7 files changed, 1379 insertions, 1 deletions
diff --git a/llvm/test/CodeGen/AMDGPU/basic-call-return.ll b/llvm/test/CodeGen/AMDGPU/basic-call-return.ll
new file mode 100644
index 00000000000..6588a1ca41e
--- /dev/null
+++ b/llvm/test/CodeGen/AMDGPU/basic-call-return.ll
@@ -0,0 +1,27 @@
+; RUN: llc -mtriple=amdgcn-amd-amdhsa -mcpu=fiji -amdgpu-function-calls -verify-machineinstrs < %s | FileCheck -check-prefix=GCN %s
+; RUN: llc -mtriple=amdgcn-amd-amdhsa -mcpu=hawaii -amdgpu-function-calls -verify-machineinstrs < %s | FileCheck -check-prefix=GCN %s
+; RUN: llc -mtriple=amdgcn-amd-amdhsa -mcpu=gfx900 -amdgpu-function-calls -verify-machineinstrs < %s | FileCheck -check-prefix=GCN %s
+
+define void @void_func_void() #2 {
+  ret void
+}
+
+; GCN-LABEL: {{^}}test_call_void_func_void:
+define amdgpu_kernel void @test_call_void_func_void() {
+  call void @void_func_void()
+  ret void
+}
+
+define void @void_func_void_clobber_s40_s41() #2 {
+  call void asm sideeffect "", "~{SGPR40_SGPR41}"() #0
+  ret void
+}
+
+define amdgpu_kernel void @test_call_void_func_void_clobber_s40_s41() {
+  call void @void_func_void_clobber_s40_s41()
+  ret void
+}
+
+attributes #0 = { nounwind }
+attributes #1 = { nounwind readnone }
+attributes #2 = { nounwind noinline }
diff --git a/llvm/test/CodeGen/AMDGPU/byval-frame-setup.ll b/llvm/test/CodeGen/AMDGPU/byval-frame-setup.ll
new file mode 100644
index 00000000000..b53447c8ffb
--- /dev/null
+++ b/llvm/test/CodeGen/AMDGPU/byval-frame-setup.ll
@@ -0,0 +1,235 @@
+; RUN: llc -mtriple=amdgcn-amd-amdhsa -mcpu=fiji -amdgpu-function-calls -verify-machineinstrs < %s | FileCheck -enable-var-scope -check-prefixes=GCN,VI %s
+; RUN: llc -mtriple=amdgcn-amd-amdhsa -mcpu=hawaii -amdgpu-function-calls -verify-machineinstrs < %s | FileCheck -enable-var-scope -check-prefixes=GCN,CI %s
+
+%struct.ByValStruct = type { [4 x i32] }
+
+; GCN-LABEL: {{^}}void_func_byval_struct:
+; GCN: s_mov_b32 s5, s32
+; GCN: buffer_load_dword [[LOAD0:v[0-9]+]], off, s[0:3], s5{{$}}
+; GCN-NOT: s32
+; GCN: buffer_store_dword [[LOAD0]], off, s[0:3], s5{{$}}
+; GCN-NOT: s32
+
+; GCN: buffer_load_dword [[LOAD1:v[0-9]+]], off, s[0:3], s5 offset:16{{$}}
+; GCN-NOT: s32
+; GCN: buffer_store_dword [[LOAD1]], off, s[0:3], s5 offset:16{{$}}
+; GCN-NOT: s32
+define void @void_func_byval_struct(%struct.ByValStruct* byval noalias nocapture align 4 %arg0, %struct.ByValStruct* byval noalias nocapture align 4 %arg1) #1 {
+entry:
+  %arrayidx = getelementptr inbounds %struct.ByValStruct, %struct.ByValStruct* %arg0, i32 0, i32 0, i32 0
+  %tmp = load volatile i32, i32* %arrayidx, align 4
+  %add = add nsw i32 %tmp, 1
+  store volatile i32 %add, i32* %arrayidx, align 4
+  %arrayidx2 = getelementptr inbounds %struct.ByValStruct, %struct.ByValStruct* %arg1, i32 0, i32 0, i32 0
+  %tmp1 = load volatile i32, i32* %arrayidx2, align 4
+  %add3 = add nsw i32 %tmp1, 2
+  store volatile i32 %add3, i32* %arrayidx2, align 4
+  store volatile i32 9, i32 addrspace(1)* null, align 4
+  ret void
+}
+
+; GCN-LABEL: {{^}}void_func_byval_struct_non_leaf:
+; GCN: s_mov_b32 s5, s32
+; GCN: buffer_store_dword v32
+; GCN: v_writelane_b32
+
+; GCN-DAG: s_add_u32 s32, s32, 0x900{{$}}
+
+; GCN-DAG: buffer_load_dword [[LOAD0:v[0-9]+]], off, s[0:3], s5{{$}}
+; GCN: v_add_i32_e32 [[ADD0:v[0-9]+]], vcc, 1, [[LOAD0]]
+; GCN: buffer_store_dword [[ADD0]], off, s[0:3], s5{{$}}
+
+; GCN: buffer_load_dword [[LOAD1:v[0-9]+]], off, s[0:3], s5 offset:16{{$}}
+; GCN: v_add_i32_e32 [[ADD1:v[0-9]+]], vcc, 2, [[LOAD1]]
+
+; GCN: s_swappc_b64
+
+; GCN: buffer_store_dword [[ADD1]], off, s[0:3], s5 offset:16{{$}}
+
+; GCN: v_readlane_b32
+; GCN: buffer_load_dword v32,
+; GCN: s_sub_u32 s32, s32, 0x900{{$}}
+; GCN: s_setpc_b64
+define void  @void_func_byval_struct_non_leaf(%struct.ByValStruct* byval noalias nocapture align 4 %arg0, %struct.ByValStruct* byval noalias nocapture align 4 %arg1) #1 {
+entry:
+  %arrayidx = getelementptr inbounds %struct.ByValStruct, %struct.ByValStruct* %arg0, i32 0, i32 0, i32 0
+  %tmp = load volatile i32, i32* %arrayidx, align 4
+  %add = add nsw i32 %tmp, 1
+  store volatile i32 %add, i32* %arrayidx, align 4
+  %arrayidx2 = getelementptr inbounds %struct.ByValStruct, %struct.ByValStruct* %arg1, i32 0, i32 0, i32 0
+  %tmp1 = load volatile i32, i32* %arrayidx2, align 4
+  %add3 = add nsw i32 %tmp1, 2
+  call void @external_void_func_void()
+  store volatile i32 %add3, i32* %arrayidx2, align 4
+  store volatile i32 9, i32 addrspace(1)* null, align 4
+  ret void
+}
+
+; GCN-LABEL: {{^}}call_void_func_byval_struct_func:
+; GCN: s_mov_b32 s5, s32
+; GCN: s_add_u32 s32, s32, 0xa00{{$}}
+; GCN: v_writelane_b32
+
+; GCN-DAG: s_add_u32 s32, s32, 0x800{{$}}
+; GCN-DAG: v_mov_b32_e32 [[NINE:v[0-9]+]], 9
+; GCN-DAG: v_mov_b32_e32 [[THIRTEEN:v[0-9]+]], 13
+
+; VI-DAG: v_lshrrev_b32_e64 v{{[0-9]+}}, 6
+; CI-DAG: v_lshr_b32_e64 v{{[0-9]+}}, s{{[0-9]+}}, 6
+
+; GCN-DAG: v_add_i32_e64 [[FI_ADD0:v[0-9]+]], {{s\[[0-9]+:[0-9]+\]}}, 8,
+; GCN-DAG: v_or_b32_e32 [[FI_OR0:v[0-9]+]], 4, [[FI_ADD0]]
+
+; GCN-DAG: buffer_store_dword [[NINE]], off, s[0:3], s5 offset:8
+; GCN-DAG: buffer_store_dword [[THIRTEEN]], off, s[0:3], s5 offset:24
+
+; GCN-DAG: buffer_load_dword [[LOAD0:v[0-9]+]], [[FI_OR0]], s[0:3], s4 offen offset:4
+; GCN-DAG: buffer_load_dword [[LOAD1:v[0-9]+]], [[FI_OR0]], s[0:3], s4 offen offset:8
+
+; FIXME: or fails to combine with add, so FI doesn't fold and scratch wave offset is used
+; VI-DAG: v_lshrrev_b32_e64 v{{[0-9]+}}, 6
+; CI-DAG: v_lshr_b32_e64 v{{[0-9]+}}, s{{[0-9]+}}, 6
+
+; GCN-DAG: v_add_i32_e64 [[FI_ADD1:v[0-9]+]], s{{\[[0-9]+:[0-9]+\]}}, 24,
+; GCN-DAG: v_or_b32_e32 [[FI_OR1:v[0-9]+]], 4, [[FI_ADD1]]
+
+; GCN-DAG: buffer_store_dword [[LOAD0]], off, s[0:3], s32 offset:8
+; GCN-DAG: buffer_store_dword [[LOAD1]], off, s[0:3], s32 offset:12
+
+
+
+; GCN-DAG: buffer_load_dword [[LOAD2:v[0-9]+]], off, s[0:3], s5 offset:8
+; GCN-DAG: buffer_load_dword [[LOAD3:v[0-9]+]], off, s[0:3], s5 offset:12
+; GCN-DAG: buffer_store_dword [[LOAD2]], off, s[0:3], s32{{$}}
+; GCN-DAG: buffer_store_dword [[LOAD3]], off, s[0:3], s32 offset:4
+
+
+; GCN: buffer_load_dword [[LOAD4:v[0-9]+]], [[FI_OR1]], s[0:3], s4 offen offset:4
+; GCN: buffer_load_dword [[LOAD5:v[0-9]+]], [[FI_OR1]], s[0:3], s4 offen offset:8
+; GCN: buffer_store_dword [[LOAD5]], off, s[0:3], s32 offset:28
+; GCN: buffer_store_dword [[LOAD4]], off, s[0:3], s32 offset:24
+
+
+; GCN-DAG: buffer_load_dword [[LOAD6:v[0-9]+]], off, s[0:3], s5 offset:24
+; GCN-DAG: buffer_load_dword [[LOAD7:v[0-9]+]], off, s[0:3], s5 offset:28
+; GCN-DAG: buffer_store_dword [[LOAD6]], off, s[0:3], s32 offset:16
+; GCN-DAG: buffer_store_dword [[LOAD7]], off, s[0:3], s32 offset:20
+
+; GCN: s_swappc_b64
+; GCN-NEXT: s_sub_u32 s32, s32, 0x800{{$}}
+
+; GCN: v_readlane_b32
+
+; GCN: s_sub_u32 s32, s32, 0xa00{{$}}
+; GCN-NEXT: s_waitcnt
