Index: lib/Target/AMDGPU/AMDGPUPromoteAlloca.cpp
===================================================================
--- lib/Target/AMDGPU/AMDGPUPromoteAlloca.cpp
+++ lib/Target/AMDGPU/AMDGPUPromoteAlloca.cpp
@@ -65,6 +65,11 @@
 
 namespace {
 
+static cl::opt<bool> DisablePromoteAllocaToVector(
+  "disable-promote-alloca-to-vector",
+  cl::desc("Disable promote alloca to vector"),
+  cl::init(false));
+
 // FIXME: This can create globals so should be a module pass.
 class AMDGPUPromoteAlloca : public FunctionPass {
 private:
@@ -335,15 +340,21 @@
 }
 
 static bool tryPromoteAllocaToVector(AllocaInst *Alloca, AMDGPUAS AS) {
-  ArrayType *AllocaTy = dyn_cast<ArrayType>(Alloca->getAllocatedType());
 
   DEBUG(dbgs() << "Alloca candidate for vectorization\n");
 
+  if (DisablePromoteAllocaToVector) {
+    DEBUG(dbgs() << "  Promotion alloca to vector is disabled\n");
+    return false;
+  }
+
+  ArrayType *AllocaTy = dyn_cast<ArrayType>(Alloca->getAllocatedType());
+
   // FIXME: There is no reason why we can't support larger arrays, we
   // are just being conservative for now.
   if (!AllocaTy ||
       AllocaTy->getElementType()->isVectorTy() ||
-      AllocaTy->getNumElements() > 4 ||
+      AllocaTy->getNumElements() > 16 ||
       AllocaTy->getNumElements() < 2) {
     DEBUG(dbgs() << "  Cannot convert type to vector\n");
     return false;
Index: lib/Target/AMDGPU/SIISelLowering.cpp
===================================================================
--- lib/Target/AMDGPU/SIISelLowering.cpp
+++ lib/Target/AMDGPU/SIISelLowering.cpp
@@ -1892,7 +1892,7 @@
       BuildMI(MBB, I, DL, TII->get(AMDGPU::S_SET_GPR_IDX_OFF));
     } else {
       BuildMI(MBB, I, DL, TII->get(AMDGPU::V_MOVRELS_B32_e32), Dst)
-        .addReg(SrcReg, RegState::Undef, SubReg)
+        .addReg(SrcReg, 0, SubReg)
         .addReg(SrcReg, RegState::Implicit);
     }
 
@@ -1929,7 +1929,7 @@
       .addReg(AMDGPU::M0, RegState::Implicit);
   } else {
     BuildMI(*LoopBB, InsPt, DL, TII->get(AMDGPU::V_MOVRELS_B32_e32), Dst)
-      .addReg(SrcReg, RegState::Undef, SubReg)
+      .addReg(SrcReg, 0, SubReg)
       .addReg(SrcReg, RegState::Implicit);
   }
 
