AMDGPU/SI: Set correct value for amd_kernel_code_t::kernarg_segment_alignment

[arsenm]

LGTM with return type fixed

lib/Target/AMDGPU/AMDGPUMachineFunction.h

53

MaxKernArgAlign is unsigned

lib/Target/AMDGPU/AMDGPUAsmPrinter.cpp

header.kernarg_segment_byte_size =
       STM.getKernArgSegmentSize(MFI->getABIArgOffset());
   header.wavefront_sgpr_count = KernelInfo.NumSGPR;
   header.workitem_vgpr_count = KernelInfo.NumVGPR;
   header.workitem_private_segment_byte_size = KernelInfo.ScratchSize;
   header.workgroup_group_segment_byte_size = KernelInfo.LDSSize;
   header.reserved_vgpr_first = KernelInfo.ReservedVGPRFirst;
   header.reserved_vgpr_count = KernelInfo.ReservedVGPRCount;
 
+  // These alignment values are specified in powers of two, so alignment =
+  // 2^n.  The minimum alignment is 2^4 = 16.
+  header.kernarg_segment_alignment = std::max((size_t)4,
+      countTrailingZeros(MFI->getMaxKernArgAlign()));
+
   if (STM.debuggerEmitPrologue()) {
     header.debug_wavefront_private_segment_offset_sgpr =
       KernelInfo.DebuggerWavefrontPrivateSegmentOffsetSGPR;
     header.debug_private_segment_buffer_sgpr =
       KernelInfo.DebuggerPrivateSegmentBufferSGPR;
   }
 
   AMDGPUTargetStreamer *TS =

lib/Target/AMDGPU/AMDGPUMachineFunction.h

uint64_t allocateKernArg(uint64_t Size, unsigned Align) {
     MaxKernArgAlign = std::max(Align, MaxKernArgAlign);
     return Result;
   }
 
   uint64_t getKernArgSize() const {
     return KernArgSize;
   }
 
+  uint64_t getMaxKernArgAlign() const {
+    return MaxKernArgAlign;
+  }
+
   void setABIArgOffset(unsigned NewOffset) {
     ABIArgOffset = NewOffset;
   }
 
   unsigned getABIArgOffset() const {
     return ABIArgOffset;
   }
 

test/CodeGen/AMDGPU/kernel-args.ll

 ; RUN: llc < %s -march=amdgcn -verify-machineinstrs | FileCheck %s --check-prefixes=SI,GCN,MESA-GCN,FUNC
 ; RUN: llc < %s -march=amdgcn -mcpu=tonga -verify-machineinstrs | FileCheck %s --check-prefixes=VI,GCN,MESA-VI,MESA-GCN,FUNC
 ; RUN: llc < %s -mtriple=amdgcn--amdhsa -mcpu=fiji -verify-machineinstrs | FileCheck %s --check-prefixes=VI,GCN,HSA-VI,FUNC
 ; RUN: llc < %s -march=r600 -mcpu=redwood | FileCheck %s --check-prefix=EG --check-prefix=FUNC
 ; RUN: llc < %s -march=r600 -mcpu=cayman | FileCheck %s --check-prefix=EG --check-prefix=FUNC
 
 ; FUNC-LABEL: {{^}}i8_arg:
+; HSA-VI: kernarg_segment_alignment = 4
 ; EG: AND_INT {{[ *]*}}T{{[0-9]+\.[XYZW]}}, KC0[2].Z
 ; SI: s_load_dword [[VAL:s[0-9]+]], s[{{[0-9]+:[0-9]+}}], 0xb
 ; MESA-VI: s_load_dword [[VAL:s[0-9]+]], s[{{[0-9]+:[0-9]+}}], 0x2c
 ; MESA-GCN: s_and_b32 s{{[0-9]+}}, [[VAL]], 0xff
 ; HSA-VI: s_add_u32 [[SPTR_LO:s[0-9]+]], s4, 8
 ; HSA-VI: s_addc_u32 [[SPTR_HI:s[0-9]+]], s5, 0
 ; HSA-VI: v_mov_b32_e32 v[[VPTR_LO:[0-9]+]], [[SPTR_LO]]
 ; HSA-VI: v_mov_b32_e32 v[[VPTR_HI:[0-9]+]], [[SPTR_HI]]
 ; FIXME: Should be using s_load_dword
 ; HSA-VI: flat_load_ubyte v{{[0-9]+}}, v{{\[}}[[VPTR_LO]]:[[VPTR_HI]]]
 
