This is an archive of the discontinued LLVM Phabricator instance.

Differential D32645

AMDGPU: GFX9 GS and HS shaders always have the scratch wave offset in SGPR5
ClosedPublic

Authored by mareko on Apr 28 2017, 7:23 AM.

Details

Reviewers
arsenm
nhaehnle
Commits
rG584d2c05d4ba: AMDGPU: GFX9 GS and HS shaders always have the scratch wave offset in SGPR5
rL302200: AMDGPU: GFX9 GS and HS shaders always have the scratch wave offset in SGPR5

Diff Detail

Event Timeline

mareko created this revision.Apr 28 2017, 7:23 AM
Herald added subscribers: t-tye, tpr, dstuttard and 3 others. · View Herald TranscriptApr 28 2017, 7:24 AM
arsenm added inline comments.Apr 28 2017, 9:25 AM
lib/Target/AMDGPU/SIFrameLowering.cpp
355–356

This shouldn't be a recorded property of a function. Can you check MRI for whether the register has uses?

test/CodeGen/AMDGPU/scratch-simple.ll
6–8

Copy paste from local-stack-slot-bug.ll.

It would probably be better to merge the two tests rathe than having the exact same function repeated in multiple files.

mareko added inline comments.Apr 28 2017, 11:34 AM
test/CodeGen/AMDGPU/scratch-simple.ll
6–8

No. This patch renames local-stack-slot-bug.ll to scratch-simple.ll and adds new tests.

mareko updated this revision to Diff 97140.Apr 28 2017, 12:28 PM

Use isPhysRegUsed.

mareko added a comment.May 2 2017, 7:44 AM

Ping

arsenm added inline comments.May 2 2017, 11:10 AM
lib/Target/AMDGPU/SIMachineFunctionInfo.cpp
74

I still don't think you need this. You can directly set the input SGPR5 in the constructor here

mareko updated this revision to Diff 97483.May 2 2017, 12:13 PM

Directly set SGPR5 in the SIMachineFunctionInfo constructor.

Harbormaster completed remote builds in B6067: Diff 97483.May 2 2017, 12:13 PM
arsenm accepted this revision.May 4 2017, 10:39 AM

LGTM, except the test is much more complicated than it needs to be. Can have a small test with a volatile access to an alloca or something? These spill tests are big enough to slow down the test suite

This revision is now accepted and ready to land.May 4 2017, 10:39 AM
mareko added a comment.May 4 2017, 12:21 PM

Yeah, we could have a simpler test, but my approach is usually to copy an existing test and modify it, so it would help if we had such a test that screams "hey I'm a simple scratch buffer test!"

Closed by commit rL302200: AMDGPU: GFX9 GS and HS shaders always have the scratch wave offset in SGPR5 (authored by mareko). · Explain WhyMay 4 2017, 3:38 PM
This revision was automatically updated to reflect the committed changes.

Revision Contents

PathSize
lib/
Target/
AMDGPU/
3 lines
14 lines
8 lines
test/
CodeGen/
AMDGPU/
103 lines

Diff 97483

lib/Target/AMDGPU/SIFrameLowering.cpp

Show First 20 Lines • Show All 346 Lines • ▼ Show 20 Lines bool CopyBufferFirst = TRI->isSubRegisterEq(PreloadedPrivateBufferReg,
ScratchWaveOffsetReg); ScratchWaveOffsetReg);
if (CopyBuffer && CopyBufferFirst) { if (CopyBuffer && CopyBufferFirst) {
BuildMI(MBB, I, DL, TII->get(AMDGPU::COPY), ScratchRsrcReg) BuildMI(MBB, I, DL, TII->get(AMDGPU::COPY), ScratchRsrcReg)
.addReg(PreloadedPrivateBufferReg, RegState::Kill); .addReg(PreloadedPrivateBufferReg, RegState::Kill);
} }
if (OffsetRegUsed && if (OffsetRegUsed &&
PreloadedScratchWaveOffsetReg != ScratchWaveOffsetReg) { PreloadedScratchWaveOffsetReg != ScratchWaveOffsetReg) {
BuildMI(MBB, I, DL, TII->get(AMDGPU::COPY), ScratchWaveOffsetReg) BuildMI(MBB, I, DL, TII->get(AMDGPU::COPY), ScratchWaveOffsetReg)
.addReg(PreloadedScratchWaveOffsetReg, RegState::Kill); .addReg(PreloadedScratchWaveOffsetReg,
arsenmUnsubmitted
Not Done
ReplyInline Actions

This shouldn't be a recorded property of a function. Can you check MRI for whether the register has uses?

arsenm: This shouldn't be a recorded property of a function. Can you check MRI for whether the register…
MRI.isPhysRegUsed(ScratchWaveOffsetReg) ? 0 : RegState::Kill);
} }
if (CopyBuffer && !CopyBufferFirst) { if (CopyBuffer && !CopyBufferFirst) {
BuildMI(MBB, I, DL, TII->get(AMDGPU::COPY), ScratchRsrcReg) BuildMI(MBB, I, DL, TII->get(AMDGPU::COPY), ScratchRsrcReg)
.addReg(PreloadedPrivateBufferReg, RegState::Kill); .addReg(PreloadedPrivateBufferReg, RegState::Kill);
} }
if (ResourceRegUsed && (ST.isMesaGfxShader(MF) || (PreloadedPrivateBufferReg == AMDGPU::NoRegister))) { if (ResourceRegUsed && (ST.isMesaGfxShader(MF) || (PreloadedPrivateBufferReg == AMDGPU::NoRegister))) {
▲ Show 20 Lines • Show All 196 Lines • Show Last 20 Lines

