feat(backend/riscv32): 实现基础的编译器功能

- 完成 RV32IMA 指令集的代码生成
- 添加整数运算、分支、调用等基本指令支持
- 实现从 IR 到机器码的转换
- 添加简单的测试用例和测试框架
This commit is contained in:
ZZY 2025-03-08 16:50:21 +08:00
parent 95bf44eb3f
commit 172d72b0a0
32 changed files with 980 additions and 469 deletions

View File

@ -3,11 +3,16 @@ all: ccompiler
run: ccompiler
./ccompiler test.c flat.bin
ccompiler: frontend
gcc -g rv32ima_codegen.c ../../middleend/ir.c -L../../frontend -lfrontend -o ccompiler
ccompiler: frontend ir
gcc -g rv32ima_codegen.c -L../../frontend -lfrontend -L../../middleend -lir -o ccompiler
frontend:
make -C ../../frontend
ir:
make -C ../../middleend
clean:
rm -f ccompiler flat.bin
rm -f ccompiler flat.bin
make -C ../../frontend clean
make -C ../../middleend clean

View File

@ -211,7 +211,7 @@ typedef enum {
#define SLTZ(rd, rs) SLT(rd, rs, REG_X0) // 小于0则置位
// sgtz rd, rs (Set if > zero)
#define SG TZ(rd, rs) SLT(rd, REG_X0, rs) // 大于0则置位
#define SGTZ(rd, rs) SLT(rd, REG_X0, rs) // 大于0则置位
// beqz rs, offset (Branch if = zero)
#define BEQZ(rs, offset) BEQ(rs, REG_X0, offset) // 为0则转移
@ -291,13 +291,16 @@ typedef enum {
#define LI(rd, num) \
LUI(rd, num), \
ADDI(rd, rd, num)
#define MV(rd, rs) ADDI(rd, rs, 0)
#define NOT(rd, rs) XORI(rd, rs, -1)
#define CALL(offset) \
AUIPC(REG_X1, offset), \
JALR(REG_X1, REG_X1, offset)
#define SEQZ(rd, rs) SLTIU(rd, rs, 1)
#define SGT(rd, rs1, rs2) SLT(rd, rs2, rs1)
// TODO call have error when outof jalr
#define CALL(offset) \
AUIPC(REG_X1, REG_X0), \
JALR(REG_X1, REG_X1, offset)
#define CALL_ABS(addr) \
AUIPC(REG_X0, addr), \
JALR(REG_X1, REG_X0, addr)