 define void @i8_arg(i32 addrspace(1)* nocapture %out, i8 %in) nounwind {
 entry:
   %0 = zext i8 %in to i32
   store i32 %0, i32 addrspace(1)* %out, align 4
   ret void
 }
 
 ; FUNC-LABEL: {{^}}i8_zext_arg:
+; HSA-VI: kernarg_segment_alignment = 4
 ; EG: MOV {{[ *]*}}T{{[0-9]+\.[XYZW]}}, KC0[2].Z
 ; SI: s_load_dword s{{[0-9]}}, s[0:1], 0xb
 ; MESA-VI: s_load_dword s{{[0-9]}}, s[0:1], 0x2c
 ; HSA-VI: s_add_u32 [[SPTR_LO:s[0-9]+]], s4, 8
 ; HSA-VI: s_addc_u32 [[SPTR_HI:s[0-9]+]], s5, 0
 ; HSA-VI: v_mov_b32_e32 v[[VPTR_LO:[0-9]+]], [[SPTR_LO]]
 ; HSA-VI: v_mov_b32_e32 v[[VPTR_HI:[0-9]+]], [[SPTR_HI]]
 ; FIXME: Should be using s_load_dword
 ; HSA-VI: flat_load_ubyte v{{[0-9]+}}, v{{\[}}[[VPTR_LO]]:[[VPTR_HI]]]
 
 define void @i8_zext_arg(i32 addrspace(1)* nocapture %out, i8 zeroext %in) nounwind {
 entry:
   %0 = zext i8 %in to i32
   store i32 %0, i32 addrspace(1)* %out, align 4
   ret void
 }
 
 ; FUNC-LABEL: {{^}}i8_sext_arg:
+; HSA-VI: kernarg_segment_alignment = 4
 ; EG: MOV {{[ *]*}}T{{[0-9]+\.[XYZW]}}, KC0[2].Z
 ; SI: s_load_dword s{{[0-9]}}, s[0:1], 0xb
 ; MESA-VI: s_load_dword s{{[0-9]}}, s[0:1], 0x2c
 ; HSA-VI: s_add_u32 [[SPTR_LO:s[0-9]+]], s4, 8
 ; HSA-VI: s_addc_u32 [[SPTR_HI:s[0-9]+]], s5, 0
 ; HSA-VI: v_mov_b32_e32 v[[VPTR_LO:[0-9]+]], [[SPTR_LO]]
 ; HSA-VI: v_mov_b32_e32 v[[VPTR_HI:[0-9]+]], [[SPTR_HI]]
 ; FIXME: Should be using s_load_dword
 ; HSA-VI: flat_load_sbyte v{{[0-9]+}}, v{{\[}}[[VPTR_LO]]:[[VPTR_HI]]]
 
 define void @i8_sext_arg(i32 addrspace(1)* nocapture %out, i8 signext %in) nounwind {
 entry:
   %0 = sext i8 %in to i32
   store i32 %0, i32 addrspace(1)* %out, align 4
   ret void
 }
 