lib/Target/AMDGPU/SIISelLowering.cpp

Show First 20 Lines • Show All 1,036 Lines • ▼ Show 20 Linesstatic void allocateHSAUserSGPRs(CCState &CCInfo,
// TODO: Add GridWorkGroupCount user SGPRs when used. For now with HSA we read // TODO: Add GridWorkGroupCount user SGPRs when used. For now with HSA we read
// these from the dispatch pointer. // these from the dispatch pointer.
} }
// Allocate special input registers that are initialized per-wave. // Allocate special input registers that are initialized per-wave.
static void allocateSystemSGPRs(CCState &CCInfo, static void allocateSystemSGPRs(CCState &CCInfo,
MachineFunction &MF, MachineFunction &MF,
SIMachineFunctionInfo &Info, SIMachineFunctionInfo &Info,
CallingConv::ID CallConv,
bool IsShader) { bool IsShader) {
if (Info.hasWorkGroupIDX()) { if (Info.hasWorkGroupIDX()) {
unsigned Reg = Info.addWorkGroupIDX(); unsigned Reg = Info.addWorkGroupIDX();
MF.addLiveIn(Reg, &AMDGPU::SReg_32_XM0RegClass); MF.addLiveIn(Reg, &AMDGPU::SReg_32_XM0RegClass);
CCInfo.AllocateReg(Reg); CCInfo.AllocateReg(Reg);
} }
if (Info.hasWorkGroupIDY()) { if (Info.hasWorkGroupIDY()) {
Show All 14 Lines if (Info.hasWorkGroupInfo()) {
CCInfo.AllocateReg(Reg); CCInfo.AllocateReg(Reg);
} }
if (Info.hasPrivateSegmentWaveByteOffset()) { if (Info.hasPrivateSegmentWaveByteOffset()) {
// Scratch wave offset passed in system SGPR. // Scratch wave offset passed in system SGPR.
unsigned PrivateSegmentWaveByteOffsetReg; unsigned PrivateSegmentWaveByteOffsetReg;
if (IsShader) { if (IsShader) {
PrivateSegmentWaveByteOffsetReg =
Info.getPrivateSegmentWaveByteOffsetSystemSGPR();
// This is true if the scratch wave byte offset doesn't have a fixed
// location.
if (PrivateSegmentWaveByteOffsetReg == AMDGPU::NoRegister) {
PrivateSegmentWaveByteOffsetReg = findFirstFreeSGPR(CCInfo); PrivateSegmentWaveByteOffsetReg = findFirstFreeSGPR(CCInfo);
Info.setPrivateSegmentWaveByteOffset(PrivateSegmentWaveByteOffsetReg); Info.setPrivateSegmentWaveByteOffset(PrivateSegmentWaveByteOffsetReg);
}
} else } else
PrivateSegmentWaveByteOffsetReg = Info.addPrivateSegmentWaveByteOffset(); PrivateSegmentWaveByteOffsetReg = Info.addPrivateSegmentWaveByteOffset();
MF.addLiveIn(PrivateSegmentWaveByteOffsetReg, &AMDGPU::SGPR_32RegClass); MF.addLiveIn(PrivateSegmentWaveByteOffsetReg, &AMDGPU::SGPR_32RegClass);
CCInfo.AllocateReg(PrivateSegmentWaveByteOffsetReg); CCInfo.AllocateReg(PrivateSegmentWaveByteOffsetReg);
} }
} }
▲ Show 20 Lines • Show All 220 Lines • ▼ Show 20 Lines if (Arg.VT.isVector()) {
continue; continue;
} }
InVals.push_back(Val); InVals.push_back(Val);
} }
// Start adding system SGPRs. // Start adding system SGPRs.
if (IsEntryFunc) if (IsEntryFunc)
allocateSystemSGPRs(CCInfo, MF, *Info, IsShader); allocateSystemSGPRs(CCInfo, MF, *Info, CallConv, IsShader);
reservePrivateMemoryRegs(getTargetMachine(), MF, *TRI, *Info); reservePrivateMemoryRegs(getTargetMachine(), MF, *TRI, *Info);
return Chains.empty() ? Chain : return Chains.empty() ? Chain :
DAG.getNode(ISD::TokenFactor, DL, MVT::Other, Chains); DAG.getNode(ISD::TokenFactor, DL, MVT::Other, Chains);
} }
SDValue SDValue
▲ Show 20 Lines • Show All 3,974 Lines • Show Last 20 Lines

lib/Target/AMDGPU/SIMachineFunctionInfo.cpp

Show First 20 Lines • Show All 65 Lines • ▼ Show 20 Lines : AMDGPUMachineFunction(MF),
GridWorkgroupCountX(false), GridWorkgroupCountX(false),
GridWorkgroupCountY(false), GridWorkgroupCountY(false),
GridWorkgroupCountZ(false), GridWorkgroupCountZ(false),
WorkGroupIDX(false), WorkGroupIDX(false),
WorkGroupIDY(false), WorkGroupIDY(false),
WorkGroupIDZ(false), WorkGroupIDZ(false),
WorkGroupInfo(false), WorkGroupInfo(false),
PrivateSegmentWaveByteOffset(false), PrivateSegmentWaveByteOffset(false),
WorkItemIDX(false), WorkItemIDX(false),
arsenmUnsubmitted
Not Done
ReplyInline Actions

