_SROA_2_0_DST_SROA_IDX:%.*]] = getelementptr inbounds i8, ptr [[DST]], i64 42
-; CHECK-NEXT:    store i8 0, ptr [[A_SROA_2_0_DST_SROA_IDX]], align 1, !tbaa [[TBAA9]]
+; CHECK-NEXT:    store i8 0, ptr [[A_SROA_2_0_DST_SROA_IDX]], align 1, !tbaa [[TBAA59]]
 ; CHECK-NEXT:    [[A_SROA_3_0_DST_SROA_IDX:%.*]] = getelementptr inbounds i8, ptr [[DST]], i64 43
-; CHECK-NEXT:    call void @llvm.memcpy.p0.p0.i32(ptr align 1 [[A_SROA_3_0_DST_SROA_IDX]], ptr align 1 [[A_SROA_3]], i32 99, i1 false), !tbaa [[TBAA9]]
+; CHECK-NEXT:    call void @llvm.memcpy.p0.p0.i32(ptr align 1 [[A_SROA_3_0_DST_SROA_IDX]], ptr align 1 [[A_SROA_3]], i32 99, i1 false), !tbaa [[TBAA59]]
 ; CHECK-NEXT:    [[A_SROA_32_0_DST_SROA_IDX:%.*]] = getelementptr inbounds i8, ptr [[DST]], i64 142
-; CHECK-NEXT:    store <16 x i8> [[A_SROA_32_SROA_0_8_VECBLEND]], ptr [[A_SROA_32_0_DST_SROA_IDX]], align 1, !tbaa [[TBAA9]]
+; CHECK-NEXT:    call void @llvm.memcpy.p0.p0.i32(ptr align 1 [[A_SROA_32_0_DST_SROA_IDX]], ptr align 1 [[A_SROA_32]], i32 16, i1 false), !tbaa [[TBAA59]]
 ; CHECK-NEXT:    [[A_SROA_15_0_DST_SROA_IDX:%.*]] = getelementptr inbounds i8, ptr [[DST]], i64 158
-; CHECK-NEXT:    store <42 x i8> [[A_SROA_15_SROA_0_39_VECBLEND]], ptr [[A_SROA_15_0_DST_SROA_IDX]], align 1, !tbaa [[TBAA9]]
+; CHECK-NEXT:    call void @llvm.memcpy.p0.p0.i32(ptr align 1 [[A_SROA_15_0_DST_SROA_IDX]], ptr align 1 [[A_SROA_15]], i32 42, i1 false), !tbaa [[TBAA59]]
 ; CHECK-NEXT:    [[A_SROA_16_0_DST_SROA_IDX:%.*]] = getelementptr inbounds i8, ptr [[DST]], i64 200
-; CHECK-NEXT:    store <7 x i8> [[A_SROA_16_SROA_0_2_VECBLEND23]], ptr [[A_SROA_16_0_DST_SROA_IDX]], align 1, !tbaa [[TBAA9]]
+; CHECK-NEXT:    call void @llvm.memcpy.p0.p0.i32(ptr align 1 [[A_SROA_16_0_DST_SROA_IDX]], ptr align 1 [[A_SROA_16]], i32 7, i1 false), !tbaa [[TBAA59]]
 ; CHECK-NEXT:    [[A_SROA_23_0_DST_SROA_IDX:%.*]] = getelementptr inbounds i8, ptr [[DST]], i64 207
-; CHECK-NEXT:    store i8 42, ptr [[A_SROA_23_0_DST_SROA_IDX]], align 1, !tbaa [[TBAA9]]
+; CHECK-NEXT:    store i8 42, ptr [[A_SROA_23_0_DST_SROA_IDX]], align 1, !tbaa [[TBAA59]]
 ; CHECK-NEXT:    [[A_SROA_235_0_DST_SROA_IDX:%.*]] = getelementptr inbounds i8, ptr [[DST]], i64 208
-; CHECK-NEXT:    store <7 x i8> [[A_SROA_235_SROA_0_2_VECBLEND16]], ptr [[A_SROA_235_0_DST_SROA_IDX]], align 1, !tbaa [[TBAA9]]
+; CHECK-NEXT:    call void @llvm.memcpy.p0.p0.i32(ptr align 1 [[A_SROA_235_0_DST_SROA_IDX]], ptr align 1 [[A_SROA_235]], i32 7, i1 false), !tbaa [[TBAA59]]
 ; CHECK-NEXT:    [[A_SROA_31_0_DST_SROA_IDX:%.*]] = getelementptr inbounds i8, ptr [[DST]], i64 215
-; CHECK-NEXT:    call void @llvm.memcpy.p0.p0.i32(ptr align 1 [[A_SROA_31_0_DST_SROA_IDX]], ptr align 1 [[A_SROA_31]], i32 85, i1 false), !tbaa [[TBAA9]]
+; CHECK-NEXT:    call void @llvm.memcpy.p0.p0.i32(ptr align 1 [[A_SROA_31_0_DST_SROA_IDX]], ptr align 1 [[A_SROA_31]], i32 85, i1 false), !tbaa [[TBAA59]]
 ; CHECK-NEXT:    ret void
 ;
 entry:
@@ -292,29 +309,60 @@ entry:
 define void @test4(ptr %dst, ptr %src) {
 ; CHECK-LABEL: @test4(
 ; CHECK-NEXT:  entry:
-; CHECK-NEXT:    [[A_SROA_0_SROA_0_0_COPYLOAD:%.*]] = load <20 x i8>, ptr [[SRC:%.*]], align 1, !tbaa [[TBAA0]]