View File

@ -9,22 +9,6 @@ typedef union rv32code {
uint8_t bytes[4];
} rv32code_t;
#define CRT_CODE_SIZE 16
// 使用示例
rv32code_t gcodes[] = {
LI(REG_SP, 0x1000),
LI(REG_RA, 0x0),
CALL_ABS(CRT_CODE_SIZE << 2),
// Exit
ECALL_EXIT2(),
};
void test_raw_gen(FILE* out) {
fwrite(gcodes, sizeof(rv32code_t), sizeof(gcodes)/sizeof(gcodes[0]), out);
}
#include "../../frontend/frontend.h"
#include "../../middleend/ir.h"
typedef struct {
@ -59,9 +43,9 @@ int write_inst(union rv32code ins, FILE* fp) {
}
#define GENCODE(code) vector_push(ctx.codes, (rv32code_t)(code)); len += 4
#define GENCODES(code) do { \
#define GENCODES(...) do { \
rv32code_t codes[] = { \
code \
__VA_ARGS__ \
}; \
for (int i = 0; i < sizeof(codes) / sizeof(codes[0]); i ++) { \
GENCODE(codes[i]); \
@ -105,13 +89,18 @@ static int func_idx(ir_func_t* tofunc) {
}
static int system_func(const char* name) {
static const char defined_func[][16] = {
"ecall_pnt_int",
static struct {
const char* name;
int ecall_num;
} defined_func[] = {
{"ecall_pnt_int", 1},
{"ecall_pnt_char", 11},
{"ecall_scan_int", 1025 + 4},
};
for (int j = 0; j < sizeof(defined_func)/sizeof(defined_func[0]); j++) {
if (strcmp(name, defined_func[j]) == 0) {
return j;
for (int i = 0; i < sizeof(defined_func)/sizeof(defined_func[0]); i++) {
if (strcmp(name, defined_func[i].name) == 0) {
return defined_func[i].ecall_num;
}
}
return -1;
@ -119,11 +108,22 @@ static int system_func(const char* name) {
static int get_node_val(ir_node_t* ptr, int reg) {
int len = 0;
if (ptr->tag == IR_NODE_CONST_INT) {
GENCODES(LI(reg, ptr->data.const_int.val));
} else {
int offset = stack_offset(ptr);
GENCODE(LW(reg, REG_SP, offset));
switch (ptr->tag) {
case IR_NODE_CONST_INT: {
GENCODES(LI(reg, ptr->data.const_int.val));
break;
}
// case IR_NODE_CALL: {
// // GENCODE(SW(REG_A0, REG_SP, ctx.stack_offset));
// // GENCODE()
// // break;
// }
default: {
int offset = stack_offset(ptr);
GENCODE(LW(reg, REG_SP, offset));
break;
}
}
return len;
}
@ -139,8 +139,6 @@ static int gen_instr(ir_bblock_t* block, ir_node_t* instr) {
// S1 = *(S0 + imm)
offset = stack_offset(instr->data.load.target);
GENCODE(LW(REG_T0, REG_SP, offset));
// offset = STACK_OFFSET(instr);
// GENCODE(SW(REG_T0, REG_SP, offset));
break;
}
case IR_NODE_STORE: {
@ -180,6 +178,27 @@ static int gen_instr(ir_bblock_t* block, ir_node_t* instr) {
case IR_OP_MOD:
GENCODE(REM(REG_T0, REG_T1, REG_T2));
break;
case IR_OP_EQ:
GENCODE(XOR(REG_T0, REG_T1, REG_T2));
GENCODE(SEQZ(REG_T0, REG_T0));
break;
case IR_OP_GE:
GENCODE(SLT(REG_T0, REG_T1, REG_T2));
GENCODE(SEQZ(REG_T0, REG_T0));
break;
case IR_OP_GT:
GENCODE(SGT(REG_T0, REG_T1, REG_T2));
break;
case IR_OP_LE:
GENCODE(SGT(REG_T0, REG_T1, REG_T2));
GENCODE(SEQZ(REG_T0, REG_T0));
break;
case IR_OP_LT:
GENCODE(SLT(REG_T0, REG_T1, REG_T2));
break;
case IR_OP_NEQ:
GENCODE(XOR(REG_T0, REG_T1, REG_T2));
break;
default:
error("ERROR gen_instr op in riscv");
break;
@ -244,13 +263,15 @@ static int gen_instr(ir_bblock_t* block, ir_node_t* instr) {
}
int system_func_idx = system_func(instr->data.call.callee->name);
if (system_func_idx == 0) {
// ecall_pnt_int
GENCODE(ADDI(REG_A7, REG_X0, 0x1));
GENCODE(ECALL());
break;
if (system_func_idx != -1) {
// ecall
GENCODES(
ADDI(REG_A7, REG_X0, system_func_idx),
ECALL()
);
goto CALL_END;
}
jmp_t* jmp = xmalloc(sizeof(jmp_t));
*jmp = (jmp_t) {
.base_offset = ctx.cur_func_offset + ctx.cur_block_offset + len,
@ -260,10 +281,11 @@ static int gen_instr(ir_bblock_t* block, ir_node_t* instr) {
.cur_idx = func_idx(ctx.cur_func),
};
vector_push(ctx.call, jmp);
GENCODES((
CALL(0)
));
GENCODES(CALL(0));
CALL_END:
offset = stack_offset(instr);
GENCODE(SW(REG_A0, REG_SP, offset));
break;
}
default:
@ -335,10 +357,9 @@ static int gen_func(ir_func_t* func) {
return len;
}
static void gen_code(ir_prog_t* prog) {
static int gen_code(ir_prog_t* prog) {
ctx.prog = prog;
for (int i = 0; i < prog->extern_funcs.size; i++) {
if (system_func(prog->extern_funcs.data[i]->name) == -1) {
error("func %s not defined and not a system func", prog->extern_funcs.data[i]->name);
@ -354,12 +375,12 @@ static void gen_code(ir_prog_t* prog) {
len += ret;
}
for (int i = 0; i < ctx.call.size; i++) {
jmp_t* jmp = vector_at(ctx.call, i);
int32_t code = 0;
// FIXME ERROR
int offset = jmp_cache[jmp->to_idx] - (jmp_cache[jmp->cur_idx] + jmp->base_offset);
assert(offset > -0xfff && offset < 0xfff);
int32_t codes[2] = {
CALL(offset)
};
@ -369,6 +390,14 @@ static void gen_code(ir_prog_t* prog) {
};
}
}
// Got Main pos;
for (int i = 0; i < prog->funcs.size; i++) {
if (strcmp(vector_at(prog->funcs, i)->name, "main") == 0) {
return jmp_cache[i];
}
}
error("main not found");
}
int main(int argc, char** argv) {
@ -390,7 +419,23 @@ int main(int argc, char** argv) {
struct ASTNode* root = frontend(infilename, in, (sread_fn)fread_s);
gen_ir_from_ast(root);
gen_code(&prog);
int main_pos = gen_code(&prog);
#define CRT_CODE_SIZE 16
rv32code_t gcodes[] = {
LI(REG_SP, 0x1000),
LI(REG_RA, 0x0),
CALL(0),
// Exit
ECALL_EXIT2(),
};
main_pos += (CRT_CODE_SIZE - 4) * 4;
assert(main_pos > -0xfff && main_pos < 0xfff);
rv32code_t call_main[2] = {
CALL(main_pos)
};
gcodes[4] = call_main[0];
gcodes[5] = call_main[1];
for (int i = 0; i < CRT_CODE_SIZE; i++) {
write_inst((union rv32code) {