I still don't think you need this. You can directly set the input SGPR5 in the constructor here

arsenm: I still don't think you need this. You can directly set the input SGPR5 in the constructor here
WorkItemIDY(false), WorkItemIDY(false),
WorkItemIDZ(false), WorkItemIDZ(false),
PrivateMemoryInputPtr(false) { PrivateMemoryInputPtr(false) {
const SISubtarget &ST = MF.getSubtarget<SISubtarget>(); const SISubtarget &ST = MF.getSubtarget<SISubtarget>();
const Function *F = MF.getFunction(); const Function *F = MF.getFunction();
FlatWorkGroupSizes = ST.getFlatWorkGroupSizes(*F); FlatWorkGroupSizes = ST.getFlatWorkGroupSizes(*F);
WavesPerEU = ST.getWavesPerEU(*F); WavesPerEU = ST.getWavesPerEU(*F);
Show All 34 LinesSIMachineFunctionInfo::SIMachineFunctionInfo(const MachineFunction &MF)
// enabled if Z is. // enabled if Z is.
if (WorkItemIDZ) if (WorkItemIDZ)
WorkItemIDY = true; WorkItemIDY = true;
const MachineFrameInfo &FrameInfo = MF.getFrameInfo(); const MachineFrameInfo &FrameInfo = MF.getFrameInfo();
bool MaySpill = ST.isVGPRSpillingEnabled(*F); bool MaySpill = ST.isVGPRSpillingEnabled(*F);
bool HasStackObjects = FrameInfo.hasStackObjects(); bool HasStackObjects = FrameInfo.hasStackObjects();
if (HasStackObjects || MaySpill) if (HasStackObjects || MaySpill) {
PrivateSegmentWaveByteOffset = true; PrivateSegmentWaveByteOffset = true;
// HS and GS always have the scratch wave offset in SGPR5 on GFX9.
if (ST.getGeneration() >= AMDGPUSubtarget::GFX9 &&
(CC == CallingConv::AMDGPU_HS || CC == CallingConv::AMDGPU_GS))
PrivateSegmentWaveByteOffsetSystemSGPR = AMDGPU::SGPR5;
}
if (ST.isAmdCodeObjectV2(MF)) { if (ST.isAmdCodeObjectV2(MF)) {
if (HasStackObjects || MaySpill) if (HasStackObjects || MaySpill)
PrivateSegmentBuffer = true; PrivateSegmentBuffer = true;
if (F->hasFnAttribute("amdgpu-dispatch-ptr")) if (F->hasFnAttribute("amdgpu-dispatch-ptr"))
DispatchPtr = true; DispatchPtr = true;
if (F->hasFnAttribute("amdgpu-queue-ptr")) if (F->hasFnAttribute("amdgpu-queue-ptr"))
▲ Show 20 Lines • Show All 123 Lines • Show Last 20 Lines

test/CodeGen/AMDGPU/local-stack-slot-bug.ll

  • This file was deleted.
; RUN: llc -march=amdgcn -mcpu=verde -mattr=+vgpr-spilling -verify-machineinstrs < %s | FileCheck %s
; RUN: llc -march=amdgcn -mcpu=tonga -mattr=-flat-for-global -mattr=+vgpr-spilling -verify-machineinstrs < %s | FileCheck %s
; This used to fail due to a v_add_i32 instruction with an illegal immediate
; operand that was created during Local Stack Slot Allocation. Test case derived
; from https://bugs.freedesktop.org/show_bug.cgi?id=96602
;
; CHECK-LABEL: {{^}}main:
; CHECK-DAG: v_mov_b32_e32 [[K:v[0-9]+]], 0x200
; CHECK-DAG: v_mov_b32_e32 [[ZERO:v[0-9]+]], 0x400{{$}}
; CHECK-DAG: v_lshlrev_b32_e32 [[BYTES:v[0-9]+]], 2, v0
; CHECK-DAG: v_and_b32_e32 [[CLAMP_IDX:v[0-9]+]], 0x1fc, [[BYTES]]
; CHECK-DAG: v_or_b32_e32 [[LO_OFF:v[0-9]+]], [[CLAMP_IDX]], [[K]]
; CHECK-DAG: v_or_b32_e32 [[HI_OFF:v[0-9]+]], [[CLAMP_IDX]], [[ZERO]]
; CHECK: buffer_load_dword {{v[0-9]+}}, [[LO_OFF]], {{s\[[0-9]+:[0-9]+\]}}, {{s[0-9]+}} offen
; CHECK: buffer_load_dword {{v[0-9]+}}, [[HI_OFF]], {{s\[[0-9]+:[0-9]+\]}}, {{s[0-9]+}} offen
define amdgpu_ps float @main(i32 %idx) {
main_body:
%v1 = extractelement <81 x float> <float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float 0x3FE41CFEA0000000, float 0xBFE7A693C0000000, float 0xBFEA477C60000000, float 0xBFEBE5DC60000000, float 0xBFEC71C720000000, float 0xBFEBE5DC60000000, float 0xBFEA477C60000000, float 0xBFE7A693C0000000, float 0xBFE41CFEA0000000, float 0x3FDF9B13E0000000, float 0x3FDF9B1380000000, float 0x3FD5C53B80000000, float 0x3FD5C53B00000000, float 0x3FC6326AC0000000, float 0x3FC63269E0000000, float 0xBEE05CEB00000000, float 0xBEE086A320000000, float 0xBFC63269E0000000, float 0xBFC6326AC0000000, float 0xBFD5C53B80000000, float 0xBFD5C53B80000000, float 0xBFDF9B13E0000000, float 0xBFDF9B1460000000, float 0xBFE41CFE80000000, float 0x3FE7A693C0000000, float 0x3FEA477C20000000, float 0x3FEBE5DC40000000, float 0x3FEC71C6E0000000, float 0x3FEBE5DC40000000, float 0x3FEA477C20000000, float 0x3FE7A693C0000000, float 0xBFE41CFE80000000>, i32 %idx
%v2 = extractelement <81 x float> <float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float 0xBFE41CFEA0000000, float 0xBFDF9B13E0000000, float 0xBFD5C53B80000000, float 0xBFC6326AC0000000, float 0x3EE0789320000000, float 0x3FC6326AC0000000, float 0x3FD5C53B80000000, float 0x3FDF9B13E0000000, float 0x3FE41CFEA0000000, float 0xBFE7A693C0000000, float 0x3FE7A693C0000000, float 0xBFEA477C20000000, float 0x3FEA477C20000000, float 0xBFEBE5DC40000000, float 0x3FEBE5DC40000000, float 0xBFEC71C720000000, float 0x3FEC71C6E0000000, float 0xBFEBE5DC60000000, float 0x3FEBE5DC40000000, float 0xBFEA477C20000000, float 0x3FEA477C20000000, float 0xBFE7A693C0000000, float 0x3FE7A69380000000, float 0xBFE41CFEA0000000, float 0xBFDF9B13E0000000, float 0xBFD5C53B80000000, float 0xBFC6326AC0000000, float 0x3EE0789320000000, float 0x3FC6326AC0000000, float 0x3FD5C53B80000000, float 0x3FDF9B13E0000000, float 0x3FE41CFE80000000>, i32 %idx
%r = fadd float %v1, %v2
ret float %r
}

