catfood/CompilerSysY
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

machine code gen part 1

Browse Source
This commit is contained in:
ridethepig 2023-05-23 23:38:48 +08:00
parent a2d0c211a4
commit 67a32e446e
10 changed files with 332 additions and 53 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

16
include/common.h
View File

@ -2,18 +2,26 @@
#include "3rdparty/easylogging++.h"
#include <any>
#include <exception>
#include <list>
#include <memory>
#include <optional>
#include <sstream>
#include <string>
#include <memory>
#include <vector>
#include <list>
#include <unordered_set>
#include <unordered_map>
#include <unordered_set>
#include <vector>
#include <iterator>
#define uptr(type) std::unique_ptr<type>
#define sptr(type) std::shared_ptr<type>
template <typename DST, typename SRC>
inline sptr(DST) shared_cast(SRC src) {
return std::dynamic_pointer_cast<DST>(src);
}
#define FIND(container, val) std::find(container.begin(), container.end(), val)
#define panic(message) \
do { \
throw GrammarException(__FILE__, __LINE__, (message)); \

4
include/llir_value.h
View File

@ -159,8 +159,8 @@ public:
std::list<InstructionPtr_t> inst_list;
std::shared_ptr<Function> parent;
BasicBlockListNode_t itr;
std::list<BasicBlockPtr_t> successors;
std::list<BasicBlockPtr_t> predecessors;
std::list<BasicBlockPtr_t> succ_list;
std::list<BasicBlockPtr_t> pred_list;
BasicBlockPtr_t IDOM; // dominating node
std::unordered_set<BasicBlockPtr_t> idom_set; // immediate dominated nodes

2
include/machcode.h Normal file
View File

@ -0,0 +1,2 @@
#pragma once
#include "mc_inst.h"

