diff --git a/llvm/test/CodeGen/AMDGPU/promote-alloca-subvecs.ll b/llvm/test/CodeGen/AMDGPU/promote-alloca-subvecs.ll
--- a/llvm/test/CodeGen/AMDGPU/promote-alloca-subvecs.ll
+++ b/llvm/test/CodeGen/AMDGPU/promote-alloca-subvecs.ll
@@ -1,5 +1,5 @@
 ; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --version 2
-; RUN: opt -S -mtriple=amdgcn-unknown-amdhsa -passes=amdgpu-promote-alloca < %s | FileCheck %s
+; RUN: opt -S -mtriple=amdgcn-unknown-amdhsa -passes=amdgpu-promote-alloca -amdgpu-promote-alloca-to-vector-limit=512 < %s | FileCheck %s
 
 target datalayout = "e-p:64:64-p1:64:64-p2:32:32-p3:32:32-p4:64:64-p5:32:32-p6:32:32-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5"
 
@@ -332,3 +332,48 @@
   store <3 x i32> %val, ptr addrspace(5) %stack
   ret void
 }
+
+define void @test_accessty_too_small(<2 x i16> %val) {
+; CHECK-LABEL: define void @test_accessty_too_small
+; CHECK-SAME: (<2 x i16> [[VAL:%.*]]) {
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[STACK:%.*]] = alloca [4 x i64], align 4, addrspace(5)
+; CHECK-NEXT:    store <2 x i16> [[VAL]], ptr addrspace(5) [[STACK]], align 4
+; CHECK-NEXT:    ret void
+;
+entry:
+  %stack = alloca [4 x i64], align 4, addrspace(5)
+  store <2 x i16> %val, ptr addrspace(5) %stack
+  ret void
+}
+
+define void @store_2xi32_into_double(double %foo) {
+; CHECK-LABEL: define void @store_2xi32_into_double
+; CHECK-SAME: (double [[FOO:%.*]]) {
+; CHECK-NEXT:    [[DUMMYUSER0:%.*]] = freeze double 0x5F0000005E
+; CHECK-NEXT:    [[DUMMYUSER1:%.*]] = freeze double undef
+; CHECK-NEXT:    [[DUMMYUSER2:%.*]] = freeze double undef
+; CHECK-NEXT:    [[DUMMYUSER3:%.*]] = freeze double undef
+; CHECK-NEXT:    ret void
+;
+  %alloca = alloca [9 x double], align 8, addrspace(5)
+
+  store <2 x i32> <i32 94, i32 95>, ptr addrspace(5) %alloca, align 8
+  %load0 = load double, ptr addrspace(5) %alloca, align 8
+  %dummyuser0 = freeze double %load0
+
+  %idx4 = getelementptr inbounds [9 x double], ptr addrspace(5) %alloca, i32 0, i32 4
+  %idx5 = getelementptr inbounds [9 x double], ptr addrspace(5) %alloca, i32 0, i32 5
+  store <4 x i32> <i32 102, i32 103, i32 104, i32 105>, ptr addrspace(5) %idx4, align 8
+  %load1 = load double, ptr addrspace(5) %idx4, align 8
+  %dummyuser1 = freeze double %load1
+  %load2 = load double, ptr addrspace(5) %idx5, align 8
+  %dummyuser2 = freeze double %load2
+
+  %idx8 = getelementptr inbounds [9 x double], ptr addrspace(5) %alloca, i32 0, i32 8
+  store <2 x i32> <i32 110, i32 111>, ptr addrspace(5) %idx8, align 8
+  %load3 = load double, ptr addrspace(5) %idx8, align 8
+  %dummyuser3 = freeze double %load3
+
+  ret void
+}