test/CodeGen/AMDGPU/scratch-simple.ll

  • This file was added.
; RUN: llc -march=amdgcn -mcpu=verde -mattr=+vgpr-spilling -verify-machineinstrs < %s | FileCheck --check-prefix=GCN --check-prefix=SI %s
; RUN: llc -march=amdgcn -mcpu=tonga -mattr=-flat-for-global -mattr=+vgpr-spilling -verify-machineinstrs < %s | FileCheck --check-prefix=GCN --check-prefix=SI %s
; RUN: llc -march=amdgcn -mcpu=gfx900 -mattr=-flat-for-global -mattr=+vgpr-spilling -verify-machineinstrs < %s | FileCheck --check-prefix=GCN --check-prefix=GFX9 %s
; This used to fail due to a v_add_i32 instruction with an illegal immediate
; operand that was created during Local Stack Slot Allocation. Test case derived
; from https://bugs.freedesktop.org/show_bug.cgi?id=96602
;
arsenmUnsubmitted
Not Done
ReplyInline Actions

Copy paste from local-stack-slot-bug.ll.

It would probably be better to merge the two tests rathe than having the exact same function repeated in multiple files.

arsenm: Copy paste from local-stack-slot-bug.ll. It would probably be better to merge the two tests…
marekoAuthorUnsubmitted
Not Done
ReplyInline Actions

No. This patch renames local-stack-slot-bug.ll to scratch-simple.ll and adds new tests.