196
include/mc_inst.h
View File

@ -1,69 +1,215 @@
#pragma once
#include "common.h"
#include "llir.h"
namespace codegen{
using std::make_shared;
namespace codegen {
// a?, t?, ra are caller-saved
// s? are callee-saved
// x0,gp,tp are preserved in user-space
enum class RV64Reg {
x0 = 0, // zero
x1, // ra, caller
x2, // sp, callee
x3, // gp
x4, // tp
x5, // t0, caller
x6,x7, // t1-2, caller
x8, // s0/fp, callee
x9, // s1, callee
x10,x11, // a0-1,caller
x12,x13,x14,x15,x16,x17, // a2-7,caller
x18,x19,x20,x21,x22,x23,x24,x25,x26,x27, // s2-11,callee
x28,x29,x30,x31, // t3-6,caller
ra, // ra, caller
sp, // sp, callee
gp, // gp
tp, // tp
t0, // t0, caller
t1,
t2, // t1-2, caller
s0, // s0/fp, callee
s1, // s1, callee
a0,
a1, // a0-1,caller
a2,
a3,
a4,
a5,
a6,
a7, // a2-7,caller
s2,
s3,
s4,
s5,
s6,
s7,
s8,
s9,
s10,
s11, // s2-11,callee
t3,
t4,
t5,
t6, // t3-6,caller
};
inline std::string Reg2Name(RV64Reg reg) {
// riscv calling convention see: https://github.com/riscv-non-isa/riscv-elf-psabi-doc/blob/master/riscv-cc.adoc
const int XLEN = 8;
inline std::string RV64_RegName(RV64Reg reg) {
std::string regname = "x" + std::to_string((int)reg);
return regname;
}
inline RV64Reg RV64_RegOffset(RV64Reg reg, int offset) {
auto xi = (int)reg + offset;
return (RV64Reg)xi;
}
class MOperand;
class MInst;
class MBasicBlock;
class MFunction;
class MOperand {
public:
enum class type {
enum class OpTag {
Virt,
Imm,
Reg,
} op_type;
int value;
} op_type = OpTag::Virt;
int value = ~0;
private:
MOperand(OpTag tag, int val) : op_type(tag), value(val) {}
public:
static sptr(MOperand) NewVirtReg(int reg_no) {
auto mop = std::make_shared<MOperand>(OpTag::Virt, reg_no);
return mop;
}
static sptr(MOperand) NewImm(int imm_val) {
auto mop = std::make_shared<MOperand>(OpTag::Imm, imm_val);
return mop;
}
static sptr(MOperand) NewReg(RV64Reg phy_reg) {
auto mop = std::make_shared<MOperand>(OpTag::Reg, (int)phy_reg);
return mop;
}
};
class MInst {
public:
enum class MInstTag {
Add, Sub, Rsb, Mul, Div, Mod, Lt, Le, Ge, Gt, Eq, Ne, And, Or,
LongMul, FMA, Mv, Branch, Jump, Return, Load, Store,Compare, Call, Global,
Add,
Sub,
Rsb,
Mul,
Div,
Mod,
Lt,
Le,
Ge,
Gt,
Eq,
Ne,
And,
Or,
Move, // actually a pseudo, mv = addi rt, rs, 0
Branch,
Jump,
Return,
Load,
Store,
Compare,
Call,
Global,
} tag;
MInst(MInstTag tag) : tag(tag) {}
sptr(MBasicBlock) parent_bb;
MInst(MInstTag tag, sptr(MBasicBlock) parent_bb) : tag(tag), parent_bb(parent_bb) {}
};
class MBasicBlock {
public:
sptr(antlrSysY::BasicBlock) ir_bb;
sptr(MFunction) parent_func;
std::list<sptr(MInst)> inst_list;
std::list<sptr(MBasicBlock)> pred_list;
std::list<sptr(MBasicBlock)> succ_list;
std::unordered_set<sptr(MOperand)> livein;
std::unordered_set<sptr(MOperand)> liveout;
};
class MFunction {
std::list<uptr(MBasicBlock)> bbs;
public:
std::list<sptr(MBasicBlock)> bb_list;
sptr(antlrSysY::Function) ir_func;
std::list<sptr(MInst)> stack_arg_reloc;
unsigned virt_reg_cnt = 0;
};
class MGlobalVar {
public:
};
class MCModule {
std::list<uptr(MFunction)> functions;
std::list<uptr(MGlobalVar)> globals;
public:
std::list<sptr(MFunction)> function_list;
std::list<sptr(MGlobalVar)> global_list;
void IR2MC(const antlrSysY::Module &ir_module);
};
}
class MInstBinary : public MInst {
public:
};
class MInstJump : public MInst {
public:
sptr(MBasicBlock) target;
MInstJump(sptr(MBasicBlock) parent_bb) : MInst(MInstTag::Jump, parent_bb) {}
static sptr(MInstJump) New(sptr(MBasicBlock) parent_bb) {
auto inst = make_shared<MInstJump>(parent_bb);
parent_bb->inst_list.push_back(inst);
return inst;
}
};
class MInstLoad : public MInst {
public:
sptr(MOperand) dst;
sptr(MOperand) addr;
sptr(MOperand) offset;
MInstLoad(sptr(MBasicBlock) parent_bb) : MInst(MInstTag::Load, parent_bb) {}
static sptr(MInstLoad) New(sptr(MBasicBlock) parent_bb) {
auto inst = make_shared<MInstLoad>(parent_bb);
parent_bb->inst_list.push_back(inst);
return inst;
}
};
class MInstMove : public MInst {
public:
sptr(MOperand) dst;
sptr(MOperand) src;
MInstMove(sptr(MBasicBlock) parent_bb) : MInst(MInstTag::Move, parent_bb) {}
static sptr(MInstMove) New(sptr(MBasicBlock) parent_bb, bool insert_begin = false) {
auto inst = make_shared<MInstMove>(parent_bb);
if (insert_begin)
parent_bb->inst_list.push_front(inst);
else
parent_bb->inst_list.push_back(inst);
return inst;
}
static sptr(MInstMove) New(sptr(MInst) rel_inst, bool insert_after = false) {
auto parent_bb = rel_inst->parent_bb;
auto inst = make_shared<MInstMove>(parent_bb);
auto rel_inst_itr = FIND(parent_bb->inst_list, rel_inst);
assert(rel_inst_itr != parent_bb->inst_list.end());
if (insert_after)
std::advance(rel_inst_itr, 1);
parent_bb->inst_list.insert(rel_inst_itr, inst);
return inst;
}
};
} // namespace codegen