View File

@ -0,0 +1,3 @@
int main() {
return 65536;
}

View File

@ -0,0 +1,8 @@
int main() {
int a;
int b;
a = 1 + 2 * 3;
b = 7;
a = a - b + 1;
return a;
}

View File

@ -0,0 +1,6 @@
int main() {
int x = 10;
x = x + 1;
return x;
}

View File

@ -0,0 +1,10 @@
int main(void) {
int a;
a = 1;
if (a) {
a = 1;
} else {
a = 2;
}
return a;
}

View File

@ -0,0 +1,10 @@
int main(void) {
int a;
a = 0;
if (a) {
a = 1;
} else {
a = 2;
}
return a;
}

View File

@ -0,0 +1,9 @@
int add(int, int);
int main(void) {
return add(1, 2);
}
int add(int a, int b) {
return a + b;
}

View File

@ -0,0 +1,5 @@
int main() {
int i = 0;
while (i < 10) i = i + 1;
return i;
}

View File

@ -0,0 +1,12 @@
// #include <stdio.h>
int main() {
int i = 0;
int pow = 1;
do {
pow = pow * 2;
i = i + 1;
} while(i < 7);
// printf("%d", pow);
return pow;
}

View File

@ -0,0 +1,7 @@
int main() {
int num = 0;
for (int i = 0; i < 10; i += 1) {
num = num + 1;
}
return num;
}

View File

@ -0,0 +1,7 @@
int add(int a, int b) {
return a + b;
}
int main(void) {
return add(1, 2);
}

View File

@ -0,0 +1,18 @@
// #include <stdio.h>
int factorial(int num);
int main() {
int num = 5;
int result = factorial(num);
// printf("%d", result);
return result;
}
int factorial(int num) {
if (num == 0) {
return 1;
} else {
return num * factorial(num - 1);
}
}

View File

@ -0,0 +1,28 @@
VM := ../../rv32-vm
CC := ../../ccompiler
STD_CC := gcc
TESTS := $(wildcard *.c)
# 定义所有测试目标
TEST_TARGETS := $(patsubst %.c, %_test, $(TESTS))
all: $(TEST_TARGETS)
%_test: %.c
@$(STD_CC) -g -o $@ $<
@$(CC) $< flat.bin
@./$@ ; ret_gcc=$$?
@$(VM) flat.bin ; ret_vm=$$?
@echo "Testing $@"
@if [ $$ret_gcc -eq $$ret_vm ]; then \
echo "$@ passed"; \
else \
echo "$@ failed: GCC returned $$ret_gcc, VM returned $$ret_vm"; \
exit 1; \
fi
clean:
rm -f $(TEST_TARGETS) flat.bin
.PHONY: all clean