mareko: No. This patch renames local-stack-slot-bug.ll to scratch-simple.ll and adds new tests.
; GCN-LABEL: {{^}}ps_main:
; GCN-DAG: s_mov_b32 [[SWO:s[0-9]+]], s0
; GCN-DAG: v_mov_b32_e32 [[K:v[0-9]+]], 0x200
; GCN-DAG: v_mov_b32_e32 [[ZERO:v[0-9]+]], 0x400{{$}}
; GCN-DAG: v_lshlrev_b32_e32 [[BYTES:v[0-9]+]], 2, v0
; GCN-DAG: v_and_b32_e32 [[CLAMP_IDX:v[0-9]+]], 0x1fc, [[BYTES]]
; GCN-DAG: v_or_b32_e32 [[LO_OFF:v[0-9]+]], [[CLAMP_IDX]], [[K]]
; GCN-DAG: v_or_b32_e32 [[HI_OFF:v[0-9]+]], [[CLAMP_IDX]], [[ZERO]]
; GCN: buffer_load_dword {{v[0-9]+}}, [[LO_OFF]], {{s\[[0-9]+:[0-9]+\]}}, [[SWO]] offen
; GCN: buffer_load_dword {{v[0-9]+}}, [[HI_OFF]], {{s\[[0-9]+:[0-9]+\]}}, [[SWO]] offen
define amdgpu_ps float @ps_main(i32 %idx) {
%v1 = extractelement <81 x float> <float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float 0x3FE41CFEA0000000, float 0xBFE7A693C0000000, float 0xBFEA477C60000000, float 0xBFEBE5DC60000000, float 0xBFEC71C720000000, float 0xBFEBE5DC60000000, float 0xBFEA477C60000000, float 0xBFE7A693C0000000, float 0xBFE41CFEA0000000, float 0x3FDF9B13E0000000, float 0x3FDF9B1380000000, float 0x3FD5C53B80000000, float 0x3FD5C53B00000000, float 0x3FC6326AC0000000, float 0x3FC63269E0000000, float 0xBEE05CEB00000000, float 0xBEE086A320000000, float 0xBFC63269E0000000, float 0xBFC6326AC0000000, float 0xBFD5C53B80000000, float 0xBFD5C53B80000000, float 0xBFDF9B13E0000000, float 0xBFDF9B1460000000, float 0xBFE41CFE80000000, float 0x3FE7A693C0000000, float 0x3FEA477C20000000, float 0x3FEBE5DC40000000, float 0x3FEC71C6E0000000, float 0x3FEBE5DC40000000, float 0x3FEA477C20000000, float 0x3FE7A693C0000000, float 0xBFE41CFE80000000>, i32 %idx
%v2 = extractelement <81 x float> <float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float 0xBFE41CFEA0000000, float 0xBFDF9B13E0000000, float 0xBFD5C53B80000000, float 0xBFC6326AC0000000, float 0x3EE0789320000000, float 0x3FC6326AC0000000, float 0x3FD5C53B80000000, float 0x3FDF9B13E0000000, float 0x3FE41CFEA0000000, float 0xBFE7A693C0000000, float 0x3FE7A693C0000000, float 0xBFEA477C20000000, float 0x3FEA477C20000000, float 0xBFEBE5DC40000000, float 0x3FEBE5DC40000000, float 0xBFEC71C720000000, float 0x3FEC71C6E0000000, float 0xBFEBE5DC60000000, float 0x3FEBE5DC40000000, float 0xBFEA477C20000000, float 0x3FEA477C20000000, float 0xBFE7A693C0000000, float 0x3FE7A69380000000, float 0xBFE41CFEA0000000, float 0xBFDF9B13E0000000, float 0xBFD5C53B80000000, float 0xBFC6326AC0000000, float 0x3EE0789320000000, float 0x3FC6326AC0000000, float 0x3FD5C53B80000000, float 0x3FDF9B13E0000000, float 0x3FE41CFE80000000>, i32 %idx
%r = fadd float %v1, %v2
ret float %r
}
; GCN-LABEL: {{^}}vs_main:
; GCN: s_mov_b32 [[SWO:s[0-9]+]], s0
; GCN: buffer_load_dword {{v[0-9]+}}, {{v[0-9]+}}, {{s\[[0-9]+:[0-9]+\]}}, [[SWO]] offen
; GCN: buffer_load_dword {{v[0-9]+}}, {{v[0-9]+}}, {{s\[[0-9]+:[0-9]+\]}}, [[SWO]] offen
define amdgpu_vs float @vs_main(i32 %idx) {
%v1 = extractelement <81 x float> <float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float 0x3FE41CFEA0000000, float 0xBFE7A693C0000000, float 0xBFEA477C60000000, float 0xBFEBE5DC60000000, float 0xBFEC71C720000000, float 0xBFEBE5DC60000000, float 0xBFEA477C60000000, float 0xBFE7A693C0000000, float 0xBFE41CFEA0000000, float 0x3FDF9B13E0000000, float 0x3FDF9B1380000000, float 0x3FD5C53B80000000, float 0x3FD5C53B00000000, float 0x3FC6326AC0000000, float 0x3FC63269E0000000, float 0xBEE05CEB00000000, float 0xBEE086A320000000, float 0xBFC63269E0000000, float 0xBFC6326AC0000000, float 0xBFD5C53B80000000, float 0xBFD5C53B80000000, float 0xBFDF9B13E0000000, float 0xBFDF9B1460000000, float 0xBFE41CFE80000000, float 0x3FE7A693C0000000, float 0x3FEA477C20000000, float 0x3FEBE5DC40000000, float 0x3FEC71C6E0000000, float 0x3FEBE5DC40000000, float 0x3FEA477C20000000, float 0x3FE7A693C0000000, float 0xBFE41CFE80000000>, i32 %idx