6
src/algo_dominance.cpp
View File

@ -51,7 +51,7 @@ void gen_dominance(FunctionPtr_t func) {
auto cur_bb = bb_list[i];
std::vector<bool> temp(N, 1);
// temp = {i} union (intersect Dom(j)), j in pred(i)
for (auto pred : cur_bb->predecessors) {
for (auto pred : cur_bb->pred_list) {
_bitwise_and(temp, dom[pred->_dom_helper_index]);
}
temp[i] = true;
@ -112,8 +112,8 @@ void gen_dominance_frontier(FunctionPtr_t func) {
bb->DF_set.clear();
}
for (auto n : func->bb_list) {
if (n->predecessors.size() >= 2) {
for (auto pred : n->predecessors) {
if (n->pred_list.size() >= 2) {
for (auto pred : n->pred_list) {
auto runner = pred;
while (runner != n->IDOM) {
runner->DF_set.insert(n);

17
src/main.cpp
View File

@ -8,6 +8,7 @@
#include "llir.h"
#include "passes.h"
#include "visitor.h"
#include "machcode.h"
#include <3rdparty/argparse.hpp>
#include <fstream>
#include <iostream>
@ -18,6 +19,7 @@ INITIALIZE_EASYLOGGINGPP
using namespace antlrSysY;
using namespace antlr4;
using namespace codegen;
class AbortErrorListener : public BaseErrorListener {
public:
@ -42,6 +44,7 @@ int main(int argc, const char **argv) {
arg_parser.add_argument("-S").implicit_value(true).help("Useless but required by the contest").required();
arg_parser.add_argument("-o").help("Output file name").required();
arg_parser.add_argument("-O1").implicit_value(true).default_value(false).help("Performance mode");
arg_parser.add_argument("-O0").implicit_value(true).default_value(false).help("Rudimentary mode");
arg_parser.add_argument("-emit-llvm").implicit_value(true).default_value(false).help("Generate llvm ir");
arg_parser.add_argument("--v").default_value(0);
try {
@ -54,6 +57,7 @@ int main(int argc, const char **argv) {
auto source_file = arg_parser.get<std::string>("source");
auto output_file = arg_parser.get<std::string>("-o");
auto flg_O1 = arg_parser["-O1"] == true;
auto flg_O0 = arg_parser["-O0"] == true;
auto emit_llvm = arg_parser["-emit-llvm"] == true;
// std::cout << source_file << " " << output_file << " " << flg_O1 <<
// std::endl;
@ -96,12 +100,20 @@ int main(int argc, const char **argv) {
std::make_shared<PassBuildCFG>(),
std::make_shared<PassMem2Reg>(),
};
if (flg_O1) {
if (!flg_O0) {
for (auto pass : passes) {
pass->run(visitor.module);
}
}
std::vector<sptr(Pass)> optpasses = {};
if (flg_O1) {
for (auto pass : optpasses) {
pass->run(visitor.module);
}
}
if (emit_llvm) {
auto llir_file = output_file.substr(0, output_file.rfind(".")) + ".ll";
std::ofstream ofs_llir_file(llir_file);
@ -114,5 +126,8 @@ int main(int argc, const char **argv) {
// std::cout << tree->toStringTree(&parser) << std::endl << std::endl;
MCModule mc_module;
mc_module.IR2MC(visitor.module);
return 0;
}

