1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
|
//===- LICMTest.cpp - LICM unit tests -------------------------------------===//
//
// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
// See https://llvm.org/LICENSE.txt for license information.
// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
//
//===----------------------------------------------------------------------===//
#include "llvm/Analysis/TargetTransformInfo.h"
#include "llvm/AsmParser/Parser.h"
#include "llvm/IR/Module.h"
#include "llvm/MC/TargetRegistry.h"
#include "llvm/Passes/PassBuilder.h"
#include "llvm/Support/TargetSelect.h"
#include "llvm/Target/TargetMachine.h"
#include "llvm/Transforms/InstCombine/InstCombine.h"
#include "gtest/gtest.h"
#include <random>
namespace llvm {
static std::unique_ptr<LLVMTargetMachine> initTM() {
LLVMInitializeX86TargetInfo();
LLVMInitializeX86Target();
LLVMInitializeX86TargetMC();
auto TT(Triple::normalize("x86_64--"));
std::string Error;
const Target *TheTarget = TargetRegistry::lookupTarget(TT, Error);
return std::unique_ptr<LLVMTargetMachine>(static_cast<LLVMTargetMachine *>(
TheTarget->createTargetMachine(TT, "", "", TargetOptions(), std::nullopt,
std::nullopt, CodeGenOptLevel::Default)));
}
struct TernTester {
unsigned NElem;
unsigned ElemWidth;
std::mt19937_64 Rng;
unsigned ImmVal;
SmallVector<uint64_t, 16> VecElems[3];
void updateImm(uint8_t NewImmVal) { ImmVal = NewImmVal; }
void updateNElem(unsigned NewNElem) {
NElem = NewNElem;
for (unsigned I = 0; I < 3; ++I) {
VecElems[I].resize(NElem);
}
}
void updateElemWidth(unsigned NewElemWidth) {
ElemWidth = NewElemWidth;
assert(ElemWidth == 32 || ElemWidth == 64);
}
uint64_t getElemMask() const {
return (~uint64_t(0)) >> ((ElemWidth - 0) % 64);
}
void RandomizeVecArgs() {
uint64_t ElemMask = getElemMask();
for (unsigned I = 0; I < 3; ++I) {
for (unsigned J = 0; J < NElem; ++J) {
VecElems[I][J] = Rng() & ElemMask;
}
}
}
std::pair<std::string, std::string> getScalarInfo() const {
switch (ElemWidth) {
case 32:
return {"i32", "d"};
case 64:
return {"i64", "q"};
default:
llvm_unreachable("Invalid ElemWidth");
}
}
std::string getScalarType() const { return getScalarInfo().first; }
std::string getScalarExt() const { return getScalarInfo().second; }
std::string getVecType() const {
return "<" + Twine(NElem).str() + " x " + getScalarType() + ">";
};
std::string getVecWidth() const { return Twine(NElem * ElemWidth).str(); }
std::string getFunctionName() const {
return "@llvm.x86.avx512.pternlog." + getScalarExt() + "." + getVecWidth();
}
std::string getFunctionDecl() const {
return "declare " + getVecType() + getFunctionName() + "(" + getVecType() +
", " + getVecType() + ", " + getVecType() + ", " + "i32 immarg)";
}
std::string getVecN(unsigned N) const {
assert(N < 3);
std::string VecStr = getVecType() + " <";
for (unsigned I = 0; I < VecElems[N].size(); ++I) {
if (I != 0)
VecStr += ", ";
VecStr += getScalarType() + " " + Twine(VecElems[N][I]).str();
}
return VecStr + ">";
}
std::string getFunctionCall() const {
return "tail call " + getVecType() + " " + getFunctionName() + "(" +
getVecN(0) + ", " + getVecN(1) + ", " + getVecN(2) + ", " + "i32 " +
Twine(ImmVal).str() + ")";
}
std::string getTestText() const {
return getFunctionDecl() + "\ndefine " + getVecType() +
"@foo() {\n%r = " + getFunctionCall() + "\nret " + getVecType() +
" %r\n}\n";
}
void checkResult(const Value *V) {
auto GetValElem = [&](unsigned Idx) -> uint64_t {
if (auto *CV = dyn_cast<ConstantDataVector>(V))
return CV->getElementAsInteger(Idx);
auto *C = dyn_cast<Constant>(V);
assert(C);
if (C->isNullValue())
return 0;
if (C->isAllOnesValue())
return ((~uint64_t(0)) >> (ElemWidth % 64));
if (C->isOneValue())
return 1;
llvm_unreachable("Unknown constant type");
};
auto ComputeBit = [&](uint64_t A, uint64_t B, uint64_t C) -> uint64_t {
unsigned BitIdx = ((A & 1) << 2) | ((B & 1) << 1) | (C & 1);
return (ImmVal >> BitIdx) & 1;
};
for (unsigned I = 0; I < NElem; ++I) {
uint64_t Expec = 0;
uint64_t AEle = VecElems[0][I];
uint64_t BEle = VecElems[1][I];
uint64_t CEle = VecElems[2][I];
for (unsigned J = 0; J < ElemWidth; ++J) {
Expec |= ComputeBit(AEle >> J, BEle >> J, CEle >> J) << J;
}
ASSERT_EQ(Expec, GetValElem(I));
}
}
void check(LLVMContext &Ctx, FunctionPassManager &FPM,
FunctionAnalysisManager &FAM) {
SMDiagnostic Error;
std::unique_ptr<Module> M = parseAssemblyString(getTestText(), Error, Ctx);
ASSERT_TRUE(M);
Function *F = M->getFunction("foo");
ASSERT_TRUE(F);
ASSERT_EQ(F->getInstructionCount(), 2u);
FPM.run(*F, FAM);
ASSERT_EQ(F->getInstructionCount(), 1u);
ASSERT_EQ(F->size(), 1u);
const Instruction *I = F->begin()->getTerminator();
ASSERT_TRUE(I);
ASSERT_EQ(I->getNumOperands(), 1u);
checkResult(I->getOperand(0));
}
};
TEST(TernlogTest, TestConstantFolding) {
LLVMContext Ctx;
FunctionAnalysisManager FAM;
FunctionPassManager FPM;
PassBuilder PB;
LoopAnalysisManager LAM;
CGSCCAnalysisManager CGAM;
ModuleAnalysisManager MAM;
TargetIRAnalysis TIRA = TargetIRAnalysis(
[&](const Function &F) { return initTM()->getTargetTransformInfo(F); });
FAM.registerPass([&] { return TIRA; });
PB.registerModuleAnalyses(MAM);
PB.registerCGSCCAnalyses(CGAM);
PB.registerFunctionAnalyses(FAM);
PB.registerLoopAnalyses(LAM);
PB.crossRegisterProxies(LAM, FAM, CGAM, MAM);
FPM.addPass(InstCombinePass());
TernTester TT;
for (unsigned NElem = 2; NElem < 16; NElem += NElem) {
TT.updateNElem(NElem);
for (unsigned ElemWidth = 32; ElemWidth <= 64; ElemWidth += ElemWidth) {
if (ElemWidth * NElem > 512 || ElemWidth * NElem < 128)
continue;
TT.updateElemWidth(ElemWidth);
TT.RandomizeVecArgs();
for (unsigned Imm = 0; Imm < 256; ++Imm) {
TT.updateImm(Imm);
TT.check(Ctx, FPM, FAM);
}
}
}
}
} // namespace llvm
|
