diff --git a/llvm/include/llvm/Transforms/Utils/Evaluator.h b/llvm/include/llvm/Transforms/Utils/Evaluator.h
--- a/llvm/include/llvm/Transforms/Utils/Evaluator.h
+++ b/llvm/include/llvm/Transforms/Utils/Evaluator.h
@@ -138,6 +138,8 @@
                        SmallVectorImpl<Constant *> &Formals);
 
   Constant *ComputeLoadResult(Constant *P, Type *Ty);
+  Constant *ComputeLoadResult(GlobalVariable *GV, Type *Ty,
+                              const APInt &Offset);
 
   /// As we compute SSA register values, we store their contents here. The back
   /// of the deque contains the current function and the stack contains the
diff --git a/llvm/lib/Transforms/Utils/Evaluator.cpp b/llvm/lib/Transforms/Utils/Evaluator.cpp
--- a/llvm/lib/Transforms/Utils/Evaluator.cpp
+++ b/llvm/lib/Transforms/Utils/Evaluator.cpp
@@ -217,10 +217,13 @@
   P = cast<Constant>(P->stripAndAccumulateConstantOffsets(
       DL, Offset, /* AllowNonInbounds */ true));
   Offset = Offset.sextOrTrunc(DL.getIndexTypeSizeInBits(P->getType()));
-  auto *GV = dyn_cast<GlobalVariable>(P);
-  if (!GV)
-    return nullptr;
+  if (auto *GV = dyn_cast<GlobalVariable>(P))
+    return ComputeLoadResult(GV, Ty, Offset);
+  return nullptr;
+}
 
+Constant *Evaluator::ComputeLoadResult(GlobalVariable *GV, Type *Ty,
+                                       const APInt &Offset) {
   auto It = MutatedMemory.find(GV);
   if (It != MutatedMemory.end())
     return It->second.read(Ty, Offset, DL);
@@ -436,16 +439,39 @@
                               << "intrinsic.\n");
             return false;
           }
+
+          auto *LenC = dyn_cast<ConstantInt>(getVal(MSI->getLength()));
+          if (!LenC) {
+            LLVM_DEBUG(dbgs() << "Memset with unknown length.\n");
+            return false;
+          }
+
           Constant *Ptr = getVal(MSI->getDest());
+          APInt Offset(DL.getIndexTypeSizeInBits(Ptr->getType()), 0);
+          Ptr = cast<Constant>(Ptr->stripAndAccumulateConstantOffsets(
+              DL, Offset, /* AllowNonInbounds */ true));
+          auto *GV = dyn_cast<GlobalVariable>(Ptr);
+          if (!GV) {
+            LLVM_DEBUG(dbgs() << "Memset with unknown base.\n");
+            return false;
+          }
+
           Constant *Val = getVal(MSI->getValue());
-          Constant *DestVal =
-              ComputeLoadResult(getVal(Ptr), MSI->getValue()->getType());
-          if (Val->isNullValue() && DestVal && DestVal->isNullValue()) {
-            // This memset is a no-op.
-            LLVM_DEBUG(dbgs() << "Ignoring no-op memset.\n");
-            ++CurInst;
-            continue;
+          APInt Len = LenC->getValue();
+          while (Len != 0) {
+            Constant *DestVal = ComputeLoadResult(GV, Val->getType(), Offset);
+            if (DestVal != Val) {
+              LLVM_DEBUG(dbgs() << "Memset is not a no-op at offset "
+                                << Offset << " of " << *GV << ".\n");
+              return false;
+            }
+            ++Offset;
+            --Len;
           }
+
+          LLVM_DEBUG(dbgs() << "Ignoring no-op memset.\n");
+          ++CurInst;
+          continue;
         }
 
         if (II->isLifetimeStartOrEnd()) {
diff --git a/llvm/test/Transforms/GlobalOpt/ctor-memset.ll b/llvm/test/Transforms/GlobalOpt/ctor-memset.ll
--- a/llvm/test/Transforms/GlobalOpt/ctor-memset.ll
+++ b/llvm/test/Transforms/GlobalOpt/ctor-memset.ll
@@ -1,7 +1,9 @@
 ; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --check-globals
 ; RUN: opt -S -globalopt < %s | FileCheck %s
 
-@llvm.global_ctors = appending global [8 x { i32, ptr, ptr }] [
+target datalayout = "p1:32:32"
+
+@llvm.global_ctors = appending global [9 x { i32, ptr, ptr }] [
   { i32, ptr, ptr } { i32 65535, ptr @ctor0, ptr null },
   { i32, ptr, ptr } { i32 65535, ptr @ctor1, ptr null },
   { i32, ptr, ptr } { i32 65535, ptr @ctor2, ptr null },
@@ -9,11 +11,12 @@
   { i32, ptr, ptr } { i32 65535, ptr @ctor4, ptr null },
   { i32, ptr, ptr } { i32 65535, ptr @ctor5, ptr null },
   { i32, ptr, ptr } { i32 65535, ptr @ctor6, ptr null },
-  { i32, ptr, ptr } { i32 65535, ptr @ctor7, ptr null }
+  { i32, ptr, ptr } { i32 65535, ptr @ctor7, ptr null },
+  { i32, ptr, ptr } { i32 65535, ptr @ctor8, ptr null }
 ]
 