 ; FUNC-LABEL: {{^}}i16_arg:
+; HSA-VI: kernarg_segment_alignment = 4
 ; EG: AND_INT {{[ *]*}}T{{[0-9]+\.[XYZW]}}, KC0[2].Z
 ; SI: s_load_dword [[VAL:s[0-9]+]], s[{{[0-9]+:[0-9]+}}], 0xb
 ; MESA-VI: s_load_dword [[VAL:s[0-9]+]], s[{{[0-9]+:[0-9]+}}], 0x2c
 ; MESA-GCN: s_and_b32 s{{[0-9]+}}, [[VAL]], 0xff
 ; HSA-VI: s_add_u32 [[SPTR_LO:s[0-9]+]], s4, 8
 ; HSA-VI: s_addc_u32 [[SPTR_HI:s[0-9]+]], s5, 0
 ; HSA-VI: v_mov_b32_e32 v[[VPTR_LO:[0-9]+]], [[SPTR_LO]]
 ; HSA-VI: v_mov_b32_e32 v[[VPTR_HI:[0-9]+]], [[SPTR_HI]]
 ; FIXME: Should be using s_load_dword
 ; HSA-VI: flat_load_ushort v{{[0-9]+}}, v{{\[}}[[VPTR_LO]]:[[VPTR_HI]]]
 
 define void @i16_arg(i32 addrspace(1)* nocapture %out, i16 %in) nounwind {
 entry:
   %0 = zext i16 %in to i32
   store i32 %0, i32 addrspace(1)* %out, align 4
   ret void
 }
 
 ; FUNC-LABEL: {{^}}i16_zext_arg:
+; HSA-VI: kernarg_segment_alignment = 4
 ; EG: MOV {{[ *]*}}T{{[0-9]+\.[XYZW]}}, KC0[2].Z
 ; SI: s_load_dword s{{[0-9]}}, s[0:1], 0xb
 ; MESA-VI: s_load_dword s{{[0-9]}}, s[0:1], 0x2c
 ; HSA-VI: s_add_u32 [[SPTR_LO:s[0-9]+]], s4, 8
 ; HSA-VI: s_addc_u32 [[SPTR_HI:s[0-9]+]], s5, 0
 ; HSA-VI: v_mov_b32_e32 v[[VPTR_LO:[0-9]+]], [[SPTR_LO]]
 ; HSA-VI: v_mov_b32_e32 v[[VPTR_HI:[0-9]+]], [[SPTR_HI]]
 ; FIXME: Should be using s_load_dword
 ; HSA-VI: flat_load_ushort v{{[0-9]+}}, v{{\[}}[[VPTR_LO]]:[[VPTR_HI]]]
 
 define void @i16_zext_arg(i32 addrspace(1)* nocapture %out, i16 zeroext %in) nounwind {
 entry:
   %0 = zext i16 %in to i32
   store i32 %0, i32 addrspace(1)* %out, align 4
   ret void
 }
 
 ; FUNC-LABEL: {{^}}i16_sext_arg:
+; HSA-VI: kernarg_segment_alignment = 4
 ; EG: MOV {{[ *]*}}T{{[0-9]+\.[XYZW]}}, KC0[2].Z
 ; SI: s_load_dword s{{[0-9]}}, s[0:1], 0xb
 ; MESA-VI: s_load_dword s{{[0-9]}}, s[0:1], 0x2c
 ; HSA-VI: s_add_u32 [[SPTR_LO:s[0-9]+]], s4, 8
 ; HSA-VI: s_addc_u32 [[SPTR_HI:s[0-9]+]], s5, 0
 ; HSA-VI: v_mov_b32_e32 v[[VPTR_LO:[0-9]+]], [[SPTR_LO]]
 ; HSA-VI: v_mov_b32_e32 v[[VPTR_HI:[0-9]+]], [[SPTR_HI]]
 ; FIXME: Should be using s_load_dword
 ; HSA-VI: flat_load_sshort v{{[0-9]+}}, v{{\[}}[[VPTR_LO]]:[[VPTR_HI]]]
 
 define void @i16_sext_arg(i32 addrspace(1)* nocapture %out, i16 signext %in) nounwind {
 entry:
   %0 = sext i16 %in to i32
   store i32 %0, i32 addrspace(1)* %out, align 4
   ret void
 }
 