View File

@ -0,0 +1,6 @@
int main() {
int a, b;
a = 1;
b = 2;
return a + b;
}

View File

@ -0,0 +1,86 @@
import subprocess
import os
from pathlib import Path
# 配置参数
TEST_DIR = Path(".")
CC_PATH = Path("../../ccompiler.exe")
VM_PATH = Path("../../rv32-vm.exe")
WORKSPACE = Path(".") # 测试工作目录
# 测试用例映射表(示例)
TEST_CASE_MAP = {
"./01_return.c": 65536,
"./02_decl_expr.c": 1,
"./03_decl_init.c": 11,
"./04_if.c": 1,
"./05_else.c": 2,
"./06_fcall.c": 3,
"./07_while.c": 10,
"./08_do_while.c": 128,
"./09_for.c": 10,
"./10_main.c": 3,
"./11_recursive.c": 120,
}
def run_command(cmd, capture_stderr=True):
"""执行命令并捕获stderr"""
result = subprocess.run(
cmd,
cwd=WORKSPACE,
stderr=subprocess.PIPE if capture_stderr else None,
text=True,
timeout=1,
)
return result.stderr.strip() if capture_stderr else None
def run_test(test_file, expected):
print(f"\nTesting {test_file}...")
# 1. 编译生成flat.bin
compile_cmd = [str(CC_PATH), str(test_file)]
compile_err = run_command(compile_cmd)
if not (WORKSPACE / "flat.bin").exists():
print(f" Compilation failed: {compile_err}")
return False
# 2. 执行虚拟机
vm_cmd = [str(VM_PATH), "flat.bin"]
# 3. 解析返回值(假设最后一行是返回值)
try:
vm_err = run_command(vm_cmd)
actual = int(vm_err.split()[-1])
except (ValueError, IndexError) as e:
print(f" Invalid VM output: {vm_err}")
return False
except subprocess.TimeoutExpired:
print(" Timeout expired")
return False
# 4. 验证结果
if actual == expected:
print(f" PASSED {test_file}")
return True
else:
print(f" FAILED: Expected {expected}, got {actual}")
return False
def main():
passed = 0
total = 0
for test_file, expected in TEST_CASE_MAP.items():
total += 1
if run_test(TEST_DIR / test_file, expected):
passed += 1
# 清理中间文件
if (WORKSPACE / "flat.bin").exists():
os.remove(WORKSPACE / "flat.bin")
print(f"\nTest Summary: {passed}/{total} passed")
if __name__ == "__main__":
main()

View File

@ -163,13 +163,14 @@ static void parse_char_literal(lexer_t* lexer, tok_t* token) {
if (*peek == '\\') {
peek++;
val = got_slash(peek);
peek++;
} else {
val = *peek;
val = *peek++;
}
if (*peek != '\'') error("Unclosed character literal");
if (*peek++ != '\'') error("Unclosed character literal");
token->val.ch = val;
lexer->cur_ptr = peek + 1;
lexer->cur_ptr = peek;
token->val.have = 1;
token->type = TOKEN_CHAR_LITERAL;
}