%v2 = extractelement <81 x float> <float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float 0xBFE41CFEA0000000, float 0xBFDF9B13E0000000, float 0xBFD5C53B80000000, float 0xBFC6326AC0000000, float 0x3EE0789320000000, float 0x3FC6326AC0000000, float 0x3FD5C53B80000000, float 0x3FDF9B13E0000000, float 0x3FE41CFEA0000000, float 0xBFE7A693C0000000, float 0x3FE7A693C0000000, float 0xBFEA477C20000000, float 0x3FEA477C20000000, float 0xBFEBE5DC40000000, float 0x3FEBE5DC40000000, float 0xBFEC71C720000000, float 0x3FEC71C6E0000000, float 0xBFEBE5DC60000000, float 0x3FEBE5DC40000000, float 0xBFEA477C20000000, float 0x3FEA477C20000000, float 0xBFE7A693C0000000, float 0x3FE7A69380000000, float 0xBFE41CFEA0000000, float 0xBFDF9B13E0000000, float 0xBFD5C53B80000000, float 0xBFC6326AC0000000, float 0x3EE0789320000000, float 0x3FC6326AC0000000, float 0x3FD5C53B80000000, float 0x3FDF9B13E0000000, float 0x3FE41CFE80000000>, i32 %idx
%r = fadd float %v1, %v2
ret float %r
}
; GCN-LABEL: {{^}}cs_main:
; GCN: s_mov_b32 [[SWO:s[0-9]+]], s0
; GCN: buffer_load_dword {{v[0-9]+}}, {{v[0-9]+}}, {{s\[[0-9]+:[0-9]+\]}}, [[SWO]] offen
; GCN: buffer_load_dword {{v[0-9]+}}, {{v[0-9]+}}, {{s\[[0-9]+:[0-9]+\]}}, [[SWO]] offen
define amdgpu_cs float @cs_main(i32 %idx) {
%v1 = extractelement <81 x float> <float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float 0x3FE41CFEA0000000, float 0xBFE7A693C0000000, float 0xBFEA477C60000000, float 0xBFEBE5DC60000000, float 0xBFEC71C720000000, float 0xBFEBE5DC60000000, float 0xBFEA477C60000000, float 0xBFE7A693C0000000, float 0xBFE41CFEA0000000, float 0x3FDF9B13E0000000, float 0x3FDF9B1380000000, float 0x3FD5C53B80000000, float 0x3FD5C53B00000000, float 0x3FC6326AC0000000, float 0x3FC63269E0000000, float 0xBEE05CEB00000000, float 0xBEE086A320000000, float 0xBFC63269E0000000, float 0xBFC6326AC0000000, float 0xBFD5C53B80000000, float 0xBFD5C53B80000000, float 0xBFDF9B13E0000000, float 0xBFDF9B1460000000, float 0xBFE41CFE80000000, float 0x3FE7A693C0000000, float 0x3FEA477C20000000, float 0x3FEBE5DC40000000, float 0x3FEC71C6E0000000, float 0x3FEBE5DC40000000, float 0x3FEA477C20000000, float 0x3FE7A693C0000000, float 0xBFE41CFE80000000>, i32 %idx
%v2 = extractelement <81 x float> <float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float 0xBFE41CFEA0000000, float 0xBFDF9B13E0000000, float 0xBFD5C53B80000000, float 0xBFC6326AC0000000, float 0x3EE0789320000000, float 0x3FC6326AC0000000, float 0x3FD5C53B80000000, float 0x3FDF9B13E0000000, float 0x3FE41CFEA0000000, float 0xBFE7A693C0000000, float 0x3FE7A693C0000000, float 0xBFEA477C20000000, float 0x3FEA477C20000000, float 0xBFEBE5DC40000000, float 0x3FEBE5DC40000000, float 0xBFEC71C720000000, float 0x3FEC71C6E0000000, float 0xBFEBE5DC60000000, float 0x3FEBE5DC40000000, float 0xBFEA477C20000000, float 0x3FEA477C20000000, float 0xBFE7A693C0000000, float 0x3FE7A69380000000, float 0xBFE41CFEA0000000, float 0xBFDF9B13E0000000, float 0xBFD5C53B80000000, float 0xBFC6326AC0000000, float 0x3EE0789320000000, float 0x3FC6326AC0000000, float 0x3FD5C53B80000000, float 0x3FDF9B13E0000000, float 0x3FE41CFE80000000>, i32 %idx
%r = fadd float %v1, %v2
ret float %r
}
; GCN-LABEL: {{^}}hs_main:
; SI: s_mov_b32 [[SWO:s[0-9]+]], s0
; GFX9: s_mov_b32 [[SWO:s[0-9]+]], s5
; GCN: buffer_load_dword {{v[0-9]+}}, {{v[0-9]+}}, {{s\[[0-9]+:[0-9]+\]}}, [[SWO]] offen
; GCN: buffer_load_dword {{v[0-9]+}}, {{v[0-9]+}}, {{s\[[0-9]+:[0-9]+\]}}, [[SWO]] offen
define amdgpu_hs float @hs_main(i32 %idx) {
%v1 = extractelement <81 x float> <float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float 0x3FE41CFEA0000000, float 0xBFE7A693C0000000, float 0xBFEA477C60000000, float 0xBFEBE5DC60000000, float 0xBFEC71C720000000, float 0xBFEBE5DC60000000, float 0xBFEA477C60000000, float 0xBFE7A693C0000000, float 0xBFE41CFEA0000000, float 0x3FDF9B13E0000000, float 0x3FDF9B1380000000, float 0x3FD5C53B80000000, float 0x3FD5C53B00000000, float 0x3FC6326AC0000000, float 0x3FC63269E0000000, float 0xBEE05CEB00000000, float 0xBEE086A320000000, float 0xBFC63269E0000000, float 0xBFC6326AC0000000, float 0xBFD5C53B80000000, float 0xBFD5C53B80000000, float 0xBFDF9B13E0000000, float 0xBFDF9B1460000000, float 0xBFE41CFE80000000, float 0x3FE7A693C0000000, float 0x3FEA477C20000000, float 0x3FEBE5DC40000000, float 0x3FEC71C6E0000000, float 0x3FEBE5DC40000000, float 0x3FEA477C20000000, float 0x3FE7A693C0000000, float 0xBFE41CFE80000000>, i32 %idx