 ; FUNC-LABEL: {{^}}i32_arg:
+; HSA-VI: kernarg_segment_alignment = 4
 ; EG: T{{[0-9]\.[XYZW]}}, KC0[2].Z
 ; SI: s_load_dword s{{[0-9]}}, s[0:1], 0xb
 ; MESA-VI: s_load_dword s{{[0-9]}}, s[0:1], 0x2c
 ; HSA-VI: s_load_dword s{{[0-9]}}, s[4:5], 0x8
 define void @i32_arg(i32 addrspace(1)* nocapture %out, i32 %in) nounwind {
 entry:
   store i32 %in, i32 addrspace(1)* %out, align 4
   ret void
 }
 
 ; FUNC-LABEL: {{^}}f32_arg:
+; HSA-VI: kernarg_segment_alignment = 4
 ; EG: T{{[0-9]\.[XYZW]}}, KC0[2].Z
 ; SI: s_load_dword s{{[0-9]}}, s[0:1], 0xb
 ; MESA-VI: s_load_dword s{{[0-9]}}, s[0:1], 0x2c
 ; HSA-VI: s_load_dword s{{[0-9]+}}, s[4:5], 0x8
 define void @f32_arg(float addrspace(1)* nocapture %out, float %in) nounwind {
 entry:
   store float %in, float addrspace(1)* %out, align 4
   ret void
 }
 
 ; FUNC-LABEL: {{^}}v2i8_arg:
+; HSA-VI: kernarg_segment_alignment = 4
 ; EG: VTX_READ_8
 ; EG: VTX_READ_8
 ; MESA-GCN: buffer_load_ubyte
 ; MESA-GCN: buffer_load_ubyte
 ; HSA-VI: flat_load_ubyte
 ; HSA-VI: flat_load_ubyte
 define void @v2i8_arg(<2 x i8> addrspace(1)* %out, <2 x i8> %in) {
 entry:
   store <2 x i8> %in, <2 x i8> addrspace(1)* %out
   ret void
 }
 
 ; FUNC-LABEL: {{^}}v2i16_arg:
+; HSA-VI: kernarg_segment_alignment = 4
 ; EG: VTX_READ_16
 ; EG: VTX_READ_16
 ; MESA-GCN: buffer_load_ushort
 ; MESA-GCN: buffer_load_ushort
 ; HSA-VI: flat_load_ushort
 ; HSA-VI: flat_load_ushort
 define void @v2i16_arg(<2 x i16> addrspace(1)* %out, <2 x i16> %in) {
 entry:
   store <2 x i16> %in, <2 x i16> addrspace(1)* %out
   ret void
 }
 
 ; FUNC-LABEL: {{^}}v2i32_arg:
+; HSA-VI: kernarg_segment_alignment = 4
 ; EG-DAG: T{{[0-9]\.[XYZW]}}, KC0[3].X
 ; EG-DAG: T{{[0-9]\.[XYZW]}}, KC0[2].W
 ; SI: s_load_dwordx2 s{{\[[0-9]:[0-9]\]}}, s[0:1], 0xb
 ; MESA-VI: s_load_dwordx2 s{{\[[0-9]:[0-9]\]}}, s[0:1], 0x2c
 ; HSA-VI: s_load_dwordx2 s[{{[0-9]+:[0-9]+}}], s[4:5], 0x8
 define void @v2i32_arg(<2 x i32> addrspace(1)* nocapture %out, <2 x i32> %in) nounwind {
 entry:
   store <2 x i32> %in, <2 x i32> addrspace(1)* %out, align 4
   ret void
 }
 
 ; FUNC-LABEL: {{^}}v2f32_arg:
+; HSA-VI: kernarg_segment_alignment = 4
 ; EG-DAG: T{{[0-9]\.[XYZW]}}, KC0[3].X
 ; EG-DAG: T{{[0-9]\.[XYZW]}}, KC0[2].W
 ; SI: s_load_dwordx2 s{{\[[0-9]:[0-9]\]}}, s[0:1], 0xb
 ; MESA-VI: s_load_dwordx2 s{{\[[0-9]:[0-9]\]}}, s[0:1], 0x2c
 ; HSA-VI: s_load_dwordx2 s{{\[[0-9]:[0-9]\]}}, s[4:5], 0x8
 define void @v2f32_arg(<2 x float> addrspace(1)* nocapture %out, <2 x float> %in) nounwind {
 entry:
   store <2 x float> %in, <2 x float> addrspace(1)* %out, align 4
   ret void
 }
 