+; CHECK-NEXT:    [[A_SROA_0:%.*]] = alloca [20 x i8], align 1
+; CHECK-NEXT:    [[A_SROA_2_SROA_4:%.*]] = alloca [7 x i8], align 1
+; CHECK-NEXT:    [[A_SROA_3:%.*]] = alloca [10 x i8], align 1
+; CHECK-NEXT:    [[A_SROA_31_SROA_5:%.*]] = alloca [7 x i8], align 1
+; CHECK-NEXT:    [[A_SROA_6_SROA_4:%.*]] = alloca [7 x i8], align 1
+; CHECK-NEXT:    [[A_SROA_7:%.*]] = alloca [40 x i8], align 1
+; CHECK-NEXT:    call void @llvm.memcpy.p0.p0.i32(ptr align 1 [[A_SROA_0]], ptr align 1 [[SRC:%.*]], i32 20, i1 false), !tbaa [[TBAA0]]
 ; CHECK-NEXT:    [[A_SROA_2_0_SRC_SROA_IDX:%.*]] = getelementptr inbounds i8, ptr [[SRC]], i64 20
-; CHECK-NEXT:    [[A_SROA_2_SROA_0_0_COPYLOAD:%.*]] = load <10 x i8>, ptr [[A_SROA_2_0_SRC_SROA_IDX]], align 1, !tbaa [[TBAA0]]
+; CHECK-NEXT:    [[A_SROA_2_SROA_0_0_COPYLOAD:%.*]] = load i16, ptr [[A_SROA_2_0_SRC_SROA_IDX]], align 1, !tbaa [[TBAA0]]
+; CHECK-NEXT:    [[A_SROA_2_SROA_3_0_A_SROA_2_0_SRC_SROA_IDX_SROA_IDX:%.*]] = getelementptr inbounds i8, ptr [[A_SROA_2_0_SRC_SROA_IDX]], i64 2
+; CHECK-NEXT:    [[A_SROA_2_SROA_3_0_COPYLOAD:%.*]] = load i8, ptr [[A_SROA_2_SROA_3_0_A_SROA_2_0_SRC_SROA_IDX_SROA_IDX]], align 1, !tbaa [[TBAA0]]
+; CHECK-NEXT:    [[A_SROA_2_SROA_4_0_A_SROA_2_0_SRC_SROA_IDX_SROA_IDX:%.*]] = getelementptr inbounds i8, ptr [[A_SROA_2_0_SRC_SROA_IDX]], i64 3
+; CHECK-NEXT:    call void @llvm.memcpy.p0.p0.i32(ptr align 1 [[A_SROA_2_SROA_4]], ptr align 1 [[A_SROA_2_SROA_4_0_A_SROA_2_0_SRC_SROA_IDX_SROA_IDX]], i32 7, i1 false), !tbaa [[TBAA0]]
 ; CHECK-NEXT:    [[A_SROA_3_0_SRC_SROA_IDX:%.*]] = getelementptr inbounds i8, ptr [[SRC]], i64 30
-; CHECK-NEXT:    [[A_SROA_3_SROA_0_0_COPYLOAD:%.*]] = load <10 x i8>, ptr [[A_SROA_3_0_SRC_SROA_IDX]], align 1, !tbaa [[TBAA0]]
+; CHECK-NEXT:    call void @llvm.memcpy.p0.p0.i32(ptr align 1 [[A_SROA_3]], ptr align 1 [[A_SROA_3_0_SRC_SROA_IDX]], i32 10, i1 false), !tbaa [[TBAA0]]
 ; CHECK-NEXT:    [[A_SROA_31_0_SRC_SROA_IDX:%.*]] = getelementptr inbounds i8, ptr [[SRC]], i64 40
-; CHECK-NEXT:    [[A_SROA_31_SROA_0_0_COPYLOAD:%.*]] = load <10 x i8>, ptr [[A_SROA_31_0_SRC_SROA_IDX]], align 1, !tbaa [[TBAA0]]
+; CHECK-NEXT:    [[A_SROA_31_SROA_0_0_COPYLOAD:%.*]] = load i16, ptr [[A_SROA_31_0_SRC_SROA_IDX]], align 1, !tbaa [[TBAA0]]
+; CHECK-NEXT:    [[A_SROA_31_SROA_4_0_A_SROA_31_0_SRC_SROA_IDX_SROA_IDX:%.*]] = getelementptr inbounds i8, ptr [[A_SROA_31_0_SRC_SROA_IDX]], i64 2
+; CHECK-NEXT:    [[A_SROA_31_SROA_4_0_COPYLOAD:%.*]] = load i8, ptr [[A_SROA_31_SROA_4_0_A_SROA_31_0_SRC_SROA_IDX_SROA_IDX]], align 1, !tbaa [[TBAA0]]
+; CHECK-NEXT:    [[A_SROA_31_SROA_5_0_A_SROA_31_0_SRC_SROA_IDX_SROA_IDX:%.*]] = getelementptr inbounds i8, ptr [[A_SROA_31_0_SRC_SROA_IDX]], i64 3
+; CHECK-NEXT:    call void @llvm.memcpy.p0.p0.i32(ptr align 1 [[A_SROA_31_SROA_5]], ptr align 1 [[A_SROA_31_SROA_5_0_A_SROA_31_0_SRC_SROA_IDX_SROA_IDX]], i32 7, i1 false), !tbaa [[TBAA0]]
 ; CHECK-NEXT:    [[A_SROA_6_0_SRC_SROA_IDX:%.*]] = getelementptr inbounds i8, ptr [[SRC]], i64 50
-; CHECK-NEXT:    [[A_SROA_6_SROA_0_0_COPYLOAD:%.*]] = load <10 x i8>, ptr [[A_SROA_6_0_SRC_SROA_IDX]], align 1, !tbaa [[TBAA0]]