108
src/mc_codegen.cpp Normal file
View File

@ -0,0 +1,108 @@
#include "common.h"
#include "llir.h"
#include "mc_inst.h"
#include "visitor.h"
using std::make_shared;
namespace codegen {
static sptr(MOperand) value2moperand(
sptr(antlrSysY::Value) ir_value,
sptr(MBasicBlock) mc_bb,
std::unordered_map<sptr(antlrSysY::Value), sptr(MOperand)> &val2mop
) {
if (auto fparam = shared_cast<antlrSysY::FParam>(ir_value)) {
auto itr = val2mop.find(ir_value);
if (itr != val2mop.end()) {
return itr->second;
}
else {
auto vr = MOperand::NewVirtReg(mc_bb->parent_func->virt_reg_cnt++);
val2mop.insert({ir_value, vr});
auto ir_func = mc_bb->parent_func->ir_func;
auto fparam_itr = FIND(ir_func->fparam_list, fparam);
assert(fparam_itr != ir_func->fparam_list.end());
auto fparam_ndx = std::distance(ir_func->fparam_list.begin(), fparam_itr);
// according to RV call conv, we can have a0-a7 for params
if (fparam_ndx < 8) {
// copy param as an vr in func entry
auto inst_move = MInstMove::New(mc_bb->parent_func->bb_list.front(), true);
inst_move->src = MOperand::NewReg(RV64_RegOffset(RV64Reg::a0, fparam_ndx));
inst_move->dst = vr;
}
else {
// read from stack, sp + (i - 8) * 8, since ABI requires the stack to be aligned by XLEN=64
// this need to be further re-located since sp may have changed
// FramePtr won't get used here, for perf reason. Ref:
// https://stackoverflow.com/questions/13006371/does-omitting-the-frame-pointers-really-have-a-positive-effect-on-performance-an
// auto vr_tmp = MOperand::NewVirtReg(mc_bb->parent_func->virt_reg_cnt++);
auto inst_load = MInstLoad::New(mc_bb); // ld vr, (i-8)*8(sp)
// auto inst_move = MInstMove::New(inst_load); // lui vr_t,
inst_load->addr = MOperand::NewReg(RV64Reg::sp);
inst_load->offset = MOperand::NewImm((fparam_ndx - 8) * XLEN);
inst_load->dst = vr;
mc_bb->parent_func->stack_arg_reloc.push_back(inst_load);
}
return vr;
}
}
else {
// plain situation
auto itr = val2mop.find(ir_value);
if (itr != val2mop.end()) {
return itr->second;
}
else {
auto vr = MOperand::NewVirtReg(mc_bb->parent_func->virt_reg_cnt++);
val2mop.insert({ir_value, vr});
return vr;
}
}
}
void MCModule::IR2MC(const antlrSysY::Module &ir_module) {
for (auto glob : ir_module.global_var_list) {
// TODO copy globals from ir
}
for (auto func : ir_module.function_list) {
if (func->is_libfunc()) continue;
auto mc_func = make_shared<MFunction>();
this->function_list.push_back(mc_func);
mc_func->ir_func = func;
// copy pred/succ info
std::unordered_map<sptr(antlrSysY::BasicBlock), sptr(MBasicBlock)> bb_ir2mc;
for (auto bb : func->bb_list) {
auto mc_bb = make_shared<MBasicBlock>();
mc_func->bb_list.push_back(mc_bb);
mc_bb->ir_bb = bb;
bb_ir2mc.insert({bb, mc_bb});
}
for (auto bb : func->bb_list) {
auto mc_bb = bb_ir2mc.at(bb);
assert(bb->succ_list.size() <= 2 && "At most 2 successors");
for (auto succ : bb->succ_list) {
auto mc_succ = bb_ir2mc.at(succ);
mc_bb->succ_list.push_back(mc_succ);
}
for (auto pred : bb->pred_list) {
auto mc_pred = bb_ir2mc.at(pred);
mc_bb->pred_list.push_back(mc_pred);
}
}
// instruction translate
for (auto bb : func->bb_list) {
auto mc_bb = bb_ir2mc[bb];
for (auto inst : bb->inst_list) {
if (auto li = shared_cast<antlrSysY::InstLoad>(inst)) {
auto mc_li = MInstLoad::New(mc_bb);
}
}
}
}
}
} // namespace codegen

16
src/pass_build_cfg.cpp
View File

@ -9,12 +9,12 @@ namespace antlrSysY {
static void print_cfg_info(BasicBlockPtr_t bb) {
std::cout << "---" << bb->name << "---\n";
std::cout << " pred: [";
for (auto pred : bb->predecessors) {
for (auto pred : bb->pred_list) {
std::cout << pred->name << ", ";
}
std::cout << "]\n";
std::cout << " succ: [";
for (auto succ : bb->successors) {
for (auto succ : bb->succ_list) {
std::cout << succ->name << ", ";
}
std::cout << "]\n";
@ -25,7 +25,7 @@ static void print_cfg(BasicBlockPtr_t rt, bool rst = false) {
static std::unordered_set<BasicBlockPtr_t> vis;
if (rst) vis.clear();
print_cfg_info(rt);
for (auto succ : rt->successors) {
for (auto succ : rt->succ_list) {
if (!vis.count(succ)) {
vis.insert(succ);
print_cfg(succ);
@ -34,8 +34,8 @@ static void print_cfg(BasicBlockPtr_t rt, bool rst = false) {
}
static void connect(BasicBlockPtr_t pred, BasicBlockPtr_t succ) {
pred->successors.push_back(succ);
succ->predecessors.push_back(pred);
pred->succ_list.push_back(succ);
succ->pred_list.push_back(pred);
}
static void _build_cfg(const FunctionPtr_t func) {
@ -73,9 +73,9 @@ static void _build_cfg(const FunctionPtr_t func) {
// remove unreachable blocks
for (auto itr = std::next(func->bb_list.begin()); itr != func->bb_list.end();) {
auto bb = *itr;
if (bb->predecessors.size() == 0) {
for (auto succ : bb->successors) {
succ->predecessors.remove(bb);
if (bb->pred_list.size() == 0) {
for (auto succ : bb->succ_list) {
succ->pred_list.remove(bb);
}
VLOG(6) << "remove unreachable block:" << bb->name;
itr = func->bb_list.erase(itr);