 ; FUNC-LABEL: {{^}}v3i8_arg:
+; HSA-VI: kernarg_segment_alignment = 4
 ; EG-DAG: VTX_READ_8 T{{[0-9]}}.X, T{{[0-9]}}.X, 40
 ; EG-DAG: VTX_READ_8 T{{[0-9]}}.X, T{{[0-9]}}.X, 41
 ; EG-DAG: VTX_READ_8 T{{[0-9]}}.X, T{{[0-9]}}.X, 42
 ; MESA-GCN: buffer_load_ubyte
 ; MESA-GCN: buffer_load_ubyte
 ; MESA-GCN: buffer_load_ubyte
 ; HSA-VI: flat_load_ubyte
 ; HSA-VI: flat_load_ubyte
 ; HSA-VI: flat_load_ubyte
 define void @v3i8_arg(<3 x i8> addrspace(1)* nocapture %out, <3 x i8> %in) nounwind {
 entry:
   store <3 x i8> %in, <3 x i8> addrspace(1)* %out, align 4
   ret void
 }
 
 ; FUNC-LABEL: {{^}}v3i16_arg:
+; HSA-VI: kernarg_segment_alignment = 4
 ; EG-DAG: VTX_READ_16 T{{[0-9]}}.X, T{{[0-9]}}.X, 44
 ; EG-DAG: VTX_READ_16 T{{[0-9]}}.X, T{{[0-9]}}.X, 46
 ; EG-DAG: VTX_READ_16 T{{[0-9]}}.X, T{{[0-9]}}.X, 48
 ; MESA-GCN: buffer_load_ushort
 ; MESA-GCN: buffer_load_ushort
 ; MESA-GCN: buffer_load_ushort
 ; HSA-VI: flat_load_ushort
 ; HSA-VI: flat_load_ushort
 ; HSA-VI: flat_load_ushort
 define void @v3i16_arg(<3 x i16> addrspace(1)* nocapture %out, <3 x i16> %in) nounwind {
 entry:
   store <3 x i16> %in, <3 x i16> addrspace(1)* %out, align 4
   ret void
 }
 ; FUNC-LABEL: {{^}}v3i32_arg:
+; HSA-VI: kernarg_segment_alignment = 4
 ; EG-DAG: T{{[0-9]\.[XYZW]}}, KC0[3].Y
 ; EG-DAG: T{{[0-9]\.[XYZW]}}, KC0[3].Z
 ; EG-DAG: T{{[0-9]\.[XYZW]}}, KC0[3].W
 ; SI: s_load_dwordx4 s{{\[[0-9]:[0-9]+\]}}, s[0:1], 0xd
 ; MESA-VI: s_load_dwordx4 s{{\[[0-9]:[0-9]+\]}}, s[0:1], 0x34
 ; HSA-VI: s_load_dwordx4 s[{{[0-9]+:[0-9]+}}], s[4:5], 0x10
 define void @v3i32_arg(<3 x i32> addrspace(1)* nocapture %out, <3 x i32> %in) nounwind {
 entry:
   store <3 x i32> %in, <3 x i32> addrspace(1)* %out, align 4
   ret void
 }
 
 ; FUNC-LABEL: {{^}}v3f32_arg:
+; HSA-VI: kernarg_segment_alignment = 4
 ; EG-DAG: T{{[0-9]\.[XYZW]}}, KC0[3].Y
 ; EG-DAG: T{{[0-9]\.[XYZW]}}, KC0[3].Z
 ; EG-DAG: T{{[0-9]\.[XYZW]}}, KC0[3].W
 ; SI: s_load_dwordx4 s{{\[[0-9]:[0-9]+\]}}, s[0:1], 0xd
 ; MESA-VI: s_load_dwordx4 s{{\[[0-9]:[0-9]+\]}}, s[0:1], 0x34
 ; HSA-VI: s_load_dwordx4 s[{{[0-9]+:[0-9]+}}], s[4:5], 0x10
 define void @v3f32_arg(<3 x float> addrspace(1)* nocapture %out, <3 x float> %in) nounwind {
 entry:
   store <3 x float> %in, <3 x float> addrspace(1)* %out, align 4
   ret void
 }
 