View File

@ -117,29 +117,23 @@ void test_literals() {
// test_lexer_string("4294967295", TOKEN_INT_LITERAL); // UINT_MAX
}
// TEST_CASE("Character literals"); {
// test_lexer_string("'a'", TOKEN_CHAR_LITERAL);
// test_lexer_string("'\\n'", TOKEN_CHAR_LITERAL);
// test_lexer_string("'\\t'", TOKEN_CHAR_LITERAL);
// test_lexer_string("'\\\\'", TOKEN_CHAR_LITERAL);
// test_lexer_string("'\\0'", TOKEN_CHAR_LITERAL);
// }
TEST_CASE("Character literals"); {
test_lexer_string("'a'", TOKEN_CHAR_LITERAL);
test_lexer_string("'\\n'", TOKEN_CHAR_LITERAL);
test_lexer_string("'\\t'", TOKEN_CHAR_LITERAL);
test_lexer_string("'\\\\'", TOKEN_CHAR_LITERAL);
test_lexer_string("'\\0'", TOKEN_CHAR_LITERAL);
}
TEST_CASE("String literals"); {
test_lexer_string("\"hello\"", TOKEN_STRING_LITERAL);
test_lexer_string("\"multi-line\\nstring\"", TOKEN_STRING_LITERAL);
test_lexer_string("\"escape\\\"quote\"", TOKEN_STRING_LITERAL);
}
// TEST_CASE("Integer literals");
// test_lexer_string("123", TOKEN_INT_LITERAL);
// test_lexer_string("0x1F", TOKEN_INT_LITERAL);
// TEST_CASE("Floating literals");
// test_lexer_string("3.14e-5", TOKEN_FLOAT_LITERAL);
// TEST_CASE("Character literals");
// test_lexer_string("'\\n'", TOKEN_CHAR_LITERAL);
}
// 边界测试

View File

@ -30,14 +30,13 @@ void init_tokbuf(tok_buf_t *tokbuf, void *stream, get_tokbuf_func gettok) {
tokbuf->cap = 0;
}
tok_t *peek_tok(tok_buf_t *tokbuf)
{
tok_t *peek_tok(tok_buf_t *tokbuf) {
int idx = tokbuf->peek;
idx = ROUND_IDX(idx + 1);
tokbuf->peek = ROUND_IDX(idx + 1);
if (tokbuf->size >= tokbuf->cap) {
error("peek too deep, outof array size");
}
if (tokbuf->peek == tokbuf->end) {
if (idx == tokbuf->end) {
if (tokbuf->size == tokbuf->cap) {
error("peek_tok buffer overflow");
}
@ -46,10 +45,9 @@ tok_t *peek_tok(tok_buf_t *tokbuf)
}
tokbuf->gettok(tokbuf->stream, &(tokbuf->buf[idx]));
tokbuf->size++;
tokbuf->end = idx;
tokbuf->end = tokbuf->peek;
}
tokbuf->peek = idx;
return &(tokbuf->buf[idx]);
}

View File

@ -173,6 +173,8 @@ typedef struct ASTNode* (*parse_func_t) (parser_t*);
void parse_prog(parser_t* parser);
ast_node_t* parse_decl(parser_t* parser);
ast_node_t* parse_decl_val(parser_t* parser);
ast_node_t* parse_block(parser_t* parser);
ast_node_t* parse_stmt(parser_t* parser);
ast_node_t* parse_expr(parser_t* parser);

View File

@ -8,6 +8,7 @@
// TODO 语义分析压入符号表
static void parse_params(parser_t* parser, tok_buf_t* cache, ast_node_t* node) {
flush_peek_tok(cache);
tok_type_t ttype;
ast_node_t *params = new_ast_node();
node->decl_func.params = params;

View File

@ -89,8 +89,12 @@ ast_node_t* parse_stmt(parser_t* parser) {
// init expr or init decl_var
// TODO need add this feature
node->for_stmt.init = parse_expr(parser);
expect_pop_tok(tokbuf, TOKEN_SEMICOLON);
if (peek_decl(tokbuf)) {
node->for_stmt.init = parse_decl_val(parser);
} else {
node->for_stmt.init = parse_expr(parser);
expect_pop_tok(tokbuf, TOKEN_SEMICOLON);
}
// cond expr or null
ttype = peek_tok_type(tokbuf);

View File

@ -0,0 +1,30 @@
# 编译器设置
CC = gcc
AR = ar
CFLAGS = -g -Wall
# 源文件列表
SRCS = \
ir.c \
ir_ast.c \
ir_lib.c \
ir_type.c
# 生成目标文件列表
OBJS = $(SRCS:.c=.o)
# 最终目标
TARGET = libir.a
all: $(TARGET)
$(TARGET): $(OBJS)
$(AR) rcs $@ $^
%.o: %.c
$(CC) $(CFLAGS) -c -o $@ $<
clean:
rm -f $(OBJS) $(TARGET)
.PHONY: all clean