+; CHECK-NEXT:    [[A_SROA_6_SROA_0_0_COPYLOAD:%.*]] = load i16, ptr [[A_SROA_6_0_SRC_SROA_IDX]], align 1, !tbaa [[TBAA0]]
+; CHECK-NEXT:    [[A_SROA_6_SROA_3_0_A_SROA_6_0_SRC_SROA_IDX_SROA_IDX:%.*]] = getelementptr inbounds i8, ptr [[A_SROA_6_0_SRC_SROA_IDX]], i64 2
+; CHECK-NEXT:    [[A_SROA_6_SROA_3_0_COPYLOAD:%.*]] = load i8, ptr [[A_SROA_6_SROA_3_0_A_SROA_6_0_SRC_SROA_IDX_SROA_IDX]], align 1, !tbaa [[TBAA0]]
+; CHECK-NEXT:    [[A_SROA_6_SROA_4_0_A_SROA_6_0_SRC_SROA_IDX_SROA_IDX:%.*]] = getelementptr inbounds i8, ptr [[A_SROA_6_0_SRC_SROA_IDX]], i64 3
+; CHECK-NEXT:    call void @llvm.memcpy.p0.p0.i32(ptr align 1 [[A_SROA_6_SROA_4]], ptr align 1 [[A_SROA_6_SROA_4_0_A_SROA_6_0_SRC_SROA_IDX_SROA_IDX]], i32 7, i1 false), !tbaa [[TBAA0]]
 ; CHECK-NEXT:    [[A_SROA_7_0_SRC_SROA_IDX:%.*]] = getelementptr inbounds i8, ptr [[SRC]], i64 60
-; CHECK-NEXT:    [[A_SROA_7_SROA_0_0_COPYLOAD:%.*]] = load <40 x i8>, ptr [[A_SROA_7_0_SRC_SROA_IDX]], align 1, !tbaa [[TBAA0]]
-; CHECK-NEXT:    [[A_SROA_31_SROA_0_2_VEC_INSERT:%.*]] = insertelement <10 x i8> [[A_SROA_2_SROA_0_0_COPYLOAD]], i8 0, i32 2
-; CHECK-NEXT:    store <20 x i8> [[A_SROA_0_SROA_0_0_COPYLOAD]], ptr [[DST:%.*]], align 1, !tbaa [[TBAA11:![0-9]+]]
+; CHECK-NEXT:    call void @llvm.memcpy.p0.p0.i32(ptr align 1 [[A_SROA_7]], ptr align 1 [[A_SROA_7_0_SRC_SROA_IDX]], i32 40, i1 false), !tbaa [[TBAA0]]
+; CHECK-NEXT:    call void @llvm.memcpy.p0.p0.i32(ptr align 1 [[A_SROA_31_SROA_5]], ptr align 1 [[A_SROA_2_SROA_4]], i32 7, i1 false), !tbaa [[TBAA3]]
+; CHECK-NEXT:    call void @llvm.memcpy.p0.p0.i32(ptr align 1 [[A_SROA_31_SROA_5]], ptr align 1 [[A_SROA_6_SROA_4]], i32 7, i1 false), !tbaa [[TBAA5]]
+; CHECK-NEXT:    call void @llvm.memcpy.p0.p0.i32(ptr align 1 [[DST:%.*]], ptr align 1 [[A_SROA_0]], i32 20, i1 false), !tbaa [[TBAA7]]
 ; CHECK-NEXT:    [[A_SROA_2_0_DST_SROA_IDX:%.*]] = getelementptr inbounds i8, ptr [[DST]], i64 20
-; CHECK-NEXT:    store <10 x i8> [[A_SROA_2_SROA_0_0_COPYLOAD]], ptr [[A_SROA_2_0_DST_SROA_IDX]], align 1, !tbaa [[TBAA11]]
+; CHECK-NEXT:    store i16 [[A_SROA_2_SROA_0_0_COPYLOAD]], ptr [[A_SROA_2_0_DST_SROA_IDX]], align 1, !tbaa [[TBAA7]]
+; CHECK-NEXT:    [[A_SROA_2_SROA_3_0_A_SROA_2_0_DST_SROA_IDX_SROA_IDX:%.*]] = getelementptr inbounds i8, ptr [[A_SROA_2_0_DST_SROA_IDX]], i64 2
+; CHECK-NEXT:    store i8 [[A_SROA_2_SROA_3_0_COPYLOAD]], ptr [[A_SROA_2_SROA_3_0_A_SROA_2_0_DST_SROA_IDX_SROA_IDX]], align 1, !tbaa [[TBAA7]]
+; CHECK-NEXT:    [[A_SROA_2_SROA_4_0_A_SROA_2_0_DST_SROA_IDX_SROA_IDX:%.*]] = getelementptr inbounds i8, ptr [[A_SROA_2_0_DST_SROA_IDX]], i64 3
+; CHECK-NEXT:    call void @llvm.memcpy.p0.p0.i32(ptr align 1 [[A_SROA_2_SROA_4_0_A_SROA_2_0_DST_SROA_IDX_SROA_IDX]], ptr align 1 [[A_SROA_2_SROA_4]], i32 7, i1 false), !tbaa [[TBAA7]]
 ; CHECK-NEXT:    [[A_SROA_3_0_DST_SROA_IDX:%.*]] = getelementptr inbounds i8, ptr [[DST]], i64 30
-; CHECK-NEXT:    store <10 x i8> [[A_SROA_3_SROA_0_0_COPYLOAD]], ptr [[A_SROA_3_0_DST_SROA_IDX]], align 1, !tbaa [[TBAA11]]
+; CHECK-NEXT:    call void @llvm.memcpy.p0.p0.i32(ptr align 1 [[A_SROA_3_0_DST_SROA_IDX]], ptr align 1 [[A_SROA_3]], i32 10, i1 false), !tbaa [[TBAA7]]