 ; FUNC-LABEL: {{^}}v4i8_arg:
+; HSA-VI: kernarg_segment_alignment = 4
 ; EG: VTX_READ_8
 ; EG: VTX_READ_8
 ; EG: VTX_READ_8
 ; EG: VTX_READ_8
 ; MESA-GCN: buffer_load_ubyte
 ; MESA-GCN: buffer_load_ubyte
 ; MESA-GCN: buffer_load_ubyte
 ; MESA-GCN: buffer_load_ubyte
 ; HSA-VI: flat_load_ubyte
 ; HSA-VI: flat_load_ubyte
 ; HSA-VI: flat_load_ubyte
 ; HSA-VI: flat_load_ubyte
 define void @v4i8_arg(<4 x i8> addrspace(1)* %out, <4 x i8> %in) {
 entry:
   store <4 x i8> %in, <4 x i8> addrspace(1)* %out
   ret void
 }
 
 ; FUNC-LABEL: {{^}}v4i16_arg:
+; HSA-VI: kernarg_segment_alignment = 4
 ; EG: VTX_READ_16
 ; EG: VTX_READ_16
 ; EG: VTX_READ_16
 ; EG: VTX_READ_16
 ; MESA-GCN: buffer_load_ushort
 ; MESA-GCN: buffer_load_ushort
 ; MESA-GCN: buffer_load_ushort
 ; MESA-GCN: buffer_load_ushort
 ; HSA-GCN: flat_load_ushort
 ; HSA-GCN: flat_load_ushort
 ; HSA-GCN: flat_load_ushort
 ; HSA-GCN: flat_load_ushort
 define void @v4i16_arg(<4 x i16> addrspace(1)* %out, <4 x i16> %in) {
 entry:
   store <4 x i16> %in, <4 x i16> addrspace(1)* %out
   ret void
 }
 
 ; FUNC-LABEL: {{^}}v4i32_arg:
+; HSA-VI: kernarg_segment_alignment = 4
 ; EG-DAG: T{{[0-9]\.[XYZW]}}, KC0[3].Y
 ; EG-DAG: T{{[0-9]\.[XYZW]}}, KC0[3].Z
 ; EG-DAG: T{{[0-9]\.[XYZW]}}, KC0[3].W
 ; EG-DAG: T{{[0-9]\.[XYZW]}}, KC0[4].X
 ; SI: s_load_dwordx4 s{{\[[0-9]:[0-9]\]}}, s[0:1], 0xd
 ; MESA-VI: s_load_dwordx4 s{{\[[0-9]:[0-9]\]}}, s[0:1], 0x34
 ; HSA-VI: s_load_dwordx4 s[{{[0-9]+:[0-9]+}}], s[4:5], 0x10
 define void @v4i32_arg(<4 x i32> addrspace(1)* nocapture %out, <4 x i32> %in) nounwind {
 entry:
   store <4 x i32> %in, <4 x i32> addrspace(1)* %out, align 4
   ret void
 }
 
 ; FUNC-LABEL: {{^}}v4f32_arg:
+; HSA-VI: kernarg_segment_alignment = 4
 ; EG-DAG: T{{[0-9]\.[XYZW]}}, KC0[3].Y
 ; EG-DAG: T{{[0-9]\.[XYZW]}}, KC0[3].Z
 ; EG-DAG: T{{[0-9]\.[XYZW]}}, KC0[3].W
 ; EG-DAG: T{{[0-9]\.[XYZW]}}, KC0[4].X
 ; SI: s_load_dwordx4 s{{\[[0-9]:[0-9]\]}}, s[0:1], 0xd
 ; MESA-VI: s_load_dwordx4 s{{\[[0-9]:[0-9]\]}}, s[0:1], 0x34
 ; HSA-VI: s_load_dwordx4 s[{{[0-9]+:[0-9]+}}], s[4:5], 0x10
 define void @v4f32_arg(<4 x float> addrspace(1)* nocapture %out, <4 x float> %in) nounwind {
 entry:
   store <4 x float> %in, <4 x float> addrspace(1)* %out, align 4
   ret void
 }
 