+; GCN-NEXT: s_setpc_b64
+define void @call_void_func_byval_struct_func() #0 {
+entry:
+  %arg0 = alloca %struct.ByValStruct, align 4
+  %arg1 = alloca %struct.ByValStruct, align 4
+  %tmp = bitcast %struct.ByValStruct* %arg0 to i8*
+  call void @llvm.lifetime.start.p0i8(i64 32, i8* %tmp)
+  %tmp1 = bitcast %struct.ByValStruct* %arg1 to i8*
+  call void @llvm.lifetime.start.p0i8(i64 32, i8* %tmp1)
+  %arrayidx = getelementptr inbounds %struct.ByValStruct, %struct.ByValStruct* %arg0, i32 0, i32 0, i32 0
+  store volatile i32 9, i32* %arrayidx, align 4
+  %arrayidx2 = getelementptr inbounds %struct.ByValStruct, %struct.ByValStruct* %arg1, i32 0, i32 0, i32 0
+  store volatile i32 13, i32* %arrayidx2, align 4
+  call void @void_func_byval_struct(%struct.ByValStruct* byval nonnull align 4 %arg0, %struct.ByValStruct* byval nonnull align 4 %arg1)
+  call void @llvm.lifetime.end.p0i8(i64 32, i8* %tmp1)
+  call void @llvm.lifetime.end.p0i8(i64 32, i8* %tmp)
+  ret void
+}
+
+; GCN-LABEL: {{^}}call_void_func_byval_struct_kernel:
+; GCN: s_mov_b32 s33, s7
+; GCN: s_add_u32 s32, s33, 0xa00{{$}}
+
+; GCN-DAG: v_mov_b32_e32 [[NINE:v[0-9]+]], 9
+; GCN-DAG: v_mov_b32_e32 [[THIRTEEN:v[0-9]+]], 13
+; GCN-DAG: buffer_store_dword [[NINE]], off, s[0:3], s33 offset:8
+; GCN: buffer_store_dword [[THIRTEEN]], off, s[0:3], s33 offset:24
+
+; GCN-DAG: s_add_u32 s32, s32, 0x800{{$}}
+
+; FIXME: Fold offset
+; GCN-DAG: v_or_b32_e32 [[OR_FI0:v[0-9]+]], 4,
+
+; GCN: buffer_load_dword [[LOAD0:v[0-9]+]], [[OR_FI0]], s[0:3], s33 offen offset:4
+; GCN: buffer_load_dword [[LOAD1:v[0-9]+]], [[OR_FI0]], s[0:3], s33 offen offset:8
+
+; FIXME: Fold offset
+; GCN-DAG: v_or_b32_e32 [[OR_FI1:v[0-9]+]], 4,
+
+; GCN: buffer_store_dword [[LOAD1]], off, s[0:3], s32 offset:12
+; GCN: buffer_store_dword [[LOAD0]], off, s[0:3], s32 offset:8
+
+
+; GCN-DAG: buffer_load_dword [[LOAD2:v[0-9]+]], off, s[0:3], s33 offset:8
+; GCN-DAG: buffer_load_dword [[LOAD3:v[0-9]+]], off, s[0:3], s33 offset:12
+; GCN: buffer_store_dword [[LOAD3]], off, s[0:3], s32 offset:4
+; GCN: buffer_store_dword [[LOAD2]], off, s[0:3], s32{{$}}
+
+
+
+; GCN: buffer_load_dword [[LOAD4:v[0-9]+]], [[OR_FI1]], s[0:3], s33 offen offset:4
+; GCN: buffer_load_dword [[LOAD5:v[0-9]+]], [[OR_FI1]], s[0:3], s33 offen offset:8
+; GCN: buffer_store_dword [[LOAD5]], off, s[0:3], s32 offset:28
+; GCN: buffer_store_dword [[LOAD4]], off, s[0:3], s32 offset:24
+
+
+; GCN-DAG: buffer_load_dword [[LOAD6:v[0-9]+]], off, s[0:3], s33 offset:24
+; GCN-DAG: buffer_load_dword [[LOAD7:v[0-9]+]], off, s[0:3], s33 offset:28
+; GCN-DAG: buffer_store_dword [[LOAD6]], off, s[0:3], s32 offset:16
+; GCN-DAG: buffer_store_dword [[LOAD7]], off, s[0:3], s32 offset:20
+
+
+; GCN: s_swappc_b64
+; FIXME: Dead SP modfication
+; GCN-NEXT: s_sub_u32 s32, s32, 0x800{{$}}
+; GCN-NEXT: s_endpgm
+define amdgpu_kernel void @call_void_func_byval_struct_kernel() #0 {
+entry:
+  %arg0 = alloca %struct.ByValStruct, align 4
+  %arg1 = alloca %struct.ByValStruct, align 4
+  %tmp = bitcast %struct.ByValStruct* %arg0 to i8*
+  call void @llvm.lifetime.start.p0i8(i64 32, i8* %tmp)
+  %tmp1 = bitcast %struct.ByValStruct* %arg1 to i8*
+  call void @llvm.lifetime.start.p0i8(i64 32, i8* %tmp1)
+  %arrayidx = getelementptr inbounds %struct.ByValStruct, %struct.ByValStruct* %arg0, i32 0, i32 0, i32 0
+  store volatile i32 9, i32* %arrayidx, align 4
+  %arrayidx2 = getelementptr inbounds %struct.ByValStruct, %struct.ByValStruct* %arg1, i32 0, i32 0, i32 0
+  store volatile i32 13, i32* %arrayidx2, align 4
+  call void @void_func_byval_struct(%struct.ByValStruct* byval nonnull align 4 %arg0, %struct.ByValStruct* byval nonnull align 4 %arg1)
+  call void @llvm.lifetime.end.p0i8(i64 32, i8* %tmp1)
+  call void @llvm.lifetime.end.p0i8(i64 32, i8* %tmp)
+  ret void
+}
+
+; GCN-LABEL: {{^}}call_void_func_byval_struct_kernel_no_frame_pointer_elim:
+define amdgpu_kernel void @call_void_func_byval_struct_kernel_no_frame_pointer_elim() #2 {
+entry:
+  %arg0 = alloca %struct.ByValStruct, align 4
+  %arg1 = alloca %struct.ByValStruct, align 4
+  %tmp = bitcast %struct.ByValStruct* %arg0 to i8*
+  call void @llvm.lifetime.start.p0i8(i64 32, i8* %tmp)
+  %tmp1 = bitcast %struct.ByValStruct* %arg1 to i8*
+  call void @llvm.lifetime.start.p0i8(i64 32, i8* %tmp1)
+  %arrayidx = getelementptr inbounds %struct.ByValStruct, %struct.ByValStruct* %arg0, i32 0, i32 0, i32 0
+  store volatile i32 9, i32* %arrayidx, align 4
+  %arrayidx2 = getelementptr inbounds %struct.ByValStruct, %struct.ByValStruct* %arg1, i32 0, i32 0, i32 0
+  store volatile i32 13, i32* %arrayidx2, align 4
+  call void @void_func_byval_struct(%struct.ByValStruct* byval nonnull align 4 %arg0, %struct.ByValStruct* byval nonnull align 4 %arg1)
+  call void @llvm.lifetime.end.p0i8(i64 32, i8* %tmp1)
+  call void @llvm.lifetime.end.p0i8(i64 32, i8* %tmp)
+  ret void
+}
+
+declare void @external_void_func_void() #0
+
+declare void @llvm.lifetime.start.p0i8(i64, i8* nocapture) #3
+declare void @llvm.lifetime.end.p0i8(i64, i8* nocapture) #3
+
+attributes #0 = { nounwind }
+attributes #1 = { noinline norecurse nounwind }
+attributes #2 = { nounwind norecurse "no-frame-pointer-elim"="true" }
diff --git a/llvm/test/CodeGen/AMDGPU/call-argument-types.ll b/llvm/test/CodeGen/AMDGPU/call-argument-types.ll
new file mode 100644
index 00000000000..0411de9673a
--- /dev/null
+++ b/llvm/test/CodeGen/AMDGPU/call-argument-types.ll
@@ -0,0 +1,527 @@
+; RUN: llc -march=amdgcn -mcpu=fiji -mattr=-flat-for-global -amdgpu-function-calls -amdgpu-scalarize-global-loads=0 -verify-machineinstrs < %s | FileCheck -enable-var-scope -check-prefixes=GCN,VI,MESA %s
+; RUN: llc -march=amdgcn -mcpu=hawaii -amdgpu-function-calls -amdgpu-scalarize-global-loads=0 -verify-machineinstrs < %s | FileCheck -enable-var-scope -check-prefixes=GCN,CI,MESA %s
+; RUN: llc -march=amdgcn -mcpu=gfx900 -mattr=-flat-for-global -amdgpu-function-calls -amdgpu-scalarize-global-loads=0 -verify-machineinstrs < %s | FileCheck -enable-var-scope -check-prefixes=GCN,GFX9,VI,MESA %s
+; RUN: llc -mtriple=amdgcn-amd-amdhsa -mcpu=fiji -mattr=-flat-for-global -amdgpu-function-calls -amdgpu-scalarize-global-loads=0 -verify-machineinstrs < %s | FileCheck -enable-var-scope -check-prefixes=GCN,VI,HSA %s
+
+declare void @external_void_func_i1(i1) #0
+declare void @external_void_func_i1_signext(i1 signext) #0
+declare void @external_void_func_i1_zeroext(i1 zeroext) #0
+
+declare void @external_void_func_i8(i8) #0
+declare void @external_void_func_i8_signext(i8 signext) #0
+declare void @external_void_func_i8_zeroext(i8 zeroext) #0
+
+declare void @external_void_func_i16(i16) #0
+declare void @external_void_func_i16_signext(i16 signext) #0
+declare void @external_void_func_i16_zeroext(i16 zeroext) #0
+
+declare void @external_void_func_i32(i32) #0
+declare void @external_void_func_i64(i64) #0