Index: test/CodeGen/AMDGPU/amdgpu.private-memory.ll
===================================================================
--- test/CodeGen/AMDGPU/amdgpu.private-memory.ll
+++ test/CodeGen/AMDGPU/amdgpu.private-memory.ll
@@ -1,14 +1,18 @@
-; RUN: llc -show-mc-encoding -mattr=+promote-alloca -amdgpu-load-store-vectorizer=0 -enable-amdgpu-aa=0 -verify-machineinstrs -march=amdgcn < %s | FileCheck %s -check-prefix=SI-PROMOTE -check-prefix=SI -check-prefix=FUNC
-; RUN: llc -show-mc-encoding -mattr=+promote-alloca -amdgpu-load-store-vectorizer=0 -enable-amdgpu-aa=0 -verify-machineinstrs -mtriple=amdgcn--amdhsa -mcpu=kaveri -mattr=-unaligned-buffer-access < %s | FileCheck %s -check-prefix=SI-PROMOTE -check-prefix=SI -check-prefix=FUNC -check-prefix=HSA-PROMOTE
+; RUN: llc -show-mc-encoding -mattr=+promote-alloca -disable-promote-alloca-to-vector -amdgpu-load-store-vectorizer=0 -enable-amdgpu-aa=0 -verify-machineinstrs -march=amdgcn < %s | FileCheck %s -check-prefix=SI-PROMOTE -check-prefix=SI -check-prefix=FUNC
+; RUN: llc -show-mc-encoding -mattr=+promote-alloca -amdgpu-load-store-vectorizer=0 -enable-amdgpu-aa=0 -verify-machineinstrs -march=amdgcn < %s | FileCheck %s -check-prefix=SI-PROMOTE-VECT -check-prefix=SI -check-prefix=FUNC
+; RUN: llc -show-mc-encoding -mattr=+promote-alloca -disable-promote-alloca-to-vector -amdgpu-load-store-vectorizer=0 -enable-amdgpu-aa=0 -verify-machineinstrs -mtriple=amdgcn--amdhsa -mcpu=kaveri -mattr=-unaligned-buffer-access < %s | FileCheck %s -check-prefix=SI-PROMOTE -check-prefix=SI -check-prefix=FUNC -check-prefix=HSA-PROMOTE
+; RUN: llc -show-mc-encoding -mattr=+promote-alloca -amdgpu-load-store-vectorizer=0 -enable-amdgpu-aa=0 -verify-machineinstrs -mtriple=amdgcn--amdhsa -mcpu=kaveri -mattr=-unaligned-buffer-access < %s | FileCheck %s -check-prefix=SI-PROMOTE-VECT -check-prefix=SI -check-prefix=FUNC
 ; RUN: llc -show-mc-encoding -mattr=-promote-alloca -amdgpu-load-store-vectorizer=0 -enable-amdgpu-aa=0 -verify-machineinstrs -march=amdgcn < %s | FileCheck %s -check-prefix=SI-ALLOCA -check-prefix=SI -check-prefix=FUNC
 ; RUN: llc -show-mc-encoding -mattr=-promote-alloca -amdgpu-load-store-vectorizer=0 -enable-amdgpu-aa=0 -verify-machineinstrs -mtriple=amdgcn-amdhsa -mcpu=kaveri -mattr=-unaligned-buffer-access < %s | FileCheck %s -check-prefix=SI-ALLOCA -check-prefix=SI -check-prefix=FUNC -check-prefix=HSA-ALLOCA
-; RUN: llc -show-mc-encoding -mattr=+promote-alloca -amdgpu-load-store-vectorizer=0 -enable-amdgpu-aa=0 -verify-machineinstrs -mtriple=amdgcn-amdhsa -march=amdgcn -mcpu=tonga -mattr=-unaligned-buffer-access < %s | FileCheck %s -check-prefix=SI-PROMOTE -check-prefix=SI -check-prefix=FUNC
+; RUN: llc -show-mc-encoding -mattr=+promote-alloca -disable-promote-alloca-to-vector -amdgpu-load-store-vectorizer=0 -enable-amdgpu-aa=0 -verify-machineinstrs -mtriple=amdgcn-amdhsa -march=amdgcn -mcpu=tonga -mattr=-unaligned-buffer-access < %s | FileCheck %s -check-prefix=SI-PROMOTE -check-prefix=SI -check-prefix=FUNC
+; RUN: llc -show-mc-encoding -mattr=+promote-alloca -amdgpu-load-store-vectorizer=0 -enable-amdgpu-aa=0 -verify-machineinstrs -mtriple=amdgcn-amdhsa -march=amdgcn -mcpu=tonga -mattr=-unaligned-buffer-access < %s | FileCheck %s -check-prefix=SI-PROMOTE-VECT -check-prefix=SI -check-prefix=FUNC
 ; RUN: llc -show-mc-encoding -mattr=-promote-alloca -amdgpu-load-store-vectorizer=0 -enable-amdgpu-aa=0 -verify-machineinstrs -mtriple=amdgcn-amdhsa -march=amdgcn -mcpu=tonga -mattr=-unaligned-buffer-access < %s | FileCheck %s -check-prefix=SI-ALLOCA -check-prefix=SI -check-prefix=FUNC
 
-; RUN: opt -S -mtriple=amdgcn-unknown-amdhsa -mcpu=kaveri -amdgpu-promote-alloca < %s | FileCheck -check-prefix=HSAOPT -check-prefix=OPT %s
-; RUN: opt -S -mtriple=amdgcn-unknown-unknown -mcpu=kaveri -amdgpu-promote-alloca < %s | FileCheck -check-prefix=NOHSAOPT -check-prefix=OPT %s
+; RUN: opt -S -mtriple=amdgcn-unknown-amdhsa -mcpu=kaveri -amdgpu-promote-alloca -disable-promote-alloca-to-vector < %s | FileCheck -check-prefix=HSAOPT -check-prefix=OPT %s
+; RUN: opt -S -mtriple=amdgcn-unknown-unknown -mcpu=kaveri -amdgpu-promote-alloca -disable-promote-alloca-to-vector < %s | FileCheck -check-prefix=NOHSAOPT -check-prefix=OPT %s
 
-; RUN: llc -march=r600 -mcpu=cypress < %s | FileCheck %s -check-prefix=R600 -check-prefix=FUNC
+; RUN: llc -march=r600 -mcpu=cypress -disable-promote-alloca-to-vector < %s | FileCheck %s -check-prefix=R600 -check-prefix=FUNC
+; RUN: llc -march=r600 -mcpu=cypress < %s | FileCheck %s -check-prefix=R600-VECT -check-prefix=FUNC
 
 
 ; HSAOPT: @mova_same_clause.stack = internal unnamed_addr addrspace(3) global [256 x [5 x i32]] undef, align 4
@@ -223,16 +227,16 @@
 
 ; FUNC-LABEL: {{^}}short_array:
 
-; R600: MOVA_INT
+; R600-VECT: MOVA_INT
 
 ; SI-ALLOCA-DAG: buffer_store_short v{{[0-9]+}}, off, s[{{[0-9]+:[0-9]+}}], s{{[0-9]+}} offset:6 ; encoding: [0x06,0x00,0x68,0xe0
 ; SI-ALLOCA-DAG: buffer_store_short v{{[0-9]+}}, off, s[{{[0-9]+:[0-9]+}}], s{{[0-9]+}} offset:4 ; encoding: [0x04,0x00,0x68,0xe0
 ; Loaded value is 0 or 1, so sext will become zext, so we get buffer_load_ushort instead of buffer_load_sshort.
 ; SI-ALLOCA: buffer_load_sshort v{{[0-9]+}}, v{{[0-9]+}}, s[{{[0-9]+:[0-9]+}}], s{{[0-9]+}}
 
-; SI-PROMOTE: s_load_dword [[IDX:s[0-9]+]]
-; SI-PROMOTE: s_lshl_b32 [[SCALED_IDX:s[0-9]+]], [[IDX]], 16
-; SI-PROMOTE: v_bfe_u32 v{{[0-9]+}}, v{{[0-9]+}}, [[SCALED_IDX]], 16
+; SI-PROMOTE-VECT: s_load_dword [[IDX:s[0-9]+]]
+; SI-PROMOTE-VECT: s_lshl_b32 [[SCALED_IDX:s[0-9]+]], [[IDX]], 16
+; SI-PROMOTE-VECT: v_bfe_u32 v{{[0-9]+}}, v{{[0-9]+}}, [[SCALED_IDX]], 16
 define amdgpu_kernel void @short_array(i32 addrspace(1)* %out, i32 %index) #0 {
 entry:
   %0 = alloca [2 x i16]
@@ -249,10 +253,10 @@
 
 ; FUNC-LABEL: {{^}}char_array:
 
-; R600: MOVA_INT
+; R600-VECT: MOVA_INT
 
-; SI-PROMOTE-DAG: buffer_store_byte v{{[0-9]+}}, off, s[{{[0-9]+:[0-9]+}}], s{{[0-9]+}} offset:4 ; encoding:
-; SI-PROMOTE-DAG: buffer_store_byte v{{[0-9]+}}, off, s[{{[0-9]+:[0-9]+}}], s{{[0-9]+}} offset:5 ; encoding:
+; SI-PROMOTE-VECT-DAG: buffer_store_byte v{{[0-9]+}}, off, s[{{[0-9]+:[0-9]+}}], s{{[0-9]+}} offset:4 ; encoding:
+; SI-PROMOTE-VECT-DAG: buffer_store_byte v{{[0-9]+}}, off, s[{{[0-9]+:[0-9]+}}], s{{[0-9]+}} offset:5 ; encoding:
 
 ; SI-ALLOCA-DAG: buffer_store_byte v{{[0-9]+}}, off, s[{{[0-9]+:[0-9]+}}], s{{[0-9]+}} offset:4 ; encoding: [0x04,0x00,0x60,0xe0
 ; SI-ALLOCA-DAG: buffer_store_byte v{{[0-9]+}}, off, s[{{[0-9]+:[0-9]+}}], s{{[0-9]+}} offset:5 ; encoding: [0x05,0x00,0x60,0xe0
Index: test/CodeGen/AMDGPU/indirect-private-64.ll
===================================================================
--- test/CodeGen/AMDGPU/indirect-private-64.ll
+++ test/CodeGen/AMDGPU/indirect-private-64.ll
@@ -1,8 +1,8 @@
 ; RUN: llc -march=amdgcn -mattr=-promote-alloca,+max-private-element-size-16 -verify-machineinstrs < %s | FileCheck -check-prefix=SI-ALLOCA16 -check-prefix=SI %s
 ; RUN: llc -march=amdgcn -mattr=-promote-alloca,+max-private-element-size-4 -verify-machineinstrs < %s | FileCheck -check-prefix=SI-ALLOCA4 -check-prefix=SI %s
-; RUN: llc -march=amdgcn -mattr=+promote-alloca -verify-machineinstrs < %s | FileCheck -check-prefix=SI-PROMOTE -check-prefix=SI %s
+; RUN: llc -march=amdgcn -mattr=+promote-alloca -disable-promote-alloca-to-vector -verify-machineinstrs < %s | FileCheck -check-prefix=SI-PROMOTE -check-prefix=SI %s
 ; RUN: llc -march=amdgcn -mcpu=tonga -mattr=-flat-for-global -mattr=-promote-alloca,+max-private-element-size-16 -verify-machineinstrs < %s | FileCheck -check-prefix=CI-ALLOCA16 -check-prefix=SI %s
-; RUN: llc -march=amdgcn -mcpu=tonga -mattr=-flat-for-global -mattr=+promote-alloca -verify-machineinstrs < %s | FileCheck -check-prefix=CI-PROMOTE -check-prefix=SI %s
+; RUN: llc -march=amdgcn -mcpu=tonga -mattr=-flat-for-global -mattr=+promote-alloca -disable-promote-alloca-to-vector -verify-machineinstrs < %s | FileCheck -check-prefix=CI-PROMOTE -check-prefix=SI %s
 
 declare void @llvm.amdgcn.s.barrier() #0
 
Index: test/CodeGen/AMDGPU/large-work-group-promote-alloca.ll
===================================================================
--- test/CodeGen/AMDGPU/large-work-group-promote-alloca.ll
+++ test/CodeGen/AMDGPU/large-work-group-promote-alloca.ll
@@ -1,5 +1,5 @@
-; RUN: opt -S -mtriple=amdgcn-unknown-unknown -amdgpu-promote-alloca < %s | FileCheck --check-prefix=SI --check-prefix=ALL %s
-; RUN: opt -S -mcpu=tonga -mtriple=amdgcn-unknown-unknown -amdgpu-promote-alloca < %s | FileCheck --check-prefix=CI --check-prefix=ALL %s
+; RUN: opt -S -mtriple=amdgcn-unknown-unknown -amdgpu-promote-alloca -disable-promote-alloca-to-vector < %s | FileCheck --check-prefix=SI --check-prefix=ALL %s
+; RUN: opt -S -mcpu=tonga -mtriple=amdgcn-unknown-unknown -amdgpu-promote-alloca -disable-promote-alloca-to-vector < %s | FileCheck --check-prefix=CI --check-prefix=ALL %s
 
 ; SI-NOT: @promote_alloca_size_63.stack = internal unnamed_addr addrspace(3) global [63 x [5 x i32]] undef, align 4
 ; CI: @promote_alloca_size_63.stack = internal unnamed_addr addrspace(3) global [63 x [5 x i32]] undef, align 4
Index: test/CodeGen/AMDGPU/private-memory-r600.ll
===================================================================
--- test/CodeGen/AMDGPU/private-memory-r600.ll
+++ test/CodeGen/AMDGPU/private-memory-r600.ll
@@ -1,5 +1,6 @@
-; RUN: llc -march=r600 -mcpu=redwood < %s | FileCheck %s -check-prefix=R600 -check-prefix=FUNC
-; RUN: opt -S -mtriple=r600-unknown-unknown -mcpu=redwood -amdgpu-promote-alloca < %s | FileCheck -check-prefix=OPT %s
+; RUN: llc -march=r600 -mcpu=redwood -disable-promote-alloca-to-vector < %s | FileCheck %s -check-prefix=R600 -check-prefix=FUNC
+; RUN: llc -march=r600 -mcpu=redwood < %s | FileCheck %s -check-prefix=R600-VECT -check-prefix=FUNC
+; RUN: opt -S -mtriple=r600-unknown-unknown -mcpu=redwood -amdgpu-promote-alloca -disable-promote-alloca-to-vector < %s | FileCheck -check-prefix=OPT %s
 
 declare i32 @llvm.r600.read.tidig.x() nounwind readnone
 
@@ -109,7 +110,7 @@
 
 ; FUNC-LABEL: {{^}}short_array:
 
-; R600: MOVA_INT
+; R600-VECT: MOVA_INT
 define amdgpu_kernel void @short_array(i32 addrspace(1)* %out, i32 %index) #0 {
 entry:
   %0 = alloca [2 x i16]
@@ -126,7 +127,7 @@
 
 ; FUNC-LABEL: {{^}}char_array:
 
-; R600: MOVA_INT
+; R600-VECT: MOVA_INT
 define amdgpu_kernel void @char_array(i32 addrspace(1)* %out, i32 %index) #0 {
 entry:
   %0 = alloca [2 x i8]
Index: test/CodeGen/AMDGPU/promote-alloca-padding-size-estimate.ll
===================================================================
--- test/CodeGen/AMDGPU/promote-alloca-padding-size-estimate.ll
+++ test/CodeGen/AMDGPU/promote-alloca-padding-size-estimate.ll
@@ -1,4 +1,4 @@
-; RUN: llc -mtriple=amdgcn-amd-amdhsa < %s | FileCheck -check-prefix=GCN %s
+; RUN: llc -mtriple=amdgcn-amd-amdhsa -disable-promote-alloca-to-vector < %s | FileCheck -check-prefix=GCN %s
 
 ; This shows that the amount of LDS estimate is sensitive to the order
 ; of the LDS globals.
Index: test/CodeGen/AMDGPU/target-cpu.ll
===================================================================
--- test/CodeGen/AMDGPU/target-cpu.ll
+++ test/CodeGen/AMDGPU/target-cpu.ll
@@ -1,4 +1,4 @@
-; RUN: llc -march=amdgcn -verify-machineinstrs < %s | FileCheck %s
+; RUN: llc -march=amdgcn -disable-promote-alloca-to-vector -verify-machineinstrs < %s | FileCheck %s
 
 declare i8 addrspace(2)* @llvm.amdgcn.kernarg.segment.ptr() #1