 ; CHECK-NEXT:    [[A_SROA_31_0_DST_SROA_IDX:%.*]] = getelementptr inbounds i8, ptr [[DST]], i64 40
-; CHECK-NEXT:    store <10 x i8> [[A_SROA_6_SROA_0_0_COPYLOAD]], ptr [[A_SROA_31_0_DST_SROA_IDX]], align 1, !tbaa [[TBAA11]]
+; CHECK-NEXT:    store i16 [[A_SROA_6_SROA_0_0_COPYLOAD]], ptr [[A_SROA_31_0_DST_SROA_IDX]], align 1, !tbaa [[TBAA7]]
+; CHECK-NEXT:    [[A_SROA_31_SROA_4_0_A_SROA_31_0_DST_SROA_IDX_SROA_IDX:%.*]] = getelementptr inbounds i8, ptr [[A_SROA_31_0_DST_SROA_IDX]], i64 2
+; CHECK-NEXT:    store i8 [[A_SROA_6_SROA_3_0_COPYLOAD]], ptr [[A_SROA_31_SROA_4_0_A_SROA_31_0_DST_SROA_IDX_SROA_IDX]], align 1, !tbaa [[TBAA7]]
+; CHECK-NEXT:    [[A_SROA_31_SROA_5_0_A_SROA_31_0_DST_SROA_IDX_SROA_IDX:%.*]] = getelementptr inbounds i8, ptr [[A_SROA_31_0_DST_SROA_IDX]], i64 3
+; CHECK-NEXT:    call void @llvm.memcpy.p0.p0.i32(ptr align 1 [[A_SROA_31_SROA_5_0_A_SROA_31_0_DST_SROA_IDX_SROA_IDX]], ptr align 1 [[A_SROA_31_SROA_5]], i32 7, i1 false), !tbaa [[TBAA7]]
 ; CHECK-NEXT:    [[A_SROA_6_0_DST_SROA_IDX:%.*]] = getelementptr inbounds i8, ptr [[DST]], i64 50
-; CHECK-NEXT:    store <10 x i8> [[A_SROA_6_SROA_0_0_COPYLOAD]], ptr [[A_SROA_6_0_DST_SROA_IDX]], align 1, !tbaa [[TBAA11]]
+; CHECK-NEXT:    store i16 [[A_SROA_6_SROA_0_0_COPYLOAD]], ptr [[A_SROA_6_0_DST_SROA_IDX]], align 1, !tbaa [[TBAA7]]
+; CHECK-NEXT:    [[A_SROA_6_SROA_3_0_A_SROA_6_0_DST_SROA_IDX_SROA_IDX:%.*]] = getelementptr inbounds i8, ptr [[A_SROA_6_0_DST_SROA_IDX]], i64 2
+; CHECK-NEXT:    store i8 [[A_SROA_6_SROA_3_0_COPYLOAD]], ptr [[A_SROA_6_SROA_3_0_A_SROA_6_0_DST_SROA_IDX_SROA_IDX]], align 1, !tbaa [[TBAA7]]
+; CHECK-NEXT:    [[A_SROA_6_SROA_4_0_A_SROA_6_0_DST_SROA_IDX_SROA_IDX:%.*]] = getelementptr inbounds i8, ptr [[A_SROA_6_0_DST_SROA_IDX]], i64 3
+; CHECK-NEXT:    call void @llvm.memcpy.p0.p0.i32(ptr align 1 [[A_SROA_6_SROA_4_0_A_SROA_6_0_DST_SROA_IDX_SROA_IDX]], ptr align 1 [[A_SROA_6_SROA_4]], i32 7, i1 false), !tbaa [[TBAA7]]
 ; CHECK-NEXT:    [[A_SROA_7_0_DST_SROA_IDX:%.*]] = getelementptr inbounds i8, ptr [[DST]], i64 60
-; CHECK-NEXT:    store <40 x i8> [[A_SROA_7_SROA_0_0_COPYLOAD]], ptr [[A_SROA_7_0_DST_SROA_IDX]], align 1, !tbaa [[TBAA11]]
+; CHECK-NEXT:    call void @llvm.memcpy.p0.p0.i32(ptr align 1 [[A_SROA_7_0_DST_SROA_IDX]], ptr align 1 [[A_SROA_7]], i32 40, i1 false), !tbaa [[TBAA7]]
 ; CHECK-NEXT:    ret void
 ;
 entry:
@@ -397,8 +445,8 @@ define void @test7(ptr %src, ptr %dst) {
 ; CHECK-NEXT:    [[A_SROA_0:%.*]] = alloca i32, align 4
 ; CHECK-NEXT:    [[A_SROA_0_0_COPYLOAD:%.*]] = load volatile i32, ptr [[SRC:%.*]], align 1, !tbaa [[TBAA0]]
 ; CHECK-NEXT:    store volatile i32 [[A_SROA_0_0_COPYLOAD]], ptr [[A_SROA_0]], align 4, !tbaa [[TBAA0]]
-; CHECK-NEXT:    [[A_SROA_0_0_A_SROA_0_0_COPYLOAD1:%.*]] = load volatile i32, ptr [[A_SROA_0]], align 4, !tbaa [[TBAA13:![0-9]+]]
-; CHECK-NEXT:    store volatile i32 [[A_SROA_0_0_A_SROA_0_0_COPYLOAD1]], ptr [[DST:%.*]], align 1, !tbaa [[TBAA13]]
+; CHECK-NEXT:    [[A_SROA_0_0_A_SROA_0_0_COPYLOAD1:%.*]] = load volatile i32, ptr [[A_SROA_0]], align 4, !tbaa [[TBAA3]]