+
+declare void @external_void_func_f16(half) #0
+declare void @external_void_func_f32(float) #0
+declare void @external_void_func_f64(double) #0
+
+declare void @external_void_func_v2i16(<2 x i16>) #0
+declare void @external_void_func_v2f16(<2 x half>) #0
+
+declare void @external_void_func_v2i32(<2 x i32>) #0
+declare void @external_void_func_v3i32(<3 x i32>) #0
+declare void @external_void_func_v4i32(<4 x i32>) #0
+declare void @external_void_func_v8i32(<8 x i32>) #0
+declare void @external_void_func_v16i32(<16 x i32>) #0
+declare void @external_void_func_v32i32(<32 x i32>) #0
+declare void @external_void_func_v32i32_i32(<32 x i32>, i32) #0
+
+; return value and argument
+declare i32 @external_i32_func_i32(i32) #0
+
+; Structs
+declare void @external_void_func_struct_i8_i32({ i8, i32 }) #0
+declare void @external_void_func_byval_struct_i8_i32({ i8, i32 }* byval) #0
+declare void @external_void_func_sret_struct_i8_i32_byval_struct_i8_i32({ i8, i32 }* sret, { i8, i32 }* byval) #0
+
+declare void @external_void_func_v16i8(<16 x i8>) #0
+
+
+; FIXME: Should be passing -1
+; GCN-LABEL: {{^}}test_call_external_void_func_i1_imm:
+; MESA: s_mov_b32 s36, SCRATCH_RSRC_DWORD
+
+; MESA-DAG: s_mov_b64 s[0:1], s[36:37]
+
+; GCN: s_getpc_b64 s{{\[}}[[PC_LO:[0-9]+]]:[[PC_HI:[0-9]+]]{{\]}}
+; GCN-NEXT: s_add_u32 s[[PC_LO]], s[[PC_LO]], external_void_func_i1@rel32@lo+4
+; GCN-NEXT: s_addc_u32 s[[PC_HI]], s[[PC_HI]], external_void_func_i1@rel32@hi+4
+; GCN-DAG: v_mov_b32_e32 v0, 1{{$}}
+; MESA-DAG: s_mov_b64 s[2:3], s[38:39]
+
+; GCN: s_swappc_b64 s[30:31], s{{\[}}[[PC_LO]]:[[PC_HI]]{{\]}}
+; GCN-NEXT: s_endpgm
+define amdgpu_kernel void @test_call_external_void_func_i1_imm() #0 {
+  call void @external_void_func_i1(i1 true)
+  ret void
+}
+
+; GCN-LABEL: {{^}}test_call_external_void_func_i1_signext:
+; MESA: s_mov_b32 s33, s3{{$}}
+; HSA: s_mov_b32 s33, s9{{$}}
+
+; GCN: s_getpc_b64 s{{\[}}[[PC_LO:[0-9]+]]:[[PC_HI:[0-9]+]]{{\]}}
+; GCN-NEXT: s_add_u32 s[[PC_LO]], s[[PC_LO]], external_void_func_i1_signext@rel32@lo+4
+; GCN-NEXT: s_addc_u32 s[[PC_HI]], s[[PC_HI]], external_void_func_i1_signext@rel32@hi+4
+; GCN-NEXT: buffer_load_ubyte [[VAR:v[0-9]+]]
+; HSA-NEXT: s_mov_b32 s4, s33
+; HSA-NEXT: s_mov_b32 s32, s33
+
+; MESA-DAG: s_mov_b32 s4, s33{{$}}
+; MESA-DAG: s_mov_b32 s32, s33{{$}}
+
+; GCN: s_waitcnt vmcnt(0)
+; GCN-NEXT: v_bfe_i32 v0, v0, 0, 1
+; GCN-NEXT: s_swappc_b64 s[30:31], s{{\[}}[[PC_LO]]:[[PC_HI]]{{\]}}
+; GCN-NEXT: s_endpgm
+define amdgpu_kernel void @test_call_external_void_func_i1_signext(i32) #0 {
+  %var = load volatile i1, i1 addrspace(1)* undef
+  call void @external_void_func_i1_signext(i1 %var)
+  ret void
+}
+
+; FIXME: load should be scheduled before getpc
+; GCN-LABEL: {{^}}test_call_external_void_func_i1_zeroext:
+; MESA: s_mov_b32 s33, s3{{$}}
+
+; GCN: s_getpc_b64 s{{\[}}[[PC_LO:[0-9]+]]:[[PC_HI:[0-9]+]]{{\]}}
+; GCN-NEXT: s_add_u32 s[[PC_LO]], s[[PC_LO]], external_void_func_i1_zeroext@rel32@lo+4
+; GCN-NEXT: s_addc_u32 s[[PC_HI]], s[[PC_HI]], external_void_func_i1_zeroext@rel32@hi+4
+; GCN-NEXT: buffer_load_ubyte v0
+
+; GCN-DAG: s_mov_b32 s4, s33{{$}}
+; GCN-DAG: s_mov_b32 s32, s33{{$}}
+
+; GCN: s_waitcnt vmcnt(0)
+; GCN-NEXT: v_and_b32_e32 v0, 1, v0
+; GCN-NEXT: s_swappc_b64 s[30:31], s{{\[}}[[PC_LO]]:[[PC_HI]]{{\]}}
+; GCN-NEXT: s_endpgm
+define amdgpu_kernel void @test_call_external_void_func_i1_zeroext(i32) #0 {
+  %var = load volatile i1, i1 addrspace(1)* undef
+  call void @external_void_func_i1_zeroext(i1 %var)
+  ret void
+}
+
+; GCN-LABEL: {{^}}test_call_external_void_func_i8_imm:
+; MESA-DAG: s_mov_b32 s33, s3{{$}}
+
+; GCN: s_getpc_b64 s{{\[}}[[PC_LO:[0-9]+]]:[[PC_HI:[0-9]+]]{{\]}}
+; GCN-NEXT: s_add_u32 s[[PC_LO]], s[[PC_LO]], external_void_func_i8@rel32@lo+4
+; GCN-NEXT: s_addc_u32 s[[PC_HI]], s[[PC_HI]], external_void_func_i8@rel32@hi+4
+; GCN-NEXT: v_mov_b32_e32 v0, 0x7b
+
+; HSA-DAG: s_mov_b32 s4, s33{{$}}
+; GCN-DAG: s_mov_b32 s32, s33{{$}}
+
+; GCN: s_swappc_b64 s[30:31], s{{\[}}[[PC_LO]]:[[PC_HI]]{{\]}}
+; GCN-NEXT: s_endpgm
+define amdgpu_kernel void @test_call_external_void_func_i8_imm(i32) #0 {
+  call void @external_void_func_i8(i8 123)
+  ret void
+}
+
+; FIXME: don't wait before call
+; GCN-LABEL: {{^}}test_call_external_void_func_i8_signext:
+; HSA-DAG: s_mov_b32 s33, s9{{$}}
+; MESA-DAG: s_mov_b32 s33, s3{{$}}
+
+; GCN-DAG: buffer_load_sbyte v0
+; GCN: s_getpc_b64 s{{\[}}[[PC_LO:[0-9]+]]:[[PC_HI:[0-9]+]]{{\]}}
+; GCN-NEXT: s_add_u32 s[[PC_LO]], s[[PC_LO]], external_void_func_i8_signext@rel32@lo+4
+; GCN-NEXT: s_addc_u32 s[[PC_HI]], s[[PC_HI]], external_void_func_i8_signext@rel32@hi+4
+
+; GCN-DAG: s_mov_b32 s4, s33
+; GCN-DAG: s_mov_b32 s32, s3
+
+; GCN: s_waitcnt vmcnt(0)
+; GCN-NEXT: s_swappc_b64 s[30:31], s{{\[}}[[PC_LO]]:[[PC_HI]]{{\]}}
+; GCN-NEXT: s_endpgm
+define amdgpu_kernel void @test_call_external_void_func_i8_signext(i32) #0 {
+  %var = load volatile i8, i8 addrspace(1)* undef
+  call void @external_void_func_i8_signext(i8 %var)
+  ret void
+}
+
+; GCN-LABEL: {{^}}test_call_external_void_func_i8_zeroext:
+; MESA-DAG: s_mov_b32 s33, s3{{$}}
+; HSA-DAG: s_mov_b32 s33, s9{{$}}
+
+; GCN-DAG: buffer_load_ubyte v0
+; GCN: s_getpc_b64 s{{\[}}[[PC_LO:[0-9]+]]:[[PC_HI:[0-9]+]]{{\]}}
+; GCN-NEXT: s_add_u32 s[[PC_LO]], s[[PC_LO]], external_void_func_i8_zeroext@rel32@lo+4
+; GCN-NEXT: s_addc_u32 s[[PC_HI]], s[[PC_HI]], external_void_func_i8_zeroext@rel32@hi+4
+
+; GCN-DAG: s_mov_b32 s4, s33
+; GCN-DAG: s_mov_b32 s32, s33
+
+; GCN: s_waitcnt vmcnt(0)
+; GCN-NEXT: s_swappc_b64 s[30:31], s{{\[}}[[PC_LO]]:[[PC_HI]]{{\]}}
+; GCN-NEXT: s_endpgm
+define amdgpu_kernel void @test_call_external_void_func_i8_zeroext(i32) #0 {
+  %var = load volatile i8, i8 addrspace(1)* undef
+  call void @external_void_func_i8_zeroext(i8 %var)
+  ret void
+}
+
+; GCN-LABEL: {{^}}test_call_external_void_func_i16_imm:
+; GCN-DAG: v_mov_b32_e32 v0, 0x7b{{$}}
+
+; GCN-DAG: s_mov_b32 s4, s33
+; GCN-DAG: s_mov_b32 s32, s33
+
+; GCN: s_swappc_b64
+define amdgpu_kernel void @test_call_external_void_func_i16_imm() #0 {
+  call void @external_void_func_i16(i16 123)
+  ret void
+}
+
+; GCN-LABEL: {{^}}test_call_external_void_func_i16_signext:
+; MESA-DAG: s_mov_b32 s33, s3{{$}}
+
+; GCN-DAG: buffer_load_sshort v0
+; GCN: s_getpc_b64 s{{\[}}[[PC_LO:[0-9]+]]:[[PC_HI:[0-9]+]]{{\]}}
+; GCN-NEXT: s_add_u32 s[[PC_LO]], s[[PC_LO]], external_void_func_i16_signext@rel32@lo+4
+; GCN-NEXT: s_addc_u32 s[[PC_HI]], s[[PC_HI]], external_void_func_i16_signext@rel32@hi+4
+
+; GCN-DAG: s_mov_b32 s4, s33
+; GCN-DAG: s_mov_b32 s32, s33
+
+; GCN: s_waitcnt vmcnt(0)
+; GCN-NEXT: s_swappc_b64 s[30:31], s{{\[}}[[PC_LO]]:[[PC_HI]]{{\]}}
+; GCN-NEXT: s_endpgm
+define amdgpu_kernel void @test_call_external_void_func_i16_signext(i32) #0 {
+  %var = load volatile i16, i16 addrspace(1)* undef
+  call void @external_void_func_i16_signext(i16 %var)