%v2 = extractelement <81 x float> <float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float 0xBFE41CFEA0000000, float 0xBFDF9B13E0000000, float 0xBFD5C53B80000000, float 0xBFC6326AC0000000, float 0x3EE0789320000000, float 0x3FC6326AC0000000, float 0x3FD5C53B80000000, float 0x3FDF9B13E0000000, float 0x3FE41CFEA0000000, float 0xBFE7A693C0000000, float 0x3FE7A693C0000000, float 0xBFEA477C20000000, float 0x3FEA477C20000000, float 0xBFEBE5DC40000000, float 0x3FEBE5DC40000000, float 0xBFEC71C720000000, float 0x3FEC71C6E0000000, float 0xBFEBE5DC60000000, float 0x3FEBE5DC40000000, float 0xBFEA477C20000000, float 0x3FEA477C20000000, float 0xBFE7A693C0000000, float 0x3FE7A69380000000, float 0xBFE41CFEA0000000, float 0xBFDF9B13E0000000, float 0xBFD5C53B80000000, float 0xBFC6326AC0000000, float 0x3EE0789320000000, float 0x3FC6326AC0000000, float 0x3FD5C53B80000000, float 0x3FDF9B13E0000000, float 0x3FE41CFE80000000>, i32 %idx
%r = fadd float %v1, %v2
ret float %r
}
; GCN-LABEL: {{^}}gs_main:
; SI: s_mov_b32 [[SWO:s[0-9]+]], s0
; GFX9: s_mov_b32 [[SWO:s[0-9]+]], s5
; GCN: buffer_load_dword {{v[0-9]+}}, {{v[0-9]+}}, {{s\[[0-9]+:[0-9]+\]}}, [[SWO]] offen
; GCN: buffer_load_dword {{v[0-9]+}}, {{v[0-9]+}}, {{s\[[0-9]+:[0-9]+\]}}, [[SWO]] offen
define amdgpu_gs float @gs_main(i32 %idx) {
%v1 = extractelement <81 x float> <float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float 0x3FE41CFEA0000000, float 0xBFE7A693C0000000, float 0xBFEA477C60000000, float 0xBFEBE5DC60000000, float 0xBFEC71C720000000, float 0xBFEBE5DC60000000, float 0xBFEA477C60000000, float 0xBFE7A693C0000000, float 0xBFE41CFEA0000000, float 0x3FDF9B13E0000000, float 0x3FDF9B1380000000, float 0x3FD5C53B80000000, float 0x3FD5C53B00000000, float 0x3FC6326AC0000000, float 0x3FC63269E0000000, float 0xBEE05CEB00000000, float 0xBEE086A320000000, float 0xBFC63269E0000000, float 0xBFC6326AC0000000, float 0xBFD5C53B80000000, float 0xBFD5C53B80000000, float 0xBFDF9B13E0000000, float 0xBFDF9B1460000000, float 0xBFE41CFE80000000, float 0x3FE7A693C0000000, float 0x3FEA477C20000000, float 0x3FEBE5DC40000000, float 0x3FEC71C6E0000000, float 0x3FEBE5DC40000000, float 0x3FEA477C20000000, float 0x3FE7A693C0000000, float 0xBFE41CFE80000000>, i32 %idx
%v2 = extractelement <81 x float> <float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float 0xBFE41CFEA0000000, float 0xBFDF9B13E0000000, float 0xBFD5C53B80000000, float 0xBFC6326AC0000000, float 0x3EE0789320000000, float 0x3FC6326AC0000000, float 0x3FD5C53B80000000, float 0x3FDF9B13E0000000, float 0x3FE41CFEA0000000, float 0xBFE7A693C0000000, float 0x3FE7A693C0000000, float 0xBFEA477C20000000, float 0x3FEA477C20000000, float 0xBFEBE5DC40000000, float 0x3FEBE5DC40000000, float 0xBFEC71C720000000, float 0x3FEC71C6E0000000, float 0xBFEBE5DC60000000, float 0x3FEBE5DC40000000, float 0xBFEA477C20000000, float 0x3FEA477C20000000, float 0xBFE7A693C0000000, float 0x3FE7A69380000000, float 0xBFE41CFEA0000000, float 0xBFDF9B13E0000000, float 0xBFD5C53B80000000, float 0xBFC6326AC0000000, float 0x3EE0789320000000, float 0x3FC6326AC0000000, float 0x3FD5C53B80000000, float 0x3FDF9B13E0000000, float 0x3FE41CFE80000000>, i32 %idx
%r = fadd float %v1, %v2
ret float %r
}
; GCN-LABEL: {{^}}hs_ir_uses_scratch_offset:
; SI: s_mov_b32 [[SWO:s[0-9]+]], s6
; GFX9: s_mov_b32 [[SWO:s[0-9]+]], s5
; GCN: buffer_load_dword {{v[0-9]+}}, {{v[0-9]+}}, {{s\[[0-9]+:[0-9]+\]}}, [[SWO]] offen
; GCN: buffer_load_dword {{v[0-9]+}}, {{v[0-9]+}}, {{s\[[0-9]+:[0-9]+\]}}, [[SWO]] offen
; GCN: s_mov_b32 s2, s5
define amdgpu_hs <{i32, i32, i32, float}> @hs_ir_uses_scratch_offset(i32 inreg, i32 inreg, i32 inreg, i32 inreg, i32 inreg, i32 inreg %swo, i32 %idx) {
%v1 = extractelement <81 x float> <float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float 0x3FE41CFEA0000000, float 0xBFE7A693C0000000, float 0xBFEA477C60000000, float 0xBFEBE5DC60000000, float 0xBFEC71C720000000, float 0xBFEBE5DC60000000, float 0xBFEA477C60000000, float 0xBFE7A693C0000000, float 0xBFE41CFEA0000000, float 0x3FDF9B13E0000000, float 0x3FDF9B1380000000, float 0x3FD5C53B80000000, float 0x3FD5C53B00000000, float 0x3FC6326AC0000000, float 0x3FC63269E0000000, float 0xBEE05CEB00000000, float 0xBEE086A320000000, float 0xBFC63269E0000000, float 0xBFC6326AC0000000, float 0xBFD5C53B80000000, float 0xBFD5C53B80000000, float 0xBFDF9B13E0000000, float 0xBFDF9B1460000000, float 0xBFE41CFE80000000, float 0x3FE7A693C0000000, float 0x3FEA477C20000000, float 0x3FEBE5DC40000000, float 0x3FEC71C6E0000000, float 0x3FEBE5DC40000000, float 0x3FEA477C20000000, float 0x3FE7A693C0000000, float 0xBFE41CFE80000000>, i32 %idx
