Generalize the transform that boosts GEP indices to the size of a pointer to

also do it for vectors of pointers.

llvm-svn: 167354

2 files changed, 13 insertions, 2 deletions

diff --git a/llvm/lib/Transforms/InstCombine/InstructionCombining.cpp b/llvm/lib/Transforms/InstCombine/InstructionCombining.cpp
index 7f8c3ae5581..ccf75bca2b9 100644
--- a/llvm/lib/Transforms/InstCombine/InstructionCombining.cpp
+++ b/llvm/lib/Transforms/InstCombine/InstructionCombining.cpp
@@ -1055,7 +1055,7 @@ Instruction *InstCombiner::visitGetElementPtrInst(GetElementPtrInst &GEP) {
   // by multiples of a zero size type with zero.
   if (TD) {
     bool MadeChange = false;
-    Type *IntPtrTy = TD->getIntPtrType(GEP.getContext());
+    Type *IntPtrTy = TD->getIntPtrType(GEP.getPointerOperandType());
 
     gep_type_iterator GTI = gep_type_begin(GEP);
     for (User::op_iterator I = GEP.op_begin() + 1, E = GEP.op_end();
@@ -1074,7 +1074,7 @@ Instruction *InstCombiner::visitGetElementPtrInst(GetElementPtrInst &GEP) {
         }
 
       Type *IndexTy = (*I)->getType();
-      if (IndexTy != IntPtrTy && !IndexTy->isVectorTy()) {
+      if (IndexTy != IntPtrTy) {
         // If we are using a wider index than needed for this platform, shrink
         // it to what we need.  If narrower, sign-extend it to what we need.
         // This explicit cast can make subsequent optimizations more obvious.
diff --git a/llvm/test/Transforms/InstCombine/vector_gep2.ll b/llvm/test/Transforms/InstCombine/vector_gep2.ll
new file mode 100644
index 00000000000..20165b11001
--- /dev/null
+++ b/llvm/test/Transforms/InstCombine/vector_gep2.ll
@@ -0,0 +1,11 @@
+; RUN: opt < %s -instcombine -S | FileCheck %s
+
+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-n8:16:32:64-S128"
+target triple = "x86_64-unknown-linux-gnu"
+
+define <2 x i8*> @testa(<2 x i8*> %a) {
+; CHECK: @testa
+  %g = getelementptr <2 x i8*> %a, <2 x i32> <i32 0, i32 1>
+; CHECK: getelementptr <2 x i8*> %a, <2 x i64> <i64 0, i64 1>
+  ret <2 x i8*> %g
+}