+  ret void
+}
+
+; GCN-LABEL: {{^}}test_call_external_void_func_i16_zeroext:
+; MESA-DAG: s_mov_b32 s33, s3{{$}}
+
+
+; GCN-DAG: buffer_load_ushort v0
+; GCN: s_getpc_b64 s{{\[}}[[PC_LO:[0-9]+]]:[[PC_HI:[0-9]+]]{{\]}}
+; GCN-NEXT: s_add_u32 s[[PC_LO]], s[[PC_LO]], external_void_func_i16_zeroext@rel32@lo+4
+; GCN-NEXT: s_addc_u32 s[[PC_HI]], s[[PC_HI]], external_void_func_i16_zeroext@rel32@hi+4
+
+; GCN-DAG: s_mov_b32 s4, s33
+; GCN-DAG: s_mov_b32 s32, s33
+
+; GCN: s_waitcnt vmcnt(0)
+; GCN-NEXT: s_swappc_b64 s[30:31], s{{\[}}[[PC_LO]]:[[PC_HI]]{{\]}}
+; GCN-NEXT: s_endpgm
+define amdgpu_kernel void @test_call_external_void_func_i16_zeroext(i32) #0 {
+  %var = load volatile i16, i16 addrspace(1)* undef
+  call void @external_void_func_i16_zeroext(i16 %var)
+  ret void
+}
+
+; GCN-LABEL: {{^}}test_call_external_void_func_i32_imm:
+; MESA-DAG: s_mov_b32 s33, s3{{$}}
+
+; GCN: s_getpc_b64 s{{\[}}[[PC_LO:[0-9]+]]:[[PC_HI:[0-9]+]]{{\]}}
+; GCN-NEXT: s_add_u32 s[[PC_LO]], s[[PC_LO]], external_void_func_i32@rel32@lo+4
+; GCN-NEXT: s_addc_u32 s[[PC_HI]], s[[PC_HI]], external_void_func_i32@rel32@hi+4
+; GCN: v_mov_b32_e32 v0, 42
+; GCN-DAG: s_mov_b32 s4, s33
+; GCN-DAG: s_mov_b32 s32, s33
+
+; GCN: s_swappc_b64 s[30:31], s{{\[}}[[PC_LO]]:[[PC_HI]]{{\]}}
+; GCN-NEXT: s_endpgm
+define amdgpu_kernel void @test_call_external_void_func_i32_imm(i32) #0 {
+  call void @external_void_func_i32(i32 42)
+  ret void
+}
+
+; GCN-LABEL: {{^}}test_call_external_void_func_i64_imm:
+; GCN-DAG: s_movk_i32 [[K0:s[0-9]+]], 0x7b{{$}}
+; GCN-DAG: s_mov_b32 [[K1:s[0-9]+]], 0{{$}}
+; GCN-DAG: v_mov_b32_e32 v0, [[K0]]
+; GCN-DAG: s_getpc_b64 s{{\[}}[[PC_LO:[0-9]+]]:[[PC_HI:[0-9]+]]{{\]}}
+; GCN-DAG: s_add_u32 s[[PC_LO]], s[[PC_LO]], external_void_func_i64@rel32@lo+4
+; GCN-DAG: s_addc_u32 s[[PC_HI]], s[[PC_HI]], external_void_func_i64@rel32@hi+4
+; GCN-DAG: v_mov_b32_e32 v1, [[K1]]
+; GCN: s_swappc_b64 s[30:31], s{{\[}}[[PC_LO]]:[[PC_HI]]{{\]}}
+; GCN-NEXT: s_endpgm
+define amdgpu_kernel void @test_call_external_void_func_i64_imm() #0 {
+  call void @external_void_func_i64(i64 123)
+  ret void
+}
+
+; GCN-LABEL: {{^}}test_call_external_void_func_f16_imm:
+; VI: v_mov_b32_e32 v0, 0x4400
+; CI: v_mov_b32_e32 v0, 4.0
+; GCN-NOT: v0
+; GCN: s_swappc_b64
+define amdgpu_kernel void @test_call_external_void_func_f16_imm() #0 {
+  call void @external_void_func_f16(half 4.0)
+  ret void
+}
+
+; GCN-LABEL: {{^}}test_call_external_void_func_f32_imm:
+; GCN: v_mov_b32_e32 v0, 4.0
+; GCN-NOT: v0
+; GCN: s_swappc_b64
+define amdgpu_kernel void @test_call_external_void_func_f32_imm() #0 {
+  call void @external_void_func_f32(float 4.0)
+  ret void
+}
+
+; GCN-LABEL: {{^}}test_call_external_void_func_f64_imm:
+; GCN: v_mov_b32_e32 v0, 0{{$}}
+; GCN: v_mov_b32_e32 v1, 0x40100000
+; GCN: s_swappc_b64
+define amdgpu_kernel void @test_call_external_void_func_f64_imm() #0 {
+  call void @external_void_func_f64(double 4.0)
+  ret void
+}
+
+; GCN-LABEL: {{^}}test_call_external_void_func_v2i16:
+; GFX9: buffer_load_dword v0
+; GFX9-NOT: v0
+; GFX9: s_swappc_b64
+define amdgpu_kernel void @test_call_external_void_func_v2i16() #0 {
+  %val = load <2 x i16>, <2 x i16> addrspace(1)* undef
+  call void @external_void_func_v2i16(<2 x i16> %val)
+  ret void
+}
+
+; GCN-LABEL: {{^}}test_call_external_void_func_v2f16:
+; GFX9: buffer_load_dword v0
+; GFX9-NOT: v0
+; GFX9: s_swappc_b64
+define amdgpu_kernel void @test_call_external_void_func_v2f16() #0 {
+  %val = load <2 x half>, <2 x half> addrspace(1)* undef
+  call void @external_void_func_v2f16(<2 x half> %val)
+  ret void
+}
+
+; GCN-LABEL: {{^}}test_call_external_void_func_v2i32:
+; GCN: buffer_load_dwordx2 v[0:1]
+; GCN: s_waitcnt
+; GCN-NEXT: s_swappc_b64
+define amdgpu_kernel void @test_call_external_void_func_v2i32() #0 {
+  %val = load <2 x i32>, <2 x i32> addrspace(1)* undef
+  call void @external_void_func_v2i32(<2 x i32> %val)
+  ret void
+}
+
+; FIXME: Passing 4th
+; GCN-LABEL: {{^}}test_call_external_void_func_v3i32_imm:
+; HSA-DAG: s_mov_b32 s33, s9
+; MESA-DAG: s_mov_b32 s33, s3{{$}}
+
+; GCN-DAG: v_mov_b32_e32 v0
+; GCN-DAG: v_mov_b32_e32 v1
+; GCN-DAG: v_mov_b32_e32 v2
+; GCN-DAG: v_mov_b32_e32 v3
+
+; GCN: s_swappc_b64
+define amdgpu_kernel void @test_call_external_void_func_v3i32_imm(i32) #0 {
+  call void @external_void_func_v3i32(<3 x i32> <i32 3, i32 4, i32 5>)
+  ret void
+}
+
+; GCN-LABEL: {{^}}test_call_external_void_func_v4i32:
+; GCN: buffer_load_dwordx4 v[0:3]
+; GCN: s_waitcnt
+; GCN-NEXT: s_swappc_b64
+define amdgpu_kernel void @test_call_external_void_func_v4i32() #0 {
+  %val = load <4 x i32>, <4 x i32> addrspace(1)* undef
+  call void @external_void_func_v4i32(<4 x i32> %val)
+  ret void
+}
+
+; GCN-LABEL: {{^}}test_call_external_void_func_v8i32:
+; GCN-DAG: buffer_load_dwordx4 v[0:3], off
+; GCN-DAG: buffer_load_dwordx4 v[4:7], off
+; GCN: s_waitcnt
+; GCN-NEXT: s_swappc_b64
+define amdgpu_kernel void @test_call_external_void_func_v8i32() #0 {
+  %ptr = load <8 x i32> addrspace(1)*, <8 x i32> addrspace(1)* addrspace(2)* undef
+  %val = load <8 x i32>, <8 x i32> addrspace(1)* %ptr
+  call void @external_void_func_v8i32(<8 x i32> %val)
+  ret void
+}
+
+; GCN-LABEL: {{^}}test_call_external_void_func_v16i32:
+; GCN-DAG: buffer_load_dwordx4 v[0:3], off
+; GCN-DAG: buffer_load_dwordx4 v[4:7], off
+; GCN-DAG: buffer_load_dwordx4 v[8:11], off
+; GCN-DAG: buffer_load_dwordx4 v[12:15], off
+; GCN: s_waitcnt
+; GCN-NEXT: s_swappc_b64
+define amdgpu_kernel void @test_call_external_void_func_v16i32() #0 {
+  %ptr = load <16 x i32> addrspace(1)*, <16 x i32> addrspace(1)* addrspace(2)* undef
+  %val = load <16 x i32>, <16 x i32> addrspace(1)* %ptr
+  call void @external_void_func_v16i32(<16 x i32> %val)
+  ret void
+}
+
+; GCN-LABEL: {{^}}test_call_external_void_func_v32i32:
+; GCN-DAG: buffer_load_dwordx4 v[0:3], off
+; GCN-DAG: buffer_load_dwordx4 v[4:7], off
+; GCN-DAG: buffer_load_dwordx4 v[8:11], off
+; GCN-DAG: buffer_load_dwordx4 v[12:15], off
+; GCN-DAG: buffer_load_dwordx4 v[16:19], off
+; GCN-DAG: buffer_load_dwordx4 v[20:23], off
+; GCN-DAG: buffer_load_dwordx4 v[24:27], off
+; GCN-DAG: buffer_load_dwordx4 v[28:31], off
+; GCN: s_waitcnt
+; GCN-NEXT: s_swappc_b64
+define amdgpu_kernel void @test_call_external_void_func_v32i32() #0 {
+  %ptr = load <32 x i32> addrspace(1)*, <32 x i32> addrspace(1)* addrspace(2)* undef
+  %val = load <32 x i32>, <32 x i32> addrspace(1)* %ptr
+  call void @external_void_func_v32i32(<32 x i32> %val)
+  ret void
+}
+
+; GCN-LABEL: {{^}}test_call_external_void_func_v32i32_i32:
+; HSA-DAG: s_mov_b32 s33, s9
+; HSA-DAG: s_add_u32 [[SP_REG:s[0-9]+]], s33, 0x100{{$}}
+
+; MESA-DAG: s_mov_b32 s33, s3{{$}}
+; MESA-DAG: s_add_u32 [[SP_REG:s[0-9]+]], s33, 0x100{{$}}
+
+; GCN-DAG: buffer_load_dword [[VAL1:v[0-9]+]], off, s[{{[0-9]+}}:{{[0-9]+}}], 0{{$}}