+; CHECK-NEXT:    store volatile i32 [[A_SROA_0_0_A_SROA_0_0_COPYLOAD1]], ptr [[DST:%.*]], align 1, !tbaa [[TBAA3]]
 ; CHECK-NEXT:    ret void
 ;
 entry:
@@ -417,9 +465,9 @@ define %S2 @test8(ptr %arg) {
 ; CHECK-NEXT:  entry:
 ; CHECK-NEXT:    [[S2_NEXT_PTR:%.*]] = getelementptr [[S2:%.*]], ptr [[ARG:%.*]], i64 0, i32 1
 ; CHECK-NEXT:    [[S2_NEXT:%.*]] = load ptr, ptr [[S2_NEXT_PTR]], align 8, !tbaa [[TBAA0]]
-; CHECK-NEXT:    [[S2_NEXT_S1:%.*]] = load ptr, ptr [[S2_NEXT]], align 8, !tbaa [[TBAA13]]
+; CHECK-NEXT:    [[S2_NEXT_S1:%.*]] = load ptr, ptr [[S2_NEXT]], align 8, !tbaa [[TBAA3]]
 ; CHECK-NEXT:    [[S2_NEXT_NEXT_PTR:%.*]] = getelementptr [[S2]], ptr [[S2_NEXT]], i64 0, i32 1
-; CHECK-NEXT:    [[S2_NEXT_NEXT:%.*]] = load ptr, ptr [[S2_NEXT_NEXT_PTR]], align 8, !tbaa [[TBAA11]]
+; CHECK-NEXT:    [[S2_NEXT_NEXT:%.*]] = load ptr, ptr [[S2_NEXT_NEXT_PTR]], align 8, !tbaa [[TBAA7]]
 ; CHECK-NEXT:    [[RESULT1:%.*]] = insertvalue [[S2]] poison, ptr [[S2_NEXT_S1]], 0
 ; CHECK-NEXT:    [[RESULT2:%.*]] = insertvalue [[S2]] [[RESULT1]], ptr [[S2_NEXT_NEXT]], 1
 ; CHECK-NEXT:    ret [[S2]] [[RESULT2]]
@@ -665,7 +713,7 @@ define void @test16(ptr %src, ptr %dst) {
 ; CHECK-LABEL: @test16(
 ; CHECK-NEXT:  entry:
 ; CHECK-NEXT:    [[A_SROA_0_0_COPYLOAD:%.*]] = load i24, ptr [[SRC:%.*]], align 1, !tbaa [[TBAA0]]
-; CHECK-NEXT:    store i24 0, ptr [[DST:%.*]], align 1, !tbaa [[TBAA15:![0-9]+]]
+; CHECK-NEXT:    store i24 0, ptr [[DST:%.*]], align 1, !tbaa [[TBAA5]]
 ; CHECK-NEXT:    ret void
 ;
 entry:
@@ -684,7 +732,7 @@ define void @test17(ptr %src, ptr %dst) {
 ; CHECK-NEXT:  entry:
 ; CHECK-NEXT:    [[A:%.*]] = alloca [3 x i8], align 1
 ; CHECK-NEXT:    call void @llvm.memcpy.p0.p0.i32(ptr [[A]], ptr [[SRC:%.*]], i32 4, i1 true), !tbaa [[TBAA0]]
-; CHECK-NEXT:    call void @llvm.memcpy.p0.p0.i32(ptr [[DST:%.*]], ptr [[A]], i32 4, i1 true), !tbaa [[TBAA13]]
+; CHECK-NEXT:    call void @llvm.memcpy.p0.p0.i32(ptr [[DST:%.*]], ptr [[A]], i32 4, i1 true), !tbaa [[TBAA3]]
 ; CHECK-NEXT:    ret void
 ;
 entry:
@@ -705,12 +753,12 @@ define void @test18(ptr %src, ptr %dst, i32 %size) {
 ; CHECK-NEXT:    [[A_SROA_0_0_COPYLOAD:%.*]] = load i32, ptr [[SRC:%.*]], align 1, !tbaa [[TBAA0]]
 ; CHECK-NEXT:    [[A_SROA_3_0_SRC_SROA_IDX:%.*]] = getelementptr inbounds i8, ptr [[SRC]], i64 4
 ; CHECK-NEXT:    [[A_SROA_3_0_COPYLOAD:%.*]] = load i32, ptr [[A_SROA_3_0_SRC_SROA_IDX]], align 1, !tbaa [[TBAA0]]
-; CHECK-NEXT:    call void @llvm.memcpy.p0.p0.i32(ptr align 1 [[A_SROA_33]], ptr [[SRC]], i32 [[SIZE:%.*]], i1 false), !tbaa [[TBAA13]]
-; CHECK-NEXT:    call void @llvm.memset.p0.i32(ptr align 1 [[A_SROA_33]], i8 42, i32 [[SIZE]], i1 false), !tbaa [[TBAA15]]
-; CHECK-NEXT:    store i32 42, ptr [[DST:%.*]], align 1, !tbaa [[TBAA17:![0-9]+]]
+; CHECK-NEXT:    call void @llvm.memcpy.p0.p0.i32(ptr align 1 [[A_SROA_33]], ptr [[SRC]], i32 [[SIZE:%.*]], i1 false), !tbaa [[TBAA3]]
+; CHECK-NEXT:    call void @llvm.memset.p0.i32(ptr align 1 [[A_SROA_33]], i8 42, i32 [[SIZE]], i1 false), !tbaa [[TBAA5]]
+; CHECK-NEXT:    store i32 42, ptr [[DST:%.*]], align 1, !tbaa [[TBAA9]]