 ;.
-; CHECK: @[[LLVM_GLOBAL_CTORS:[a-zA-Z0-9_$"\\.-]+]] = appending global [2 x { i32, ptr, ptr }] [{ i32, ptr, ptr } { i32 65535, ptr @ctor6, ptr null }, { i32, ptr, ptr } { i32 65535, ptr @ctor7, ptr null }]
+; CHECK: @[[LLVM_GLOBAL_CTORS:[a-zA-Z0-9_$"\\.-]+]] = appending global [3 x { i32, ptr, ptr }] [{ i32, ptr, ptr } { i32 65535, ptr @ctor3, ptr null }, { i32, ptr, ptr } { i32 65535, ptr @ctor4, ptr null }, { i32, ptr, ptr } { i32 65535, ptr @ctor7, ptr null }]
 ; CHECK: @[[G0:[a-zA-Z0-9_$"\\.-]+]] = local_unnamed_addr global { i32, i32 } zeroinitializer
 ; CHECK: @[[G1:[a-zA-Z0-9_$"\\.-]+]] = local_unnamed_addr global { i32, i32, i32 } { i32 0, i32 0, i32 1 }
 ; CHECK: @[[G2:[a-zA-Z0-9_$"\\.-]+]] = local_unnamed_addr global { i32, i32, i32 } { i32 1, i32 0, i32 0 }
@@ -22,11 +25,11 @@
 ; CHECK: @[[G5:[a-zA-Z0-9_$"\\.-]+]] = local_unnamed_addr global { i16, i32 } { i16 0, i32 1 }
 ; CHECK: @[[G6:[a-zA-Z0-9_$"\\.-]+]] = local_unnamed_addr global { i32, i32 } { i32 -1, i32 -1 }
 ; CHECK: @[[G7:[a-zA-Z0-9_$"\\.-]+]] = local_unnamed_addr global { i32, i32 } { i32 -1, i32 1 }
+; CHECK: @[[G8:[a-zA-Z0-9_$"\\.-]+]] = local_unnamed_addr addrspace(1) global { i32, i32 } zeroinitializer
 ;.
 
 ; memset of all-zero global
 @g0 = global { i32, i32 } zeroinitializer
-
 define internal void @ctor0() {
   call void @llvm.memset.p0.i64(ptr @g0, i8 0, i64 8, i1 false)
   ret void
@@ -52,6 +55,10 @@
 @g3 = global { i32, i32 } { i32 0, i32 1 }
 
 define internal void @ctor3() {
+; CHECK-LABEL: @ctor3(
+; CHECK-NEXT:    call void @llvm.memset.p0.i64(ptr @g3, i8 0, i64 8, i1 false)
+; CHECK-NEXT:    ret void
+;
   call void @llvm.memset.p0.i64(ptr @g3, i8 0, i64 8, i1 false)
   ret void
 }
@@ -60,11 +67,17 @@
 @g4 = global { i32, i32 } { i32 0, i32 undef }
 
 define internal void @ctor4() {
+; CHECK-LABEL: @ctor4(
+; CHECK-NEXT:    call void @llvm.memset.p0.i64(ptr @g4, i8 0, i64 8, i1 false)
+; CHECK-NEXT:    ret void
+;
   call void @llvm.memset.p0.i64(ptr @g4, i8 0, i64 8, i1 false)
   ret void
 }
 
 ; memset including padding bytes
+; FIXME: We still incorrectly optimize the memset away here, even though code
+; might access the padding.
 @g5 = global { i16, i32 } { i16 0, i32 1 }
 
 define internal void @ctor5() {
@@ -76,10 +89,6 @@
 @g6 = global { i32, i32 } { i32 -1, i32 -1 }
 
 define internal void @ctor6() {
-; CHECK-LABEL: @ctor6(
-; CHECK-NEXT:    call void @llvm.memset.p0.i64(ptr @g6, i8 -1, i64 8, i1 false)
-; CHECK-NEXT:    ret void
-;
   call void @llvm.memset.p0.i64(ptr @g6, i8 -1, i64 8, i1 false)
   ret void
 }
@@ -96,6 +105,14 @@
   ret void
 }
 
+; memset of zero value in differently-sized address space
+@g8 = addrspace(1) global { i32, i32 } zeroinitializer
+
+define internal void @ctor8() {
+  call void @llvm.memset.p0.i64(ptr addrspacecast (ptr addrspace(1) @g8 to ptr), i8 0, i64 8, i1 false)
+  ret void
+}
+
 declare void @llvm.memset.p0.i64(ptr, i8, i64, i1)
 ;.
 ; CHECK: attributes #[[ATTR0:[0-9]+]] = { argmemonly nofree nounwind willreturn writeonly }