+; GCN-DAG: buffer_load_dwordx4 v[0:3], off
+; GCN-DAG: buffer_load_dwordx4 v[4:7], off
+; GCN-DAG: buffer_load_dwordx4 v[8:11], off
+; GCN-DAG: buffer_load_dwordx4 v[12:15], off
+; GCN-DAG: buffer_load_dwordx4 v[16:19], off
+; GCN-DAG: buffer_load_dwordx4 v[20:23], off
+; GCN-DAG: buffer_load_dwordx4 v[24:27], off
+; GCN-DAG: buffer_load_dwordx4 v[28:31], off
+
+; GCN: buffer_store_dword [[VAL1]], off, s[{{[0-9]+}}:{{[0-9]+}}], [[SP_REG]]{{$}}
+; GCN: s_waitcnt
+; GCN-NEXT: s_swappc_b64
+; GCN-NEXT: s_endpgm
+define amdgpu_kernel void @test_call_external_void_func_v32i32_i32(i32) #0 {
+  %ptr0 = load <32 x i32> addrspace(1)*, <32 x i32> addrspace(1)* addrspace(2)* undef
+  %val0 = load <32 x i32>, <32 x i32> addrspace(1)* %ptr0
+  %val1 = load i32, i32 addrspace(1)* undef
+  call void @external_void_func_v32i32_i32(<32 x i32> %val0, i32 %val1)
+  ret void
+}
+
+; FIXME: No wait after call
+; GCN-LABEL: {{^}}test_call_external_i32_func_i32_imm:
+; GCN: v_mov_b32_e32 v0, 42
+; GCN: s_swappc_b64 s[30:31],
+; GCN-NEXT: s_waitcnt lgkmcnt(0)
+; GCN-NEXT: buffer_store_dword v0, off, s[36:39], 0
+define amdgpu_kernel void @test_call_external_i32_func_i32_imm(i32 addrspace(1)* %out) #0 {
+  %val = call i32 @external_i32_func_i32(i32 42)
+  store volatile i32 %val, i32 addrspace(1)* %out
+  ret void
+}
+
+; GCN-LABEL: {{^}}test_call_external_void_func_struct_i8_i32:
+; GCN: buffer_load_ubyte v0, off
+; GCN: buffer_load_dword v1, off
+; GCN: s_waitcnt vmcnt(0)
+; GCN-NEXT: s_swappc_b64
+define amdgpu_kernel void @test_call_external_void_func_struct_i8_i32() #0 {
+  %ptr0 = load { i8, i32 } addrspace(1)*, { i8, i32 } addrspace(1)* addrspace(2)* undef
+  %val = load { i8, i32 }, { i8, i32 } addrspace(1)* %ptr0
+  call void @external_void_func_struct_i8_i32({ i8, i32 } %val)
+  ret void
+}
+
+; GCN-LABEL: {{^}}test_call_external_void_func_byval_struct_i8_i32:
+; GCN-DAG: s_add_u32 [[SP:s[0-9]+]], s33, 0x400{{$}}
+
+; GCN-DAG: v_mov_b32_e32 [[VAL0:v[0-9]+]], 3
+; GCN-DAG: v_mov_b32_e32 [[VAL1:v[0-9]+]], 8
+; MESA-DAG: buffer_store_byte [[VAL0]], off, s[36:39], s33 offset:8
+; MESA-DAG: buffer_store_dword [[VAL1]], off, s[36:39], s33 offset:12
+
+; HSA-DAG: buffer_store_byte [[VAL0]], off, s[0:3], s33 offset:8
+; HSA-DAG: buffer_store_dword [[VAL1]], off, s[0:3], s33 offset:12
+
+; GCN: s_add_u32 [[SP]], [[SP]], 0x200
+
+; HSA: buffer_load_dword [[RELOAD_VAL0:v[0-9]+]], off, s[0:3], s33 offset:8
+; HSA: buffer_load_dword [[RELOAD_VAL1:v[0-9]+]], off, s[0:3], s33 offset:12
+
+; HSA: buffer_store_dword [[RELOAD_VAL1]], off, s[0:3], [[SP]] offset:4
+; HSA: buffer_store_dword [[RELOAD_VAL0]], off, s[0:3], [[SP]]{{$}}
+
+
+; MESA: buffer_load_dword [[RELOAD_VAL0:v[0-9]+]], off, s[36:39], s33 offset:8
+; MESA: buffer_load_dword [[RELOAD_VAL1:v[0-9]+]], off, s[36:39], s33 offset:12
+
+; MESA: buffer_store_dword [[RELOAD_VAL1]], off, s[36:39], [[SP]] offset:4
+; MESA: buffer_store_dword [[RELOAD_VAL0]], off, s[36:39], [[SP]]{{$}}
+
+; GCN-NEXT: s_swappc_b64
+; GCN-NEXT: s_sub_u32 [[SP]], [[SP]], 0x200
+define amdgpu_kernel void @test_call_external_void_func_byval_struct_i8_i32() #0 {
+  %val = alloca { i8, i32 }, align 4
+  %gep0 = getelementptr inbounds { i8, i32 }, { i8, i32 }* %val, i32 0, i32 0
+  %gep1 = getelementptr inbounds { i8, i32 }, { i8, i32 }* %val, i32 0, i32 1
+  store i8 3, i8* %gep0
+  store i32 8, i32* %gep1
+  call void @external_void_func_byval_struct_i8_i32({ i8, i32 }* %val)
+  ret void
+}
+
+; GCN-LABEL: {{^}}test_call_external_void_func_sret_struct_i8_i32_byval_struct_i8_i32:
+; MESA-DAG: s_add_u32 [[SP:s[0-9]+]], [[FP_REG:s[0-9]+]], 0x600{{$}}
+; HSA-DAG: s_add_u32 [[SP:s[0-9]+]], [[FP_REG:s[0-9]+]], 0x600{{$}}
+
+; GCN-DAG: v_mov_b32_e32 [[VAL0:v[0-9]+]], 3
+; GCN-DAG: v_mov_b32_e32 [[VAL1:v[0-9]+]], 8
+; GCN-DAG: buffer_store_byte [[VAL0]], off, s{{\[[0-9]+:[0-9]+\]}}, [[FP_REG]] offset:8
+; GCN-DAG: buffer_store_dword [[VAL1]], off, s{{\[[0-9]+:[0-9]+\]}}, [[FP_REG]] offset:12
+
+; GCN-DAG: buffer_load_dword [[RELOAD_VAL0:v[0-9]+]], off, s{{\[[0-9]+:[0-9]+\]}}, [[FP_REG]] offset:8
+; GCN-DAG: buffer_load_dword [[RELOAD_VAL1:v[0-9]+]], off, s{{\[[0-9]+:[0-9]+\]}}, [[FP_REG]] offset:12
+
+; GCN-DAG: s_add_u32 [[SP]], [[SP]], 0x200
+; GCN: buffer_store_dword [[RELOAD_VAL1]], off, s{{\[[0-9]+:[0-9]+\]}}, [[SP]] offset:4
+; GCN: buffer_store_dword [[RELOAD_VAL0]], off, s{{\[[0-9]+:[0-9]+\]}}, [[SP]]{{$}}
+; GCN-NEXT: s_swappc_b64
+; GCN-DAG: buffer_load_ubyte [[LOAD_OUT_VAL0:v[0-9]+]], off, s{{\[[0-9]+:[0-9]+\]}}, [[FP_REG]] offset:16
+; GCN-DAG: buffer_load_dword [[LOAD_OUT_VAL1:v[0-9]+]], off, s{{\[[0-9]+:[0-9]+\]}}, [[FP_REG]] offset:20
+; GCN: s_sub_u32 [[SP]], [[SP]], 0x200
+
+; GCN: buffer_store_byte [[LOAD_OUT_VAL0]], off
+; GCN: buffer_store_dword [[LOAD_OUT_VAL1]], off
+define amdgpu_kernel void @test_call_external_void_func_sret_struct_i8_i32_byval_struct_i8_i32(i32) #0 {
+  %in.val = alloca { i8, i32 }, align 4
+  %out.val = alloca { i8, i32 }, align 4
+  %in.gep0 = getelementptr inbounds { i8, i32 }, { i8, i32 }* %in.val, i32 0, i32 0
+  %in.gep1 = getelementptr inbounds { i8, i32 }, { i8, i32 }* %in.val, i32 0, i32 1
+  store i8 3, i8* %in.gep0
+  store i32 8, i32* %in.gep1
+  call void @external_void_func_sret_struct_i8_i32_byval_struct_i8_i32({ i8, i32 }* %out.val, { i8, i32 }* %in.val)
+  %out.gep0 = getelementptr inbounds { i8, i32 }, { i8, i32 }* %out.val, i32 0, i32 0
+  %out.gep1 = getelementptr inbounds { i8, i32 }, { i8, i32 }* %out.val, i32 0, i32 1
+  %out.val0 = load i8, i8* %out.gep0
+  %out.val1 = load i32, i32* %out.gep1
+
+  store volatile i8 %out.val0, i8 addrspace(1)* undef
+  store volatile i32 %out.val1, i32 addrspace(1)* undef
+  ret void
+}
+
+; GCN-LABEL: {{^}}test_call_external_void_func_v16i8:
+define amdgpu_kernel void @test_call_external_void_func_v16i8() #0 {
+  %ptr = load <16 x i8> addrspace(1)*, <16 x i8> addrspace(1)* addrspace(2)* undef
+  %val = load <16 x i8>, <16 x i8> addrspace(1)* %ptr
+  call void @external_void_func_v16i8(<16 x i8> %val)
+  ret void
+}
+
+attributes #0 = { nounwind }
+attributes #1 = { nounwind readnone }
+attributes #2 = { nounwind noinline }
diff --git a/llvm/test/CodeGen/AMDGPU/call-preserved-registers.ll b/llvm/test/CodeGen/AMDGPU/call-preserved-registers.ll
new file mode 100644
index 00000000000..8dac066786e
--- /dev/null
+++ b/llvm/test/CodeGen/AMDGPU/call-preserved-registers.ll
@@ -0,0 +1,251 @@
+; RUN: llc -mtriple=amdgcn-amd-amdhsa -mcpu=fiji -amdgpu-function-calls -verify-machineinstrs < %s | FileCheck -check-prefix=GCN %s
+; RUN: llc -mtriple=amdgcn-amd-amdhsa -mcpu=hawaii -amdgpu-function-calls -verify-machineinstrs < %s | FileCheck -check-prefix=GCN %s
+; RUN: llc -mtriple=amdgcn-amd-amdhsa -mcpu=gfx900 -amdgpu-function-calls -verify-machineinstrs < %s | FileCheck -check-prefix=GCN %s
+
+declare void @external_void_func_void() #0
