Index: llvm/lib/Analysis/Loads.cpp
===================================================================
--- llvm/lib/Analysis/Loads.cpp
+++ llvm/lib/Analysis/Loads.cpp
@@ -664,6 +664,37 @@
   return nullptr;
 }
 
+static bool mayOverlapLoadedMemory(const StoreInst *Store, const LoadInst *Load,
+                                   const DataLayout &DL) {
+  if (Store->getValueOperand() != Load) {
+    return true;
+  }
+  auto *Ty = Load->getType();
+  if (Ty->isScalableTy()) {
+    return true;
+  }
+  auto AlignValue = Load->getAlign().value();
+  if (Store->getAlign().value() != AlignValue) {
+    return true;
+  }
+  auto TyStoreSize = DL.getTypeStoreSize(Ty).getFixedSize();
+  if (TyStoreSize > AlignValue) {
+    return true;
+  }
+  return false;
+}
+
+static bool writesUnchangedLoadedValueToSameLocation(const Instruction *Inst,
+                                                     const Value *Available,
+                                                     const DataLayout &DL) {
+  auto *Store = dyn_cast_or_null<StoreInst>(Inst);
+  auto *Load = dyn_cast_or_null<LoadInst>(Available);
+  if (!Store || !Load || mayOverlapLoadedMemory(Store, Load, DL)) {
+    return false;
+  }
+  return true;
+}
+
 Value *llvm::FindAvailableLoadedValue(LoadInst *Load, AAResults &AA,
                                       bool *IsLoadCSE,
                                       unsigned MaxInstsToScan) {
@@ -701,9 +732,13 @@
   // did not modify the memory location.
   if (Available) {
     MemoryLocation Loc = MemoryLocation::get(Load);
-    for (Instruction *Inst : MustNotAliasInsts)
+    for (Instruction *Inst : MustNotAliasInsts) {
+      if (writesUnchangedLoadedValueToSameLocation(Inst, Available, DL)) {
+        continue;
+      }
       if (isModSet(AA.getModRefInfo(Inst, Loc)))
         return nullptr;
+    }
   }
 
   return Available;
Index: llvm/test/Transforms/InstCombine/load-elimination.ll
===================================================================
--- /dev/null
+++ llvm/test/Transforms/InstCombine/load-elimination.ll
@@ -0,0 +1,71 @@
+; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
+; RUN: opt -passes=instcombine -S < %s | FileCheck %s
+
+define void @load_elimination_i8(ptr noundef %a, ptr noundef %b, ptr noundef %c) {
+; CHECK-LABEL: @load_elimination_i8(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i8, ptr [[C:%.*]], align 1
+; CHECK-NEXT:    store i8 [[TMP0]], ptr [[A:%.*]], align 1
+; CHECK-NEXT:    store i8 [[TMP0]], ptr [[B:%.*]], align 1
+; CHECK-NEXT:    ret void
+;
+entry:
+  %0 = load i8, ptr %c, align 1
+  store i8 %0, ptr %a, align 1
+  %1 = load i8, ptr %c, align 1
+  store i8 %1, ptr %b, align 1
+  ret void
+}
+
+define void @load_elimination_i32(ptr noundef %a, ptr noundef %b, ptr noundef %c) {
+; CHECK-LABEL: @load_elimination_i32(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, ptr [[C:%.*]], align 4
+; CHECK-NEXT:    store i32 [[TMP0]], ptr [[A:%.*]], align 4
+; CHECK-NEXT:    store i32 [[TMP0]], ptr [[B:%.*]], align 4
+; CHECK-NEXT:    ret void
+;
+entry:
+  %0 = load i32, ptr %c, align 4
+  store i32 %0, ptr %a, align 4
+  %1 = load i32, ptr %c, align 4
+  store i32 %1, ptr %b, align 4
+  ret void
+}
+
+define void @load_elimination_i32_i8(ptr noundef %a, ptr noundef %b, ptr noundef %c) {
+; CHECK-LABEL: @load_elimination_i32_i8(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i8, ptr [[C:%.*]], align 1
+; CHECK-NEXT:    [[CONV:%.*]] = zext i8 [[TMP0]] to i32
+; CHECK-NEXT:    store i32 [[CONV]], ptr [[A:%.*]], align 4
+; CHECK-NEXT:    [[TMP1:%.*]] = load i8, ptr [[C]], align 1
+; CHECK-NEXT:    [[CONV1:%.*]] = zext i8 [[TMP1]] to i32
+; CHECK-NEXT:    store i32 [[CONV1]], ptr [[B:%.*]], align 4
+; CHECK-NEXT:    ret void
+;
+entry:
+  %0 = load i8, ptr %c, align 1
+  %conv = zext i8 %0 to i32
+  store i32 %conv, ptr %a, align 4
+  %1 = load i8, ptr %c, align 1
+  %conv1 = zext i8 %1 to i32
+  store i32 %conv1, ptr %b, align 4
+  ret void
+}
+
+define void @load_elimination_float(ptr %x, ptr %a, ptr %b, ptr %c) {
+; CHECK-LABEL: @load_elimination_float(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[X1:%.*]] = load float, ptr [[X:%.*]], align 4
+; CHECK-NEXT:    store float [[X1]], ptr [[A:%.*]], align 4
+; CHECK-NEXT:    store float [[X1]], ptr [[C:%.*]], align 4
+; CHECK-NEXT:    ret void
+;
+entry:
+  %x1 = load float, ptr %x
+  store float %x1, ptr %a
+  %x2 = load float, ptr %x
+  store float %x2, ptr %c
+  ret void
+}
Index: llvm/test/Transforms/InstCombine/load.ll
===================================================================
--- llvm/test/Transforms/InstCombine/load.ll
+++ llvm/test/Transforms/InstCombine/load.ll
@@ -213,9 +213,7 @@
 ; CHECK-NEXT:    [[X1:%.*]] = load float, ptr [[X:%.*]], align 4
 ; CHECK-NEXT:    store float [[X1]], ptr [[A:%.*]], align 4
 ; CHECK-NEXT:    store float [[X1]], ptr [[B:%.*]], align 4
-; CHECK-NEXT:    [[X2:%.*]] = load float, ptr [[X]], align 4
-; CHECK-NEXT:    store float [[X2]], ptr [[B]], align 4
-; CHECK-NEXT:    store float [[X2]], ptr [[C:%.*]], align 4
+; CHECK-NEXT:    store float [[X1]], ptr [[C:%.*]], align 4
 ; CHECK-NEXT:    ret void
 ;
 entry:
@@ -238,9 +236,7 @@
 ; CHECK-NEXT:    [[X1:%.*]] = load <4 x i8>, ptr [[X:%.*]], align 4
 ; CHECK-NEXT:    store <4 x i8> [[X1]], ptr [[A:%.*]], align 4
 ; CHECK-NEXT:    store <4 x i8> [[X1]], ptr [[B:%.*]], align 4
-; CHECK-NEXT:    [[X2:%.*]] = load <4 x i8>, ptr [[X]], align 4
-; CHECK-NEXT:    store <4 x i8> [[X2]], ptr [[B]], align 4
-; CHECK-NEXT:    store <4 x i8> [[X2]], ptr [[C:%.*]], align 4
+; CHECK-NEXT:    store <4 x i8> [[X1]], ptr [[C:%.*]], align 4
 ; CHECK-NEXT:    ret void
 ;
 entry: