Index: lib/Target/AMDGPU/AMDGPUPromoteAlloca.cpp
===================================================================
--- lib/Target/AMDGPU/AMDGPUPromoteAlloca.cpp
+++ lib/Target/AMDGPU/AMDGPUPromoteAlloca.cpp
@@ -549,6 +549,13 @@
     if (UseInst->getOpcode() == Instruction::PtrToInt)
       return false;
 
+    if (LoadInst *LI = dyn_cast_or_null<LoadInst>(UseInst)) {
+      if (LI->isVolatile())
+        return false;
+
+      continue;
+    }
+
     if (StoreInst *SI = dyn_cast<StoreInst>(UseInst)) {
       if (SI->isVolatile())
         return false;
@@ -556,9 +563,6 @@
       // Reject if the stored value is not the pointer operand.
       if (SI->getPointerOperand() != Val)
         return false;
-    } else if (LoadInst *LI = dyn_cast_or_null<LoadInst>(UseInst)) {
-      if (LI->isVolatile())
-        return false;
     } else if (AtomicRMWInst *RMW = dyn_cast_or_null<AtomicRMWInst>(UseInst)) {
       if (RMW->isVolatile())
         return false;
Index: test/CodeGen/AMDGPU/amdgpu.private-memory.ll
===================================================================
--- test/CodeGen/AMDGPU/amdgpu.private-memory.ll
+++ test/CodeGen/AMDGPU/amdgpu.private-memory.ll
@@ -1,14 +1,16 @@
-; RUN: llc -march=r600 -mcpu=cypress < %s | FileCheck %s -check-prefix=R600 -check-prefix=FUNC
 ; RUN: llc -show-mc-encoding -mattr=+promote-alloca -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 -verify-machineinstrs -mtriple=amdgcn--amdhsa -mcpu=kaveri < %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 -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 -verify-machineinstrs -mtriple=amdgcn-amdhsa -mcpu=kaveri < %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 -verify-machineinstrs -march=amdgcn -mcpu=tonga < %s | FileCheck %s -check-prefix=SI-PROMOTE -check-prefix=SI -check-prefix=FUNC
-; RUN: llc -show-mc-encoding -mattr=-promote-alloca -verify-machineinstrs -march=amdgcn -mcpu=tonga < %s | FileCheck %s -check-prefix=SI-ALLOCA -check-prefix=SI -check-prefix=FUNC
+; RUN: llc -show-mc-encoding -mattr=+promote-alloca -verify-machineinstrs -mtriple=amdgcn-amdhsa -march=amdgcn -mcpu=tonga < %s | FileCheck %s -check-prefix=SI-PROMOTE -check-prefix=SI -check-prefix=FUNC
+; RUN: llc -show-mc-encoding -mattr=-promote-alloca -verify-machineinstrs -mtriple=amdgcn-amdhsa -march=amdgcn -mcpu=tonga < %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: llc -march=r600 -mcpu=cypress < %s | FileCheck %s -check-prefix=R600 -check-prefix=FUNC
+
+
 ; HSAOPT: @mova_same_clause.stack = internal unnamed_addr addrspace(3) global [256 x [5 x i32]] undef, align 4
 ; HSAOPT: @high_alignment.stack = internal unnamed_addr addrspace(3) global [256 x [8 x i32]] undef, align 16
 
@@ -396,6 +398,25 @@
   ret void
 }
 
+; OPT-LABEL: @pointer_typed_alloca(
+; OPT:  getelementptr inbounds [256 x i32 addrspace(1)*], [256 x i32 addrspace(1)*] addrspace(3)* @pointer_typed_alloca.A.addr, i32 0, i32 %{{[0-9]+}}
+; OPT: load i32 addrspace(1)*, i32 addrspace(1)* addrspace(3)* %{{[0-9]+}}, align 4
+define void @pointer_typed_alloca(i32 addrspace(1)* %A) {
+entry:
+  %A.addr = alloca i32 addrspace(1)*, align 4
+  store i32 addrspace(1)* %A, i32 addrspace(1)** %A.addr, align 4
+  %ld0 = load i32 addrspace(1)*, i32 addrspace(1)** %A.addr, align 4
+  %arrayidx = getelementptr inbounds i32, i32 addrspace(1)* %ld0, i32 0
+  store i32 1, i32 addrspace(1)* %arrayidx, align 4
+  %ld1 = load i32 addrspace(1)*, i32 addrspace(1)** %A.addr, align 4
+  %arrayidx1 = getelementptr inbounds i32, i32 addrspace(1)* %ld1, i32 1
+  store i32 2, i32 addrspace(1)* %arrayidx1, align 4
+  %ld2 = load i32 addrspace(1)*, i32 addrspace(1)** %A.addr, align 4
+  %arrayidx2 = getelementptr inbounds i32, i32 addrspace(1)* %ld2, i32 2
+  store i32 3, i32 addrspace(1)* %arrayidx2, align 4
+  ret void
+}
+
 ; HSAOPT: !0 = !{}
 ; HSAOPT: !1 = !{i32 0, i32 2048}
 
Index: test/CodeGen/AMDGPU/promote-alloca-volatile.ll
===================================================================
--- test/CodeGen/AMDGPU/promote-alloca-volatile.ll
+++ test/CodeGen/AMDGPU/promote-alloca-volatile.ll
@@ -24,3 +24,22 @@
   store volatile i32 %tmp, i32* %arrayidx1
  ret void
 }
+
+; Has on OK non-volatile user but also a volatile user
+; CHECK-LABEL: @volatile_and_non_volatile_load(
+; CHECK: alloca double
+; CHECK: load double
+; CHECK: load volatile double
+define void @volatile_and_non_volatile_load(double addrspace(1)* nocapture %arg, i32 %arg1) #0 {
+bb:
+  %tmp = alloca double, align 8
+  store double 0.000000e+00, double* %tmp, align 8
+
+  %tmp4 = load double, double* %tmp, align 8
+  %tmp5 = load volatile double, double* %tmp, align 8
+
+  store double %tmp4, double addrspace(1)* %arg
+  ret void
+}
+
+attributes #0 = { nounwind }