+
+; GCN-LABEL: {{^}}test_kernel_call_external_void_func_void_clobber_s30_s31_call_external_void_func_void:
+; GCN: s_mov_b32 s33, s7
+; GCN: s_getpc_b64 s[34:35]
+; GCN-NEXT: s_add_u32 s34, s34,
+; GCN-NEXT: s_addc_u32 s35, s35,
+; GCN-NEXT: s_mov_b32 s4, s33
+; GCN-NEXT: s_mov_b32 s32, s33
+; GCN: s_swappc_b64 s[30:31], s[34:35]
+
+; GCN-NEXT: s_mov_b32 s4, s33
+; GCN-NEXT: #ASMSTART
+; GCN-NEXT: #ASMEND
+; GCN-NEXT: s_swappc_b64 s[30:31], s[34:35]
+define amdgpu_kernel void @test_kernel_call_external_void_func_void_clobber_s30_s31_call_external_void_func_void() #0 {
+  call void @external_void_func_void()
+  call void asm sideeffect "", ""() #0
+  call void @external_void_func_void()
+  ret void
+}
+
+; GCN-LABEL: {{^}}test_func_call_external_void_func_void_clobber_s30_s31_call_external_void_func_void:
+; GCN: v_writelane_b32 v32, s33, 0
+; GCN: v_writelane_b32 v32, s34, 1
+; GCN: v_writelane_b32 v32, s35, 2
+; GCN: v_writelane_b32 v32, s36, 3
+; GCN: v_writelane_b32 v32, s37, 4
+
+; GCN: s_mov_b32 s33, s5
+; GCN: s_swappc_b64
+; GCN-NEXT: ;;#ASMSTART
+; GCN-NEXT: ;;#ASMEND
+; GCN-NEXT: s_swappc_b64
+; GCN: s_mov_b32 s5, s33
+; GCN: v_readlane_b32 s37, v32, 4
+; GCN: v_readlane_b32 s36, v32, 3
+; GCN: v_readlane_b32 s35, v32, 2
+; GCN: v_readlane_b32 s34, v32, 1
+; GCN: v_readlane_b32 s33, v32, 0
+; GCN: s_setpc_b64
+define void @test_func_call_external_void_func_void_clobber_s30_s31_call_external_void_func_void() #0 {
+  call void @external_void_func_void()
+  call void asm sideeffect "", ""() #0
+  call void @external_void_func_void()
+  ret void
+}
+
+; GCN-LABEL: {{^}}void_func_void_clobber_s30_s31:
+; GCN: s_waitcnt
+; GCN-NEXT: s_mov_b64 [[SAVEPC:s\[[0-9]+:[0-9]+\]]], s[30:31]
+; GCN-NEXT: #ASMSTART
+; GCN: ; clobber
+; GCN-NEXT: #ASMEND
+; GCN-NEXT: s_mov_b64 s[30:31], [[SAVEPC]]
+; GCN-NEXT: s_setpc_b64 s[30:31]
+define void @void_func_void_clobber_s30_s31() #2 {
+  call void asm sideeffect "; clobber", "~{s[30:31]}"() #0
+  ret void
+}
+
+; GCN-LABEL: {{^}}void_func_void_clobber_vcc:
+; GCN: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
+; GCN-NEXT: ;;#ASMSTART
+; GCN-NEXT: ;;#ASMEND
+; GCN-NEXT: s_setpc_b64 s[30:31]
+define void @void_func_void_clobber_vcc() #2 {
+  call void asm sideeffect "", "~{VCC}"() #0
+  ret void
+}
+
+; GCN-LABEL: {{^}}test_call_void_func_void_clobber_vcc:
+; GCN: s_getpc_b64
+; GCN-NEXT: s_add_u32
+; GCN-NEXT: s_addc_u32
+; GCN: s_mov_b64 s[34:35], vcc
+; GCN-NEXT: s_swappc_b64
+; GCN: s_mov_b64 vcc, s[34:35]
+define amdgpu_kernel void @test_call_void_func_void_clobber_vcc(i32 addrspace(1)* %out) #0 {
+  %vcc = call i64 asm sideeffect "; def $0", "={vcc}"()
+  call void @void_func_void_clobber_vcc()
+  %val0 = load volatile i32, i32 addrspace(1)* undef
+  %val1 = load volatile i32, i32 addrspace(1)* undef
+  call void asm sideeffect "; use $0", "{vcc}"(i64 %vcc)
+  ret void
+}
+
+; GCN-LABEL: {{^}}test_call_void_func_void_mayclobber_s31:
+; GCN: s_mov_b32 s33, s31
+; GCN-NEXT: s_swappc_b64
+; GCN-NEXT: s_mov_b32 s31, s33
+define amdgpu_kernel void @test_call_void_func_void_mayclobber_s31(i32 addrspace(1)* %out) #0 {
+  %s31 = call i32 asm sideeffect "; def $0", "={s31}"()
+  call void @external_void_func_void()
+  call void asm sideeffect "; use $0", "{s31}"(i32 %s31)
+  ret void
+}
+
+; GCN-LABEL: {{^}}test_call_void_func_void_mayclobber_v31:
+; GCN: v_mov_b32_e32 v32, v31
+; GCN-NEXT: s_swappc_b64
+; GCN-NEXT: v_mov_b32_e32 v31, v32
+define amdgpu_kernel void @test_call_void_func_void_mayclobber_v31(i32 addrspace(1)* %out) #0 {
+  %v31 = call i32 asm sideeffect "; def $0", "={v31}"()
+  call void @external_void_func_void()
+  call void asm sideeffect "; use $0", "{v31}"(i32 %v31)
+  ret void
+}
+
+; GCN-LABEL: {{^}}test_call_void_func_void_preserves_s33:
+; GCN: s_mov_b32 s34, s9
+; GCN: ; def s33
+; GCN-NEXT: #ASMEND
+; GCN: s_getpc_b64 s[6:7]
+; GCN-NEXT: s_add_u32 s6, s6, external_void_func_void@rel32@lo+4
+; GCN-NEXT: s_addc_u32 s7, s7, external_void_func_void@rel32@hi+4
+; GCN-NEXT: s_mov_b32 s4, s34
+; GCN-NEXT: s_mov_b32 s32, s34
+; GCN-NEXT: s_swappc_b64 s[30:31], s[6:7]
+; GCN-NEXT: ;;#ASMSTART
+; GCN-NEXT: ; use s33
+; GCN-NEXT: ;;#ASMEND
+; GCN-NEXT: s_endpgm
+define amdgpu_kernel void @test_call_void_func_void_preserves_s33(i32 addrspace(1)* %out) #0 {
+  %s33 = call i32 asm sideeffect "; def $0", "={s33}"()
+  call void @external_void_func_void()
+  call void asm sideeffect "; use $0", "{s33}"(i32 %s33)
+  ret void
+}
+
+; GCN-LABEL: {{^}}test_call_void_func_void_preserves_v32:
+; GCN: s_mov_b32 s33, s9
+; GCN: ; def v32
+; GCN-NEXT: #ASMEND
+; GCN: s_getpc_b64 s[6:7]
+; GCN-NEXT: s_add_u32 s6, s6, external_void_func_void@rel32@lo+4
+; GCN-NEXT: s_addc_u32 s7, s7, external_void_func_void@rel32@hi+4
+; GCN-NEXT: s_mov_b32 s4, s33
+; GCN-NEXT: s_mov_b32 s32, s33
+; GCN-NEXT: s_swappc_b64 s[30:31], s[6:7]
+; GCN-NEXT: ;;#ASMSTART
+; GCN-NEXT: ; use v32
+; GCN-NEXT: ;;#ASMEND
+; GCN-NEXT: s_endpgm
+define amdgpu_kernel void @test_call_void_func_void_preserves_v32(i32 addrspace(1)* %out) #0 {
+  %v32 = call i32 asm sideeffect "; def $0", "={v32}"()
+  call void @external_void_func_void()
+  call void asm sideeffect "; use $0", "{v32}"(i32 %v32)
+  ret void
+}
+
+; GCN-LABEL: {{^}}void_func_void_clobber_s33:
+; GCN: v_writelane_b32 v0, s33, 0
+; GCN-NEXT: #ASMSTART
+; GCN-NEXT: ; clobber
+; GCN-NEXT: #ASMEND
+; GCN-NEXT:	v_readlane_b32 s33, v0, 0
+; GCN-NEXT: s_setpc_b64
+define void @void_func_void_clobber_s33() #2 {
+  call void asm sideeffect "; clobber", "~{s33}"() #0
+  ret void
+}
+
+; GCN-LABEL: {{^}}test_call_void_func_void_clobber_s33:
+; GCN: s_mov_b32 s33, s7
+; GCN: s_getpc_b64
+; GCN-NEXT: s_add_u32
+; GCN-NEXT: s_addc_u32
+; GCN-NEXT: s_mov_b32 s4, s33
+; GCN-NEXT: s_mov_b32 s32, s33
+; GCN: s_swappc_b64
+; GCN-NEXT: s_endpgm
+define amdgpu_kernel void @test_call_void_func_void_clobber_s33() #0 {
+  call void @void_func_void_clobber_s33()
+  ret void
+}
+
+; GCN-LABEL: {{^}}callee_saved_sgpr_func:
+; GCN-NOT: s40
+; GCN: v_writelane_b32 v32, s40
+; GCN: s_swappc_b64
+; GCN-NOT: s40
+; GCN: ; use s40
+; GCN-NOT: s40
+; GCN: v_readlane_b32 s40, v32
+; GCN-NOT: s40
+define void @callee_saved_sgpr_func() #2 {
+  %s40 = call i32 asm sideeffect "; def s40", "={s40}"() #0
+  call void @external_void_func_void()
+  call void asm sideeffect "; use $0", "s"(i32 %s40) #0
+  ret void
+}
+
+; GCN-LABEL: {{^}}callee_saved_sgpr_kernel:
+; GCN-NOT: s40
+; GCN: ; def s40
+; GCN-NOT: s40
+; GCN: s_swappc_b64
+; GCN-NOT: s40
+; GCN: ; use s40
+; GCN-NOT: s40
+define amdgpu_kernel void @callee_saved_sgpr_kernel() #2 {
+  %s40 = call i32 asm sideeffect "; def s40", "={s40}"() #0
+  call void @external_void_func_void()
+  call void asm sideeffect "; use $0", "s"(i32 %s40) #0
+  ret void
+}
+
+; First call preserved VGPR is used so it can't be used for SGPR spills.