 ; FUNC-LABEL: {{^}}v8i8_arg:
+; HSA-VI: kernarg_segment_alignment = 4
 ; EG: VTX_READ_8
 ; EG: VTX_READ_8
 ; EG: VTX_READ_8
 ; EG: VTX_READ_8
 ; EG: VTX_READ_8
 ; EG: VTX_READ_8
 ; EG: VTX_READ_8
 ; EG: VTX_READ_8

 ; HSA-GCN: float_load_ubyte
 define void @v8i8_arg(<8 x i8> addrspace(1)* %out, <8 x i8> %in) {
 entry:
   store <8 x i8> %in, <8 x i8> addrspace(1)* %out
   ret void
 }
 
 ; FUNC-LABEL: {{^}}v8i16_arg:
+; HSA-VI: kernarg_segment_alignment = 4
 ; EG: VTX_READ_16
 ; EG: VTX_READ_16
 ; EG: VTX_READ_16
 ; EG: VTX_READ_16
 ; EG: VTX_READ_16
 ; EG: VTX_READ_16
 ; EG: VTX_READ_16
 ; EG: VTX_READ_16

 ; HSA-VI: flat_load_ushort
 define void @v8i16_arg(<8 x i16> addrspace(1)* %out, <8 x i16> %in) {
 entry:
   store <8 x i16> %in, <8 x i16> addrspace(1)* %out
   ret void
 }
 
 ; FUNC-LABEL: {{^}}v8i32_arg:
+; HSA-VI: kernarg_segment_alignment = 5
 ; EG-DAG: T{{[0-9]\.[XYZW]}}, KC0[4].Y
 ; EG-DAG: T{{[0-9]\.[XYZW]}}, KC0[4].Z
 ; EG-DAG: T{{[0-9]\.[XYZW]}}, KC0[4].W
 ; EG-DAG: T{{[0-9]\.[XYZW]}}, KC0[5].X
 ; EG-DAG: T{{[0-9]\.[XYZW]}}, KC0[5].Y
 ; EG-DAG: T{{[0-9]\.[XYZW]}}, KC0[5].Z
 ; EG-DAG: T{{[0-9]\.[XYZW]}}, KC0[5].W
 ; EG-DAG: T{{[0-9]\.[XYZW]}}, KC0[6].X
 ; SI: s_load_dwordx8 s{{\[[0-9]+:[0-9]+\]}}, s[0:1], 0x11
 ; MESA-VI: s_load_dwordx8 s{{\[[0-9]+:[0-9]+\]}}, s[0:1], 0x44
 ; HSA-VI: s_load_dwordx8 s[{{[0-9]+:[0-9]+}}], s[4:5], 0x20
 define void @v8i32_arg(<8 x i32> addrspace(1)* nocapture %out, <8 x i32> %in) nounwind {
 entry:
   store <8 x i32> %in, <8 x i32> addrspace(1)* %out, align 4
   ret void
 }
 
 ; FUNC-LABEL: {{^}}v8f32_arg:
+; HSA-VI: kernarg_segment_alignment = 5
 ; EG-DAG: T{{[0-9]\.[XYZW]}}, KC0[4].Y
 ; EG-DAG: T{{[0-9]\.[XYZW]}}, KC0[4].Z
 ; EG-DAG: T{{[0-9]\.[XYZW]}}, KC0[4].W
 ; EG-DAG: T{{[0-9]\.[XYZW]}}, KC0[5].X
 ; EG-DAG: T{{[0-9]\.[XYZW]}}, KC0[5].Y
 ; EG-DAG: T{{[0-9]\.[XYZW]}}, KC0[5].Z
 ; EG-DAG: T{{[0-9]\.[XYZW]}}, KC0[5].W
 ; EG-DAG: T{{[0-9]\.[XYZW]}}, KC0[6].X
 ; SI: s_load_dwordx8 s{{\[[0-9]+:[0-9]+\]}}, s[0:1], 0x11
 define void @v8f32_arg(<8 x float> addrspace(1)* nocapture %out, <8 x float> %in) nounwind {
 entry:
   store <8 x float> %in, <8 x float> addrspace(1)* %out, align 4
   ret void
 }
 