 ; CHECK-NEXT:    [[A_SROA_3_0_DST_SROA_IDX:%.*]] = getelementptr inbounds i8, ptr [[DST]], i64 4
-; CHECK-NEXT:    store i32 [[A_SROA_3_0_COPYLOAD]], ptr [[A_SROA_3_0_DST_SROA_IDX]], align 1, !tbaa [[TBAA17]]
-; CHECK-NEXT:    call void @llvm.memcpy.p0.p0.i32(ptr [[DST]], ptr align 1 [[A_SROA_33]], i32 [[SIZE]], i1 false), !tbaa [[TBAA19:![0-9]+]]
+; CHECK-NEXT:    store i32 [[A_SROA_3_0_COPYLOAD]], ptr [[A_SROA_3_0_DST_SROA_IDX]], align 1, !tbaa [[TBAA9]]
+; CHECK-NEXT:    call void @llvm.memcpy.p0.p0.i32(ptr [[DST]], ptr align 1 [[A_SROA_33]], i32 [[SIZE]], i1 false), !tbaa [[TBAA11]]
 ; CHECK-NEXT:    ret void
 ;
 entry:
@@ -945,7 +993,8 @@ define void @PR14034(ptr %ptr, ptr %ptr2) {
 ;
 ; CHECK-LABEL: @PR14034(
 ; CHECK-NEXT:  entry:
-; CHECK-NEXT:    store <12 x i8> undef, ptr [[PTR2:%.*]], align 1
+; CHECK-NEXT:    [[A_SROA_0:%.*]] = alloca [12 x i8], align 8
+; CHECK-NEXT:    call void @llvm.memcpy.p0.p0.i32(ptr align 1 [[PTR2:%.*]], ptr align 8 [[A_SROA_0]], i32 12, i1 false)
 ; CHECK-NEXT:    ret void
 ;
 entry:
@@ -1521,8 +1570,8 @@ define void @test24(ptr %src, ptr %dst) {
 ; CHECK-NEXT:    [[A:%.*]] = alloca i64, align 16
 ; CHECK-NEXT:    [[A_0_COPYLOAD:%.*]] = load volatile i64, ptr [[SRC:%.*]], align 1, !tbaa [[TBAA0]]
 ; CHECK-NEXT:    store volatile i64 [[A_0_COPYLOAD]], ptr [[A]], align 16, !tbaa [[TBAA0]]
-; CHECK-NEXT:    [[A_0_COPYLOAD1:%.*]] = load volatile i64, ptr [[A]], align 16, !tbaa [[TBAA13]]
-; CHECK-NEXT:    store volatile i64 [[A_0_COPYLOAD1]], ptr [[DST:%.*]], align 1, !tbaa [[TBAA13]]
+; CHECK-NEXT:    [[A_0_COPYLOAD1:%.*]] = load volatile i64, ptr [[A]], align 16, !tbaa [[TBAA3]]
+; CHECK-NEXT:    store volatile i64 [[A_0_COPYLOAD1]], ptr [[DST:%.*]], align 1, !tbaa [[TBAA3]]
 ; CHECK-NEXT:    ret void
 ;
 entry:

diff  --git a/llvm/test/Transforms/SROA/pointer-offset-size.ll b/llvm/test/Transforms/SROA/pointer-offset-size.ll
index b2856bca05026..779b6d5001234 100644
--- a/llvm/test/Transforms/SROA/pointer-offset-size.ll
+++ b/llvm/test/Transforms/SROA/pointer-offset-size.ll
@@ -9,7 +9,8 @@ target datalayout = "e-p:64:64:64:32"
 define i16 @test(ptr %ts2.i) {
 ; CHECK-LABEL: @test(
 ; CHECK-NEXT:  entry:
-; CHECK-NEXT:    store <3 x i8> undef, ptr [[TS2_I:%.*]], align 1
+; CHECK-NEXT:    [[S_SROA_0:%.*]] = alloca [3 x i8], align 8
+; CHECK-NEXT:    call void @llvm.memcpy.p0.p0.i32(ptr align 1 [[TS2_I:%.*]], ptr align 8 [[S_SROA_0]], i32 3, i1 false)
 ; CHECK-NEXT:    [[TMP0:%.*]] = load i16, ptr [[TS2_I]], align 2
 ; CHECK-NEXT:    ret i16 [[TMP0]]
 ;

diff  --git a/llvm/test/Transforms/SROA/scalable-vectors.ll b/llvm/test/Transforms/SROA/scalable-vectors.ll
index 35cf0e8f664bc..2b2f6f17bbd42 100644
--- a/llvm/test/Transforms/SROA/scalable-vectors.ll
+++ b/llvm/test/Transforms/SROA/scalable-vectors.ll
@@ -67,7 +67,7 @@ define <vscale x 4 x i32> @cast_alloca_to_svint32_t(<vscale x 4 x i32> %type.coe
 define <vscale x 4 x i32> @cast_alloca_from_svint32_t() {
 ; CHECK-LABEL: @cast_alloca_from_svint32_t(
 ; CHECK-NEXT:    [[RETVAL_COERCE:%.*]] = alloca <vscale x 4 x i32>, align 16
-; CHECK-NEXT:    store <64 x i8> undef, ptr [[RETVAL_COERCE]], align 16