+; GCN-LABEL: {{^}}callee_saved_sgpr_vgpr_func:
+; GCN-NOT: s40
+; GCN: v_writelane_b32 v33, s40
+; GCN: s_swappc_b64
+; GCN-NOT: s40
+; GCN: ; use s40
+; GCN-NOT: s40
+; GCN: v_readlane_b32 s40, v33
+; GCN-NOT: s40
+define void @callee_saved_sgpr_vgpr_func() #2 {
+  %s40 = call i32 asm sideeffect "; def s40", "={s40}"() #0
+  %v32 = call i32 asm sideeffect "; def v32", "={v32}"() #0
+  call void @external_void_func_void()
+  call void asm sideeffect "; use $0", "s"(i32 %s40) #0
+  call void asm sideeffect "; use $0", "v"(i32 %v32) #0
+  ret void
+}
+
+; GCN-LABEL: {{^}}callee_saved_sgpr_vgpr_kernel:
+; GCN-NOT: s40
+; GCN: ; def s40
+; GCN-NOT: s40
+; GCN: s_swappc_b64
+; GCN-NOT: s40
+; GCN: ; use s40
+; GCN-NOT: s40
+define amdgpu_kernel void @callee_saved_sgpr_vgpr_kernel() #2 {
+  %s40 = call i32 asm sideeffect "; def s40", "={s40}"() #0
+  %v32 = call i32 asm sideeffect "; def v32", "={v32}"() #0
+  call void @external_void_func_void()
+  call void asm sideeffect "; use $0", "s"(i32 %s40) #0
+  call void asm sideeffect "; use $0", "v"(i32 %v32) #0
+  ret void
+}
+
+attributes #0 = { nounwind }
+attributes #1 = { nounwind readnone }
+attributes #2 = { nounwind noinline }
diff --git a/llvm/test/CodeGen/AMDGPU/call-return-types.ll b/llvm/test/CodeGen/AMDGPU/call-return-types.ll
new file mode 100644
index 00000000000..2d6d5092412
--- /dev/null
+++ b/llvm/test/CodeGen/AMDGPU/call-return-types.ll
@@ -0,0 +1,241 @@
+; RUN: llc -mtriple=amdgcn-amd-amdhsa -mcpu=fiji -mattr=-flat-for-global -amdgpu-function-calls -verify-machineinstrs < %s | FileCheck -check-prefix=GCN %s
+; RUN: llc -mtriple=amdgcn-amd-amdhsa -mcpu=hawaii -amdgpu-function-calls -verify-machineinstrs < %s | FileCheck -check-prefix=GCN %s
+; RUN: llc -mtriple=amdgcn-amd-amdhsa -mcpu=gfx900 -mattr=-flat-for-global -amdgpu-function-calls -verify-machineinstrs < %s | FileCheck -check-prefix=GCN %s
+
+declare void @external_void_func_void() #0
+
+declare i1 @external_i1_func_void() #0
+declare zeroext i1 @external_i1_zeroext_func_void() #0
+declare signext i1 @external_i1_signext_func_void() #0
+
+declare i8 @external_i8_func_void() #0
+declare zeroext i8 @external_i8_zeroext_func_void() #0
+declare signext i8 @external_i8_signext_func_void() #0
+
+declare i16 @external_i16_func_void() #0
+declare zeroext i16 @external_i16_zeroext_func_void() #0
+declare signext i16 @external_i16_signext_func_void() #0
+
+declare i32 @external_i32_func_void() #0
+declare i64 @external_i64_func_void() #0
+declare half @external_f16_func_void() #0
+declare float @external_f32_func_void() #0
+declare double @external_f64_func_void() #0
+
+declare <2 x i32> @external_v2i32_func_void() #0
+declare <3 x i32> @external_v3i32_func_void() #0
+declare <4 x i32> @external_v4i32_func_void() #0
+declare <5 x i32> @external_v5i32_func_void() #0
+declare <8 x i32> @external_v8i32_func_void() #0
+declare <16 x i32> @external_v16i32_func_void() #0
+declare <32 x i32> @external_v32i32_func_void() #0
+declare { <32 x i32>, i32 } @external_v32i32_i32_func_void() #0
+declare <2 x i16> @external_v2i16_func_void() #0
+declare <2 x half> @external_v2f16_func_void() #0
+
+declare { i32, i64 } @external_i32_i64_func_void() #0
+
+; GCN-LABEL: {{^}}test_call_external_void_func_void:
+define amdgpu_kernel void @test_call_external_void_func_void() #0 {
+  call void @external_void_func_void()
+  ret void
+}
+
+; GCN-LABEL: {{^}}test_call_external_void_func_void_x2:
+define amdgpu_kernel void @test_call_external_void_func_void_x2() #0 {
+  call void @external_void_func_void()
+  call void @external_void_func_void()
+  ret void
+}
+
+; GCN-LABEL: {{^}}test_call_external_i1_func_void:
+define amdgpu_kernel void @test_call_external_i1_func_void() #0 {
+  %val = call i1 @external_i1_func_void()
+  store volatile i1 %val, i1 addrspace(1)* undef
+  ret void
+}
+
+; GCN-LABEL: {{^}}test_call_external_i1_zeroext_func_void:
+define amdgpu_kernel void @test_call_external_i1_zeroext_func_void() #0 {
+  %val = call i1 @external_i1_zeroext_func_void()
+  %val.ext = zext i1 %val to i32
+  store volatile i32 %val.ext, i32 addrspace(1)* undef
+  ret void
+}
+
+; GCN-LABEL: {{^}}test_call_external_i1_signext_func_void:
+define amdgpu_kernel void @test_call_external_i1_signext_func_void() #0 {
+  %val = call i1 @external_i1_signext_func_void()
+  %val.ext = zext i1 %val to i32
+  store volatile i32 %val.ext, i32 addrspace(1)* undef
+  ret void
+}
+
+; GCN-LABEL: {{^}}test_call_external_i8_func_void:
+define amdgpu_kernel void @test_call_external_i8_func_void() #0 {
+  %val = call i8 @external_i8_func_void()
+  store volatile i8 %val, i8 addrspace(1)* undef
+  ret void
+}
+
+; GCN-LABEL: {{^}}test_call_external_i8_zeroext_func_void:
+define amdgpu_kernel void @test_call_external_i8_zeroext_func_void() #0 {
+  %val = call i8 @external_i8_zeroext_func_void()
+  %val.ext = zext i8 %val to i32
+  store volatile i32 %val.ext, i32 addrspace(1)* undef
+  ret void
+}
+
+; GCN-LABEL: {{^}}test_call_external_i8_signext_func_void:
+define amdgpu_kernel void @test_call_external_i8_signext_func_void() #0 {
+  %val = call i8 @external_i8_signext_func_void()
+  %val.ext = zext i8 %val to i32
+  store volatile i32 %val.ext, i32 addrspace(1)* undef
+  ret void
+}
+
+; GCN-LABEL: {{^}}test_call_external_i16_func_void:
+define amdgpu_kernel void @test_call_external_i16_func_void() #0 {
+  %val = call i16 @external_i16_func_void()
+  store volatile i16 %val, i16 addrspace(1)* undef
+  ret void
+}
+
+; GCN-LABEL: {{^}}test_call_external_i16_zeroext_func_void:
+define amdgpu_kernel void @test_call_external_i16_zeroext_func_void() #0 {
+  %val = call i16 @external_i16_zeroext_func_void()
+  %val.ext = zext i16 %val to i32
+  store volatile i32 %val.ext, i32 addrspace(1)* undef
+  ret void
+}
+
+; GCN-LABEL: {{^}}test_call_external_i16_signext_func_void:
+define amdgpu_kernel void @test_call_external_i16_signext_func_void() #0 {
+  %val = call i16 @external_i16_signext_func_void()
+  %val.ext = zext i16 %val to i32
+  store volatile i32 %val.ext, i32 addrspace(1)* undef
+  ret void
+}
+
+; GCN-LABEL: {{^}}test_call_external_i32_func_void:
+define amdgpu_kernel void @test_call_external_i32_func_void() #0 {
+  %val = call i32 @external_i32_func_void()
+  store volatile i32 %val, i32 addrspace(1)* undef
+  ret void
+}
+
+; GCN-LABEL: {{^}}test_call_external_i64_func_void:
+define amdgpu_kernel void @test_call_external_i64_func_void() #0 {
+  %val = call i64 @external_i64_func_void()
+  store volatile i64 %val, i64 addrspace(1)* undef
+  ret void
+}
+
+; GCN-LABEL: {{^}}test_call_external_f16_func_void:
+define amdgpu_kernel void @test_call_external_f16_func_void() #0 {
+  %val = call half @external_f16_func_void()
+  store volatile half %val, half addrspace(1)* undef
+  ret void
+}
+
+; GCN-LABEL: {{^}}test_call_external_f32_func_void:
+define amdgpu_kernel void @test_call_external_f32_func_void() #0 {
+  %val = call float @external_f32_func_void()
+  store volatile float %val, float addrspace(1)* undef
+  ret void
+}
+
+; GCN-LABEL: {{^}}test_call_external_f64_func_void:
+define amdgpu_kernel void @test_call_external_f64_func_void() #0 {
+  %val = call double @external_f64_func_void()
+  store volatile double %val, double addrspace(1)* undef
+  ret void
+}
+
+; GCN-LABEL: {{^}}test_call_external_v2i32_func_void:
+define amdgpu_kernel void @test_call_external_v2i32_func_void() #0 {
+  %val = call <2 x i32> @external_v2i32_func_void()
+  store volatile <2 x i32> %val, <2 x i32> addrspace(1)* undef
+  ret void
+}
+
+; GCN-LABEL: {{^}}test_call_external_v3i32_func_void:
