diff --git a/llvm/lib/Target/AMDGPU/AMDGPUAsmPrinter.cpp b/llvm/lib/Target/AMDGPU/AMDGPUAsmPrinter.cpp
index e05e9159427abfa919a36c82e8585e29f23cb08e..7b27e16081be826bc057e669fc4c6f4ac8878b75 100644
--- a/llvm/lib/Target/AMDGPU/AMDGPUAsmPrinter.cpp
+++ b/llvm/lib/Target/AMDGPU/AMDGPUAsmPrinter.cpp
@@ -504,6 +504,11 @@ void AMDGPUAsmPrinter::getSIProgramInfo(SIProgramInfo &ProgInfo,
     Ctx.emitError("too many user SGPRs used");
   }
 
+  if (MFI->LDSSize > static_cast<unsigned>(STM.getLocalMemorySize())) {
+    LLVMContext &Ctx = MF.getFunction()->getContext();
+    Ctx.emitError("LDS size exceeds device maximum");
+  }
+
   ProgInfo.VGPRBlocks = (ProgInfo.NumVGPR - 1) / 4;
   ProgInfo.SGPRBlocks = (ProgInfo.NumSGPR - 1) / 8;
   // Set the value to initialize FP_ROUND and FP_DENORM parts of the mode
diff --git a/llvm/test/CodeGen/AMDGPU/32-bit-local-address-space.ll b/llvm/test/CodeGen/AMDGPU/32-bit-local-address-space.ll
index c7bcfd2ddab20cc60c24cb4a6d7f02c5cf1a62af..ff8c904578763cb426df8ec19b33ee84b2d06c07 100644
--- a/llvm/test/CodeGen/AMDGPU/32-bit-local-address-space.ll
+++ b/llvm/test/CodeGen/AMDGPU/32-bit-local-address-space.ll
@@ -1,4 +1,4 @@
-; RUN: llc -march=amdgcn -mcpu=verde -verify-machineinstrs < %s | FileCheck -check-prefix=SI -check-prefix=FUNC %s
+; RUN: llc -march=amdgcn -mcpu=bonaire -verify-machineinstrs < %s | FileCheck -check-prefix=SI -check-prefix=FUNC %s
 ; RUN: llc -march=amdgcn -mcpu=tonga -verify-machineinstrs < %s | FileCheck -check-prefix=SI -check-prefix=FUNC %s
 
 ; On Southern Islands GPUs the local address space(3) uses 32-bit pointers and
@@ -91,12 +91,12 @@ define void @infer_ptr_alignment_global_offset(float addrspace(1)* %out, i32 %ti
 
 
 @ptr = addrspace(3) global i32 addrspace(3)* undef
-@dst = addrspace(3) global [16384 x i32] undef
+@dst = addrspace(3) global [16383 x i32] undef
 
 ; FUNC-LABEL: {{^}}global_ptr:
 ; SI: ds_write_b32
 define void @global_ptr() nounwind {
-  store i32 addrspace(3)* getelementptr ([16384 x i32], [16384 x i32] addrspace(3)* @dst, i32 0, i32 16), i32 addrspace(3)* addrspace(3)* @ptr
+  store i32 addrspace(3)* getelementptr ([16383 x i32], [16383 x i32] addrspace(3)* @dst, i32 0, i32 16), i32 addrspace(3)* addrspace(3)* @ptr
   ret void
 }
 
diff --git a/llvm/test/CodeGen/AMDGPU/over-max-lds-size.ll b/llvm/test/CodeGen/AMDGPU/over-max-lds-size.ll
new file mode 100644
index 0000000000000000000000000000000000000000..459d913ef301dbf2907653e86cdc431ee32045f7
--- /dev/null
+++ b/llvm/test/CodeGen/AMDGPU/over-max-lds-size.ll
@@ -0,0 +1,14 @@
+; RUN: not llc -march=amdgcn -mcpu=tahiti < %s 2>&1 | FileCheck -check-prefix=ERROR %s
+; RUN: not llc -march=amdgcn -mcpu=hawaii < %s 2>&1 | FileCheck -check-prefix=ERROR %s
+; RUN: not llc -march=amdgcn -mcpu=fiji < %s 2>&1 | FileCheck -check-prefix=ERROR %s
+
+; ERROR: error: LDS size exceeds device maximum
+
+@huge = internal unnamed_addr addrspace(3) global [100000 x i32] undef, align 4
+
+define void @promote_alloca_size_256() {
+entry:
+  %v0 = getelementptr inbounds [100000 x i32], [100000 x i32] addrspace(3)* @huge, i32 0, i32 0
+  store i32 0, i32 addrspace(3)* %v0
+  ret void
+}