%v2 = extractelement <81 x float> <float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float 0xBFE41CFEA0000000, float 0xBFDF9B13E0000000, float 0xBFD5C53B80000000, float 0xBFC6326AC0000000, float 0x3EE0789320000000, float 0x3FC6326AC0000000, float 0x3FD5C53B80000000, float 0x3FDF9B13E0000000, float 0x3FE41CFEA0000000, float 0xBFE7A693C0000000, float 0x3FE7A693C0000000, float 0xBFEA477C20000000, float 0x3FEA477C20000000, float 0xBFEBE5DC40000000, float 0x3FEBE5DC40000000, float 0xBFEC71C720000000, float 0x3FEC71C6E0000000, float 0xBFEBE5DC60000000, float 0x3FEBE5DC40000000, float 0xBFEA477C20000000, float 0x3FEA477C20000000, float 0xBFE7A693C0000000, float 0x3FE7A69380000000, float 0xBFE41CFEA0000000, float 0xBFDF9B13E0000000, float 0xBFD5C53B80000000, float 0xBFC6326AC0000000, float 0x3EE0789320000000, float 0x3FC6326AC0000000, float 0x3FD5C53B80000000, float 0x3FDF9B13E0000000, float 0x3FE41CFE80000000>, i32 %idx
%f = fadd float %v1, %v2
%r1 = insertvalue <{i32, i32, i32, float}> undef, i32 %swo, 2
%r2 = insertvalue <{i32, i32, i32, float}> %r1, float %f, 3
ret <{i32, i32, i32, float}> %r2
}
; GCN-LABEL: {{^}}gs_ir_uses_scratch_offset:
; SI: s_mov_b32 [[SWO:s[0-9]+]], s6
; GFX9: s_mov_b32 [[SWO:s[0-9]+]], s5
; GCN: buffer_load_dword {{v[0-9]+}}, {{v[0-9]+}}, {{s\[[0-9]+:[0-9]+\]}}, [[SWO]] offen
; GCN: buffer_load_dword {{v[0-9]+}}, {{v[0-9]+}}, {{s\[[0-9]+:[0-9]+\]}}, [[SWO]] offen
; GCN: s_mov_b32 s2, s5
define amdgpu_gs <{i32, i32, i32, float}> @gs_ir_uses_scratch_offset(i32 inreg, i32 inreg, i32 inreg, i32 inreg, i32 inreg, i32 inreg %swo, i32 %idx) {
%v1 = extractelement <81 x float> <float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float 0x3FE41CFEA0000000, float 0xBFE7A693C0000000, float 0xBFEA477C60000000, float 0xBFEBE5DC60000000, float 0xBFEC71C720000000, float 0xBFEBE5DC60000000, float 0xBFEA477C60000000, float 0xBFE7A693C0000000, float 0xBFE41CFEA0000000, float 0x3FDF9B13E0000000, float 0x3FDF9B1380000000, float 0x3FD5C53B80000000, float 0x3FD5C53B00000000, float 0x3FC6326AC0000000, float 0x3FC63269E0000000, float 0xBEE05CEB00000000, float 0xBEE086A320000000, float 0xBFC63269E0000000, float 0xBFC6326AC0000000, float 0xBFD5C53B80000000, float 0xBFD5C53B80000000, float 0xBFDF9B13E0000000, float 0xBFDF9B1460000000, float 0xBFE41CFE80000000, float 0x3FE7A693C0000000, float 0x3FEA477C20000000, float 0x3FEBE5DC40000000, float 0x3FEC71C6E0000000, float 0x3FEBE5DC40000000, float 0x3FEA477C20000000, float 0x3FE7A693C0000000, float 0xBFE41CFE80000000>, i32 %idx
%v2 = extractelement <81 x float> <float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float undef, float 0xBFE41CFEA0000000, float 0xBFDF9B13E0000000, float 0xBFD5C53B80000000, float 0xBFC6326AC0000000, float 0x3EE0789320000000, float 0x3FC6326AC0000000, float 0x3FD5C53B80000000, float 0x3FDF9B13E0000000, float 0x3FE41CFEA0000000, float 0xBFE7A693C0000000, float 0x3FE7A693C0000000, float 0xBFEA477C20000000, float 0x3FEA477C20000000, float 0xBFEBE5DC40000000, float 0x3FEBE5DC40000000, float 0xBFEC71C720000000, float 0x3FEC71C6E0000000, float 0xBFEBE5DC60000000, float 0x3FEBE5DC40000000, float 0xBFEA477C20000000, float 0x3FEA477C20000000, float 0xBFE7A693C0000000, float 0x3FE7A69380000000, float 0xBFE41CFEA0000000, float 0xBFDF9B13E0000000, float 0xBFD5C53B80000000, float 0xBFC6326AC0000000, float 0x3EE0789320000000, float 0x3FC6326AC0000000, float 0x3FD5C53B80000000, float 0x3FDF9B13E0000000, float 0x3FE41CFE80000000>, i32 %idx
%f = fadd float %v1, %v2
%r1 = insertvalue <{i32, i32, i32, float}> undef, i32 %swo, 2
%r2 = insertvalue <{i32, i32, i32, float}> %r1, float %f, 3
ret <{i32, i32, i32, float}> %r2
}
lib/Target/AMDGPU/SIMachineFunctionInfo.cpp