+; CHECK-NEXT:    store <16 x i32> undef, ptr [[RETVAL_COERCE]], align 16
 ; CHECK-NEXT:    [[TMP1:%.*]] = load <vscale x 4 x i32>, ptr [[RETVAL_COERCE]], align 16
 ; CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 ;

diff  --git a/llvm/test/Transforms/SROA/slice-width.ll b/llvm/test/Transforms/SROA/slice-width.ll
index cbf0ace8174ac..43d8a0a19763b 100644
--- a/llvm/test/Transforms/SROA/slice-width.ll
+++ b/llvm/test/Transforms/SROA/slice-width.ll
@@ -47,7 +47,8 @@ load_i1:
 
 define void @memcpy_fp80_padding() {
 ; CHECK-LABEL: @memcpy_fp80_padding(
-; CHECK-NEXT:    [[X_SROA_0_SROA_0_0_COPYLOAD:%.*]] = load <16 x i8>, ptr @foo_copy_source, align 16
+; CHECK-NEXT:    [[X_SROA_0:%.*]] = alloca x86_fp80, align 16
+; CHECK-NEXT:    call void @llvm.memcpy.p0.p0.i32(ptr align 16 [[X_SROA_0]], ptr align 16 @foo_copy_source, i32 16, i1 false)
 ; CHECK-NEXT:    [[X_SROA_1_0_COPYLOAD:%.*]] = load i64, ptr getelementptr inbounds (i8, ptr @foo_copy_source, i64 16), align 16
 ; CHECK-NEXT:    [[X_SROA_2_0_COPYLOAD:%.*]] = load i64, ptr getelementptr inbounds (i8, ptr @foo_copy_source, i64 24), align 8
 ; CHECK-NEXT:    store i64 [[X_SROA_1_0_COPYLOAD]], ptr @i64_sink, align 4

diff  --git a/llvm/test/Transforms/SROA/tbaa-struct.ll b/llvm/test/Transforms/SROA/tbaa-struct.ll
index 2ec2599204dc3..6d15a3a4adc99 100644
--- a/llvm/test/Transforms/SROA/tbaa-struct.ll
+++ b/llvm/test/Transforms/SROA/tbaa-struct.ll
@@ -11,8 +11,7 @@ declare <2 x float> @foo(ptr %0)
 define void @bar(ptr %y2) {
 ; CHECK-LABEL: @bar(
 ; CHECK-NEXT:    [[X14:%.*]] = call <2 x float> @foo(ptr [[Y2:%.*]])
-; CHECK-NEXT:    [[TMP1:%.*]] = bitcast <2 x float> [[X14]] to <2 x i32>
-; CHECK-NEXT:    store <2 x i32> [[TMP1]], ptr [[Y2]], align 4, !tbaa.struct !0
+; CHECK-NEXT:    store <2 x float> [[X14]], ptr [[Y2]], align 4, !tbaa.struct !0
 ; CHECK-NEXT:    ret void
 ;
   %x7 = alloca %vector

diff  --git a/llvm/test/Transforms/SROA/tbaa-struct2.ll b/llvm/test/Transforms/SROA/tbaa-struct2.ll
index ab62ef819be8a..f2e4cf86fd273 100644
--- a/llvm/test/Transforms/SROA/tbaa-struct2.ll
+++ b/llvm/test/Transforms/SROA/tbaa-struct2.ll
@@ -10,11 +10,12 @@ declare double @subcall(double %g, i32 %m)
 
 define double @bar(ptr %wishart) {
 ; CHECK-LABEL: @bar(
+; CHECK-NEXT:    [[TMP_SROA_3:%.*]] = alloca [4 x i8], align 4
 ; CHECK-NEXT:    [[TMP_SROA_0_0_COPYLOAD:%.*]] = load double, ptr [[WISHART:%.*]], align 8, !tbaa.struct !0
 ; CHECK-NEXT:    [[TMP_SROA_2_0_WISHART_SROA_IDX:%.*]] = getelementptr inbounds i8, ptr [[WISHART]], i64 8
 ; CHECK-NEXT:    [[TMP_SROA_2_0_COPYLOAD:%.*]] = load i32, ptr [[TMP_SROA_2_0_WISHART_SROA_IDX]], align 8, !tbaa.struct !7
 ; CHECK-NEXT:    [[TMP_SROA_3_0_WISHART_SROA_IDX:%.*]] = getelementptr inbounds i8, ptr [[WISHART]], i64 12
-; CHECK-NEXT:    [[TMP_SROA_3_SROA_0_0_COPYLOAD:%.*]] = load <4 x i8>, ptr [[TMP_SROA_3_0_WISHART_SROA_IDX]], align 4, !tbaa.struct !8
+; CHECK-NEXT:    call void @llvm.memcpy.p0.p0.i64(ptr align 4 [[TMP_SROA_3]], ptr align 4 [[TMP_SROA_3_0_WISHART_SROA_IDX]], i64 4, i1 false), !tbaa.struct !8
