diff --git a/lib/Transforms/Scalar/InstructionCombining.cpp b/lib/Transforms/Scalar/InstructionCombining.cpp
index f02a7108fd6..1cb4fa2c213 100644
--- a/lib/Transforms/Scalar/InstructionCombining.cpp
+++ b/lib/Transforms/Scalar/InstructionCombining.cpp
@@ -9795,7 +9795,8 @@ Instruction *InstCombiner::visitGetElementPtrInst(GetElementPtrInst &GEP) {
         }
       }
       // If we are using a wider index than needed for this platform, shrink it
-      // to what we need.  If the incoming value needs a cast instruction,
+      // to what we need.  If narrower, sign-extend it to what we need.
+      // If the incoming value needs a cast instruction,
       // insert it.  This explicit cast can make subsequent optimizations more
       // obvious.
       Value *Op = *i;
@@ -9809,6 +9810,16 @@ Instruction *InstCombiner::visitGetElementPtrInst(GetElementPtrInst &GEP) {
           *i = Op;
           MadeChange = true;
         }
+      } else if (TD->getTypeSizeInBits(Op->getType()) < TD->getPointerSizeInBits()) {
+        if (Constant *C = dyn_cast<Constant>(Op)) {
+          *i = ConstantExpr::getSExt(C, TD->getIntPtrType());
+          MadeChange = true;
+        } else {
+          Op = InsertCastBefore(Instruction::SExt, Op, TD->getIntPtrType(),
+                                GEP);
+          *i = Op;
+          MadeChange = true;
+        }
       }
     }
   }
diff --git a/test/Transforms/InstCombine/getelementptr_promote.ll b/test/Transforms/InstCombine/getelementptr_promote.ll
new file mode 100644
index 00000000000..811c5a95916
--- /dev/null
+++ b/test/Transforms/InstCombine/getelementptr_promote.ll
@@ -0,0 +1,58 @@
+; RUN: llvm-as < %s | opt -instcombine | llvm-dis | grep getelementptr | grep {, i64}
+
+; Instcombine should promote the getelementptr index up to the target's
+; pointer size, making the conversion explicit, which helps expose it to
+; other optimizations.
+
+target datalayout = "e-p:64:64:64-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:64:64-f32:32:32-f64:64:64-v64:64:64-v128:128:128-a0:0:64-s0:64:64-f80:128:128"
+target triple = "x86_64-apple-darwin8"
+
+define i64 @test(i64* %first, i32 %count) nounwind {
+entry:
+	%first_addr = alloca i64*		; <i64**> [#uses=2]
+	%count_addr = alloca i32		; <i32*> [#uses=2]
+	%retval = alloca i64		; <i64*> [#uses=2]
+	%n = alloca i32		; <i32*> [#uses=5]
+	%result = alloca i64		; <i64*> [#uses=4]
+	%0 = alloca i64		; <i64*> [#uses=2]
+	%"alloca point" = bitcast i32 0 to i32		; <i32> [#uses=0]
+	store i64* %first, i64** %first_addr
+	store i32 %count, i32* %count_addr
+	store i64 0, i64* %result, align 8
+	store i32 0, i32* %n, align 4
+	br label %bb1
+
+bb:		; preds = %bb1
+	%1 = load i64** %first_addr, align 8		; <i64*> [#uses=1]
+	%2 = load i32* %n, align 4		; <i32> [#uses=1]
+	%3 = bitcast i32 %2 to i32		; <i64> [#uses=1]
+	%4 = getelementptr i64* %1, i32 %3		; <i64*> [#uses=1]
+	%5 = load i64* %4, align 8		; <i64> [#uses=1]
+	%6 = lshr i64 %5, 4		; <i64> [#uses=1]
+	%7 = load i64* %result, align 8		; <i64> [#uses=1]
+	%8 = add i64 %6, %7		; <i64> [#uses=1]
+	store i64 %8, i64* %result, align 8
+	%9 = load i32* %n, align 4		; <i32> [#uses=1]
+	%10 = add i32 %9, 1		; <i32> [#uses=1]
+	store i32 %10, i32* %n, align 4
+	br label %bb1
+
+bb1:		; preds = %bb, %entry
+	%11 = load i32* %n, align 4		; <i32> [#uses=1]
+	%12 = load i32* %count_addr, align 4		; <i32> [#uses=1]
+	%13 = icmp slt i32 %11, %12		; <i1> [#uses=1]
+	%14 = zext i1 %13 to i8		; <i8> [#uses=1]
+	%toBool = icmp ne i8 %14, 0		; <i1> [#uses=1]
+	br i1 %toBool, label %bb, label %bb2
+
+bb2:		; preds = %bb1
+	%15 = load i64* %result, align 8		; <i64> [#uses=1]
+	store i64 %15, i64* %0, align 8
+	%16 = load i64* %0, align 8		; <i64> [#uses=1]
+	store i64 %16, i64* %retval, align 8
+	br label %return
+
+return:		; preds = %bb2
+	%retval3 = load i64* %retval		; <i64> [#uses=1]
+	ret i64 %retval3
+}