 ; FUNC-LABEL: {{^}}v16i8_arg:
+; HSA-VI: kernarg_segment_alignment = 4
 ; EG: VTX_READ_8
 ; EG: VTX_READ_8
 ; EG: VTX_READ_8
 ; EG: VTX_READ_8
 ; EG: VTX_READ_8
 ; EG: VTX_READ_8
 ; EG: VTX_READ_8
 ; EG: VTX_READ_8

 ; HSA-VI: flat_load_ubyte
 define void @v16i8_arg(<16 x i8> addrspace(1)* %out, <16 x i8> %in) {
 entry:
   store <16 x i8> %in, <16 x i8> addrspace(1)* %out
   ret void
 }
 
 ; FUNC-LABEL: {{^}}v16i16_arg:
+; HSA-VI: kernarg_segment_alignment = 5
 ; EG: VTX_READ_16
 ; EG: VTX_READ_16
 ; EG: VTX_READ_16
 ; EG: VTX_READ_16
 ; EG: VTX_READ_16
 ; EG: VTX_READ_16
 ; EG: VTX_READ_16
 ; EG: VTX_READ_16

 ; HSA-VI: flat_load_ushort
 define void @v16i16_arg(<16 x i16> addrspace(1)* %out, <16 x i16> %in) {
 entry:
   store <16 x i16> %in, <16 x i16> addrspace(1)* %out
   ret void
 }
 
 ; FUNC-LABEL: {{^}}v16i32_arg:
+; HSA-VI: kernarg_segment_alignment = 6
 ; EG-DAG: T{{[0-9]\.[XYZW]}}, KC0[6].Y
 ; EG-DAG: T{{[0-9]\.[XYZW]}}, KC0[6].Z
 ; EG-DAG: T{{[0-9]\.[XYZW]}}, KC0[6].W
 ; EG-DAG: T{{[0-9]\.[XYZW]}}, KC0[7].X
 ; EG-DAG: T{{[0-9]\.[XYZW]}}, KC0[7].Y
 ; EG-DAG: T{{[0-9]\.[XYZW]}}, KC0[7].Z
 ; EG-DAG: T{{[0-9]\.[XYZW]}}, KC0[7].W
 ; EG-DAG: T{{[0-9]\.[XYZW]}}, KC0[8].X

 ; HSA-VI: s_load_dwordx16 s[{{[0-9]+:[0-9]+}}], s[4:5], 0x40
 define void @v16i32_arg(<16 x i32> addrspace(1)* nocapture %out, <16 x i32> %in) nounwind {
 entry:
   store <16 x i32> %in, <16 x i32> addrspace(1)* %out, align 4
   ret void
 }
 
 ; FUNC-LABEL: {{^}}v16f32_arg:
+; HSA-VI: kernarg_segment_alignment = 6
 ; EG-DAG: T{{[0-9]\.[XYZW]}}, KC0[6].Y
 ; EG-DAG: T{{[0-9]\.[XYZW]}}, KC0[6].Z
 ; EG-DAG: T{{[0-9]\.[XYZW]}}, KC0[6].W
 ; EG-DAG: T{{[0-9]\.[XYZW]}}, KC0[7].X
 ; EG-DAG: T{{[0-9]\.[XYZW]}}, KC0[7].Y
 ; EG-DAG: T{{[0-9]\.[XYZW]}}, KC0[7].Z
 ; EG-DAG: T{{[0-9]\.[XYZW]}}, KC0[7].W
 ; EG-DAG: T{{[0-9]\.[XYZW]}}, KC0[8].X