 ; CHECK-NEXT:    [[CALL:%.*]] = call double @subcall(double [[TMP_SROA_0_0_COPYLOAD]], i32 [[TMP_SROA_2_0_COPYLOAD]])
 ; CHECK-NEXT:    ret double [[CALL]]
 ;

diff  --git a/llvm/test/Transforms/SROA/vector-promotion.ll b/llvm/test/Transforms/SROA/vector-promotion.ll
index c83363d4efbdb..2c48fb7726d94 100644
--- a/llvm/test/Transforms/SROA/vector-promotion.ll
+++ b/llvm/test/Transforms/SROA/vector-promotion.ll
@@ -573,9 +573,9 @@ define <4 x float> @test12(<4 x i32> %val) {
 
 define void @swap-8bytes(ptr %x, ptr %y) {
 ; CHECK-LABEL: @swap-8bytes(
-; CHECK-NEXT:    [[TMP_SROA_0_0_COPYLOAD:%.*]] = load <8 x i8>, ptr [[X:%.*]], align 1
+; CHECK-NEXT:    [[TMP_SROA_0_0_COPYLOAD:%.*]] = load i64, ptr [[X:%.*]], align 1
 ; CHECK-NEXT:    tail call void @llvm.memcpy.p0.p0.i64(ptr [[X]], ptr [[Y:%.*]], i64 8, i1 false)
-; CHECK-NEXT:    store <8 x i8> [[TMP_SROA_0_0_COPYLOAD]], ptr [[Y]], align 1
+; CHECK-NEXT:    store i64 [[TMP_SROA_0_0_COPYLOAD]], ptr [[Y]], align 1
 ; CHECK-NEXT:    ret void
 ;
   %tmp = alloca [2 x i32]
@@ -587,9 +587,10 @@ define void @swap-8bytes(ptr %x, ptr %y) {
 
 define void @swap-7bytes(ptr %x, ptr %y) {
 ; CHECK-LABEL: @swap-7bytes(
-; CHECK-NEXT:    [[TMP_SROA_0_0_COPYLOAD:%.*]] = load <7 x i8>, ptr [[X:%.*]], align 1
+; CHECK-NEXT:    [[TMP:%.*]] = alloca [7 x i8], align 1
+; CHECK-NEXT:    call void @llvm.memcpy.p0.p0.i64(ptr [[TMP]], ptr [[X:%.*]], i64 7, i1 false)
 ; CHECK-NEXT:    tail call void @llvm.memcpy.p0.p0.i64(ptr [[X]], ptr [[Y:%.*]], i64 7, i1 false)
-; CHECK-NEXT:    store <7 x i8> [[TMP_SROA_0_0_COPYLOAD]], ptr [[Y]], align 1
+; CHECK-NEXT:    call void @llvm.memcpy.p0.p0.i64(ptr [[Y]], ptr [[TMP]], i64 7, i1 false)
 ; CHECK-NEXT:    ret void
 ;
   %tmp = alloca [7 x i8]
@@ -601,9 +602,10 @@ define void @swap-7bytes(ptr %x, ptr %y) {
 
 define void @swap-16bytes(ptr %x, ptr %y) {
 ; CHECK-LABEL: @swap-16bytes(
-; CHECK-NEXT:    [[TMP_SROA_0_0_COPYLOAD:%.*]] = load <16 x i8>, ptr [[X:%.*]], align 1
+; CHECK-NEXT:    [[TMP:%.*]] = alloca [2 x i64], align 8
+; CHECK-NEXT:    call void @llvm.memcpy.p0.p0.i64(ptr [[TMP]], ptr [[X:%.*]], i64 16, i1 false)
 ; CHECK-NEXT:    tail call void @llvm.memcpy.p0.p0.i64(ptr [[X]], ptr [[Y:%.*]], i64 16, i1 false)
-; CHECK-NEXT:    store <16 x i8> [[TMP_SROA_0_0_COPYLOAD]], ptr [[Y]], align 1
+; CHECK-NEXT:    call void @llvm.memcpy.p0.p0.i64(ptr [[Y]], ptr [[TMP]], i64 16, i1 false)
 ; CHECK-NEXT:    ret void
 ;
   %tmp = alloca [2 x i64]
@@ -615,9 +617,10 @@ define void @swap-16bytes(ptr %x, ptr %y) {
 
 define void @swap-15bytes(ptr %x, ptr %y) {
 ; CHECK-LABEL: @swap-15bytes(
-; CHECK-NEXT:    [[TMP_SROA_0_0_COPYLOAD:%.*]] = load <15 x i8>, ptr [[X:%.*]], align 1
+; CHECK-NEXT:    [[TMP:%.*]] = alloca [15 x i8], align 1
+; CHECK-NEXT:    call void @llvm.memcpy.p0.p0.i64(ptr [[TMP]], ptr [[X:%.*]], i64 15, i1 false)
 ; CHECK-NEXT:    tail call void @llvm.memcpy.p0.p0.i64(ptr [[X]], ptr [[Y:%.*]], i64 15, i1 false)
-; CHECK-NEXT:    store <15 x i8> [[TMP_SROA_0_0_COPYLOAD]], ptr [[Y]], align 1
+; CHECK-NEXT:    call void @llvm.memcpy.p0.p0.i64(ptr [[Y]], ptr [[TMP]], i64 15, i1 false)
 ; CHECK-NEXT:    ret void
 ;
   %tmp = alloca [15 x i8]