+define amdgpu_kernel void @test_call_external_v3i32_func_void() #0 {
+  %val = call <3 x i32> @external_v3i32_func_void()
+  store volatile <3 x i32> %val, <3 x i32> addrspace(1)* undef, align 8
+  ret void
+}
+
+; GCN-LABEL: {{^}}test_call_external_v4i32_func_void:
+define amdgpu_kernel void @test_call_external_v4i32_func_void() #0 {
+  %val = call <4 x i32> @external_v4i32_func_void()
+  store volatile <4 x i32> %val, <4 x i32> addrspace(1)* undef, align 8
+  ret void
+}
+
+; GCN-LABEL: {{^}}test_call_external_v5i32_func_void:
+define amdgpu_kernel void @test_call_external_v5i32_func_void() #0 {
+  %val = call <5 x i32> @external_v5i32_func_void()
+  store volatile <5 x i32> %val, <5 x i32> addrspace(1)* undef, align 8
+  ret void
+}
+
+; GCN-LABEL: {{^}}test_call_external_v8i32_func_void:
+define amdgpu_kernel void @test_call_external_v8i32_func_void() #0 {
+  %val = call <8 x i32> @external_v8i32_func_void()
+  store volatile <8 x i32> %val, <8 x i32> addrspace(1)* undef, align 8
+  ret void
+}
+
+; GCN-LABEL: {{^}}test_call_external_v16i32_func_void:
+define amdgpu_kernel void @test_call_external_v16i32_func_void() #0 {
+  %val = call <16 x i32> @external_v16i32_func_void()
+  store volatile <16 x i32> %val, <16 x i32> addrspace(1)* undef, align 8
+  ret void
+}
+
+; GCN-LABEL: {{^}}test_call_external_v32i32_func_void:
+define amdgpu_kernel void @test_call_external_v32i32_func_void() #0 {
+  %val = call <32 x i32> @external_v32i32_func_void()
+  store volatile <32 x i32> %val, <32 x i32> addrspace(1)* undef, align 8
+  ret void
+}
+
+; GCN-LABEL: {{^}}test_call_external_v2i16_func_void:
+define amdgpu_kernel void @test_call_external_v2i16_func_void() #0 {
+  %val = call <2 x i16> @external_v2i16_func_void()
+  store volatile <2 x i16> %val, <2 x i16> addrspace(1)* undef
+  ret void
+}
+
+; GCN-LABEL: {{^}}test_call_external_v2f16_func_void:
+define amdgpu_kernel void @test_call_external_v2f16_func_void() #0 {
+  %val = call <2 x half> @external_v2f16_func_void()
+  store volatile <2 x half> %val, <2 x half> addrspace(1)* undef
+  ret void
+}
+
+; GCN-LABEL: {{^}}test_call_external_i32_i64_func_void:
+define amdgpu_kernel void @test_call_external_i32_i64_func_void() #0 {
+  %val = call { i32, i64 } @external_i32_i64_func_void()
+  %val.0 = extractvalue { i32, i64 } %val, 0
+  %val.1 = extractvalue { i32, i64 } %val, 1
+  store volatile i32 %val.0, i32 addrspace(1)* undef
+  store volatile i64 %val.1, i64 addrspace(1)* undef
+  ret void
+}
+
+; Requires writing results to stack
+; GCN-LABEL: {{^}}test_call_external_v32i32_i32_func_void:
+define amdgpu_kernel void @test_call_external_v32i32_i32_func_void() #0 {
+  %val = call { <32 x i32>, i32 } @external_v32i32_i32_func_void()
+  %val0 = extractvalue { <32 x i32>, i32 } %val, 0
+  %val1 = extractvalue { <32 x i32>, i32 } %val, 1
+  store volatile <32 x i32> %val0, <32 x i32> addrspace(1)* undef, align 8
+  store volatile i32 %val1, i32 addrspace(1)* undef
+  ret void
+}
+
+attributes #0 = { nounwind }
+attributes #1 = { nounwind readnone }
+attributes #2 = { nounwind noinline }
diff --git a/llvm/test/CodeGen/AMDGPU/callee-frame-setup.ll b/llvm/test/CodeGen/AMDGPU/callee-frame-setup.ll
index 6c3594bb44e..a07199c1a09 100644
--- a/llvm/test/CodeGen/AMDGPU/callee-frame-setup.ll
+++ b/llvm/test/CodeGen/AMDGPU/callee-frame-setup.ll
@@ -1,4 +1,5 @@
-; RUN: llc -march=amdgcn -mcpu=hawaii -verify-machineinstrs < %s | FileCheck  -enable-var-scope -check-prefix=GCN -check-prefix=CI %s
+; RUN: llc -amdgpu-function-calls -march=amdgcn -mcpu=hawaii -verify-machineinstrs < %s | FileCheck  -enable-var-scope -check-prefix=GCN -check-prefix=CI %s
+; RUN: llc -amdgpu-function-calls -march=amdgcn -mcpu=gfx900 -verify-machineinstrs < %s | FileCheck  -enable-var-scope -check-prefix=GCN -check-prefix=GFX9 %s
 
 ; GCN-LABEL: {{^}}callee_no_stack:
 ; GCN: ; BB#0:
@@ -8,6 +9,14 @@ define void @callee_no_stack() #0 {
   ret void
 }
 
+; GCN-LABEL: {{^}}callee_no_stack_no_fp_elim:
+; GCN: ; BB#0:
+; GCN-NEXT: s_waitcnt
+; GCN-NEXT: s_setpc_b64
+define void @callee_no_stack_no_fp_elim() #1 {
+  ret void
+}
+
 ; Requires frame pointer for access to local regular object.
 
 ; GCN-LABEL: {{^}}callee_with_stack:
@@ -24,4 +33,51 @@ define void @callee_with_stack() #0 {
   ret void
 }
 
+; GCN-LABEL: {{^}}callee_with_stack_and_call:
+; GCN: ; BB#0:
+; GCN-NEXT: s_waitcnt
+
+; GCN-DAG: s_mov_b32 s5, s32
+; GCN-DAG: v_writelane_b32 v32, s33,
+; GCN-DAG: v_writelane_b32 v32, s34,
+; GCN-DAG: v_writelane_b32 v32, s35,
+; GCN-DAG: buffer_store_dword v0, off, s[0:3], s5 offset:4{{$}}
+; GCN-DAG: s_add_u32 s32, s32, 0x200{{$}}
+; GCN-DAG: v_mov_b32_e32 v0, 0{{$}}
+; GCN-DAG: s_mov_b32 s33, s5
+
+
+; GCN: s_swappc_b64
+; GCN: s_mov_b32 s5, s33
+; GCN-DAG: v_readlane_b32 s35,
+; GCN-DAG: v_readlane_b32 s34,
+; GCN-DAG: v_readlane_b32 s33,
+; GCN: s_waitcnt
+; GCN-NEXT: s_setpc_b64
+define void @callee_with_stack_and_call() #0 {
+  %alloca = alloca i32
+  store volatile i32 0, i32* %alloca
+  call void @external_void_func_void()
+  ret void
+}
+
+; Should be able to copy incoming stack pointer directly to inner
+; call's stack pointer argument.
+
+; GCN-LABEL: {{^}}callee_no_stack_with_call:
+; GCN: s_waitcnt
+; GCN-NOT: s32
+; GCN: s_mov_b32 s33, s5
+; GCN: s_swappc_b64
+; GCN: s_mov_b32 s5, s33
+; GCN-NOT: s32
+; GCN: s_setpc_b64
+define void @callee_no_stack_with_call() #0 {
+  call void @external_void_func_void()
+  ret void
+}
+
+declare void @external_void_func_void() #0
+
 attributes #0 = { nounwind }
+attributes #1 = { nounwind "no-frame-pointer-elim"="true" }
diff --git a/llvm/test/CodeGen/AMDGPU/nested-calls.ll b/llvm/test/CodeGen/AMDGPU/nested-calls.ll
new file mode 100644
index 00000000000..2d8d666a26b
--- /dev/null
+++ b/llvm/test/CodeGen/AMDGPU/nested-calls.ll
@@ -0,0 +1,41 @@
+; RUN: llc -march=amdgcn -mcpu=fiji -mattr=-flat-for-global -amdgpu-sroa=0 -amdgpu-function-calls -verify-machineinstrs < %s | FileCheck -enable-var-scope -check-prefix=GCN -check-prefix=VI %s
+; RUN: llc -march=amdgcn -mcpu=hawaii -amdgpu-function-calls -amdgpu-sroa=0 -verify-machineinstrs < %s | FileCheck -enable-var-scope -check-prefix=GCN -check-prefix=CI %s
+; RUN: llc -march=amdgcn -mcpu=gfx900 -mattr=-flat-for-global -amdgpu-sroa=0 -amdgpu-function-calls -verify-machineinstrs < %s | FileCheck -enable-var-scope -check-prefix=GCN -check-prefix=GFX9 -check-prefix=VI %s
+
+; Test calls when called by other callable functions rather than
+; kernels.
+
+declare void @external_void_func_i32(i32) #0
+
+; GCN-LABEL: {{^}}test_func_call_external_void_func_i32_imm:
+; GCN: s_waitcnt
+; GCN-NOT: s32
+; GCN: s_swappc_b64
+; GCN-NOT: s32
+; GCN: s_setpc_b64
+define void @test_func_call_external_void_func_i32_imm() #0 {
+  call void @external_void_func_i32(i32 42)
+  ret void
+}
+
+; GCN-LABEL: {{^}}test_func_call_external_void_func_i32_imm_stack_use:
+; GCN: s_waitcnt
+; GCN: s_mov_b32 s5, s32
+; GCN: s_add_u32 s32, s32, 0x1100{{$}}
+; GCN: buffer_store_dword v{{[0-9]+}}, off, s[0:3], s5 offset
+; GCN: s_swappc_b64
+; GCN: s_sub_u32 s32, s32, 0x1100{{$}}
+; GCN: s_setpc_b64
+define void @test_func_call_external_void_func_i32_imm_stack_use() #0 {
+  %alloca = alloca [16 x i32], align 4
+  %gep0 = getelementptr inbounds [16 x i32], [16 x i32]* %alloca, i32 0, i32 0
+  %gep15 = getelementptr inbounds [16 x i32], [16 x i32]* %alloca, i32 0, i32 16
+  store volatile i32 0, i32* %gep0
+  store volatile i32 0, i32* %gep15
+  call void @external_void_func_i32(i32 42)
+  ret void
+}
+
+attributes #0 = { nounwind }
+attributes #1 = { nounwind readnone }
+attributes #2 = { nounwind noinline }