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
|
//===---- X86FixupSetCC.cpp - optimize usage of LEA instructions ----------===//
//
// 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
//
//===----------------------------------------------------------------------===//
//
// This file defines a pass that fixes zero-extension of setcc patterns.
// X86 setcc instructions are modeled to have no input arguments, and a single
// GR8 output argument. This is consistent with other similar instructions
// (e.g. movb), but means it is impossible to directly generate a setcc into
// the lower GR8 of a specified GR32.
// This means that ISel must select (zext (setcc)) into something like
// seta %al; movzbl %al, %eax.
// Unfortunately, this can cause a stall due to the partial register write
// performed by the setcc. Instead, we can use:
// xor %eax, %eax; seta %al
// This both avoids the stall, and encodes shorter.
//===----------------------------------------------------------------------===//
#include "X86.h"
#include "X86InstrInfo.h"
#include "X86Subtarget.h"
#include "llvm/ADT/Statistic.h"
#include "llvm/CodeGen/MachineFunctionPass.h"
#include "llvm/CodeGen/MachineInstrBuilder.h"
#include "llvm/CodeGen/MachineRegisterInfo.h"
using namespace llvm;
#define DEBUG_TYPE "x86-fixup-setcc"
STATISTIC(NumSubstZexts, "Number of setcc + zext pairs substituted");
namespace {
class X86FixupSetCCPass : public MachineFunctionPass {
public:
static char ID;
X86FixupSetCCPass() : MachineFunctionPass(ID) {}
StringRef getPassName() const override { return "X86 Fixup SetCC"; }
bool runOnMachineFunction(MachineFunction &MF) override;
private:
MachineRegisterInfo *MRI = nullptr;
const X86InstrInfo *TII = nullptr;
enum { SearchBound = 16 };
};
} // end anonymous namespace
char X86FixupSetCCPass::ID = 0;
INITIALIZE_PASS(X86FixupSetCCPass, DEBUG_TYPE, DEBUG_TYPE, false, false)
FunctionPass *llvm::createX86FixupSetCC() { return new X86FixupSetCCPass(); }
bool X86FixupSetCCPass::runOnMachineFunction(MachineFunction &MF) {
bool Changed = false;
MRI = &MF.getRegInfo();
TII = MF.getSubtarget<X86Subtarget>().getInstrInfo();
SmallVector<MachineInstr*, 4> ToErase;
for (auto &MBB : MF) {
MachineInstr *FlagsDefMI = nullptr;
for (auto &MI : MBB) {
// Remember the most recent preceding eflags defining instruction.
if (MI.definesRegister(X86::EFLAGS))
FlagsDefMI = &MI;
// Find a setcc that is used by a zext.
// This doesn't have to be the only use, the transformation is safe
// regardless.
if (MI.getOpcode() != X86::SETCCr)
continue;
MachineInstr *ZExt = nullptr;
for (auto &Use : MRI->use_instructions(MI.getOperand(0).getReg()))
if (Use.getOpcode() == X86::MOVZX32rr8)
ZExt = &Use;
if (!ZExt)
continue;
if (!FlagsDefMI)
continue;
// We'd like to put something that clobbers eflags directly before
// FlagsDefMI. This can't hurt anything after FlagsDefMI, because
// it, itself, by definition, clobbers eflags. But it may happen that
// FlagsDefMI also *uses* eflags, in which case the transformation is
// invalid.
if (FlagsDefMI->readsRegister(X86::EFLAGS))
continue;
// On 32-bit, we need to be careful to force an ABCD register.
const TargetRegisterClass *RC = MF.getSubtarget<X86Subtarget>().is64Bit()
? &X86::GR32RegClass
: &X86::GR32_ABCDRegClass;
if (!MRI->constrainRegClass(ZExt->getOperand(0).getReg(), RC)) {
// If we cannot constrain the register, we would need an additional copy
// and are better off keeping the MOVZX32rr8 we have now.
continue;
}
++NumSubstZexts;
Changed = true;
// Initialize a register with 0. This must go before the eflags def
Register ZeroReg = MRI->createVirtualRegister(RC);
BuildMI(MBB, FlagsDefMI, MI.getDebugLoc(), TII->get(X86::MOV32r0),
ZeroReg);
// X86 setcc only takes an output GR8, so fake a GR32 input by inserting
// the setcc result into the low byte of the zeroed register.
BuildMI(*ZExt->getParent(), ZExt, ZExt->getDebugLoc(),
TII->get(X86::INSERT_SUBREG), ZExt->getOperand(0).getReg())
.addReg(ZeroReg)
.addReg(MI.getOperand(0).getReg())
.addImm(X86::sub_8bit);
ToErase.push_back(ZExt);
}
}
for (auto &I : ToErase)
I->eraseFromParent();
return Changed;
}
|