12
src/pass_mem2reg.cpp
View File

@ -94,7 +94,7 @@ static void live_in_blocks(
if (!livein_blocks.insert(bb).second) continue; // already done
// 如果alloca没有在某个块中被定义，那么它一定是从前面的某个块里面live in过来的
// 所以把它也塞进队列进行处理
for (auto pred : bb->predecessors) {
for (auto pred : bb->pred_list) {
if (def_blocks.count(pred)) continue;
worklist.push_back(pred);
}
@ -111,12 +111,12 @@ static void print_dom_info(BasicBlockPtr_t bb) {
std::cout << bb->name << ":\n";
std::cout << " level: " << bb->dom_level << "\n";
std::cout << " pred: [";
for (auto pred : bb->predecessors) {
for (auto pred : bb->pred_list) {
std::cout << pred->name << ", ";
}
std::cout << "]\n";
std::cout << " succ: [";
for (auto succ : bb->successors) {
for (auto succ : bb->succ_list) {
std::cout << succ->name << ", ";
}
std::cout << "]\n";
@ -236,7 +236,7 @@ static void _mem_2_reg(FunctionPtr_t func) {
visited[frontier_index] = true;
// livein-block opt is workable in IDF, but not here
// if (livein_blocks.count(frontier)) {
auto inst_phi = build_InstPhi(TypeHelper::TYPE_I32, frontier->predecessors, frontier, ai->name);
auto inst_phi = build_InstPhi(TypeHelper::TYPE_I32, frontier->pred_list, frontier, ai->name);
phi_to_allocaid.insert({inst_phi, i});
// }
if (!def_blocks.count(frontier)) {
@ -263,7 +263,7 @@ static void _mem_2_reg(FunctionPtr_t func) {
auto phi = Value::as<InstPhi>(inst);
auto alloca_index = phi_to_allocaid.at(phi);
int pred_index = -1;
for (auto pred : rename_info.bb->predecessors) {
for (auto pred : rename_info.bb->pred_list) {
pred_index++;
if (pred == rename_info.pred) break;
}
@ -315,7 +315,7 @@ static void _mem_2_reg(FunctionPtr_t func) {
rename_info.value_list[alloca_index] = phi;
}
}
for (auto succ : rename_info.bb->successors) {
for (auto succ : rename_info.bb->succ_list) {
rename_list.push_back({succ, rename_info.bb, rename_info.value_list});
}
}

4
src/visitor_llir_gen.cpp
View File

@ -150,7 +150,7 @@ static void _gen_blocks(std::ostream &ostr, const std::list<BasicBlockPtr_t> &bl
if (block_itr != block_list.begin()) {
ostr << block->ir_seqno << ":";
ostr << " ;" << block->name << " pred=[";
for (auto pred : block->predecessors) {
for (auto pred : block->pred_list) {
ostr << pred->ir_seqno << ", ";
}
ostr << "]" << std::endl;
@ -460,7 +460,7 @@ static void _gen_blocks(std::ostream &ostr, const std::list<BasicBlockPtr_t> &bl
LOG(WARNING) << "Unexpected type of op: " << op->to_string();
assert(0);
}
auto pred = inst->parent_bb->predecessors.begin();
auto pred = inst->parent_bb->pred_list.begin();
std::advance(pred, i);
ostr << ", %" << (*pred)->ir_seqno << "]";
if (i < inst->operand_list.size() - 1) {