View File

@ -1,380 +0,0 @@
#include "ir.h"
#include "../frontend/frontend.h"
typedef struct ASTNode ASTNode;
// 上下文结构,记录生成过程中的状态
typedef struct {
ir_func_t* cur_func; // 当前处理的函数
ir_bblock_t* cur_block; // 当前基本块
} IRGenContext;
IRGenContext ctx;
ir_prog_t prog;
ir_type_t type_i32 = {
.tag = IR_TYPE_INT32,
};
static inline void init_ir_node_t(ir_node_t* node) {
node->name = NULL;
node->type = NULL;
vector_init(node->used_by);
}
static inline ir_node_t* new_irnode() {
ir_node_t* node = xmalloc(sizeof(ir_node_t));
init_ir_node_t(node);
}
static inline ir_bblock_t* new_irbblock(const char* name) {
ir_bblock_t* block = xmalloc(sizeof(ir_bblock_t));
block->label = name;
vector_init(block->instrs);
return block;
}
ir_node_t* emit_instr(ir_bblock_t* block) {
if (block == NULL) block = ctx.cur_block;
ir_node_t *node = new_irnode();
vector_push(block->instrs, node);
return vector_at(block->instrs, block->instrs.size - 1);
}
ir_node_t* emit_br(ir_node_t* cond, ir_bblock_t* trueb, ir_bblock_t* falseb) {
ir_node_t* br = emit_instr(NULL);
*br = (ir_node_t) {
.tag = IR_NODE_BRANCH,
.data.branch = {
.cond = cond,
.true_bblock = trueb,
.false_bblock = falseb,
}
};
return br;
}
ir_node_t* gen_ir_expr(ASTNode* node) {
switch (node->type) {
case NT_TERM_VAL: {
ir_node_t* ir = new_irnode();
*ir = (ir_node_t) {
.tag = IR_NODE_CONST_INT,
.data.const_int = {
.val = node->syms.tok.val.i,
},
};
return ir;
}
case NT_TERM_IDENT: {
ir_node_t* decl = node->syms.decl_node->decl_val.data;
return decl;
}
case NT_TERM_CALL: {
ir_node_t* ir = emit_instr(NULL);
*ir = (ir_node_t) {
.tag = IR_NODE_CALL,
.data.call = {
.callee = node->call.func_decl->decl_func.def->func.data,
},
};
vector_init(ir->data.call.args);
for (int i = 0; i < node->call.params->params.params.size; i++) {
vector_push(ir->data.call.args, \
gen_ir_expr(node->call.params->params.params.data[i]));
}
return ir;
}
default:
goto NEXT;
}
return NULL;
NEXT:
ir_node_t* lhs = gen_ir_expr(node->expr.left);
ir_node_t* rhs = node->expr.right ? gen_ir_expr(node->expr.right) : NULL;
if (node->type == NT_COMMA) {
return rhs;
}
ir_node_t* instr = emit_instr(NULL);
vector_push(lhs->used_by, instr);
if (rhs) { vector_push(rhs->used_by, instr); }
ir_node_t* ret;
#define BINOP(operand) do { \
*instr = (ir_node_t){ \
.tag = IR_NODE_OP, \
.data.op = { \
.op = operand, \
.lhs = lhs, \
.rhs = rhs, \
}, \
}; \
ret = instr; \
} while (0)
switch (node->type) {
case NT_ADD :// (expr) + (expr)
BINOP(IR_OP_ADD);
break;
case NT_SUB :// (expr) - (expr)
BINOP(IR_OP_SUB);
break;
case NT_MUL :// (expr) * (expr)
BINOP(IR_OP_MUL);
break;
case NT_DIV :// (expr) / (expr)
BINOP(IR_OP_DIV);
break;
case NT_MOD :// (expr) % (expr)
BINOP(IR_OP_MOD);
break;
case NT_AND :// (expr) & (expr)
BINOP(IR_OP_AND);
break;
case NT_OR :// (expr) | (expr)
BINOP(IR_OP_OR);
break;
case NT_XOR :// (expr) ^ (expr)
BINOP(IR_OP_XOR);
break;
case NT_BIT_NOT :// ~ (expr)
// TODO
// BINOP(IR_OP_NOT);
break;
case NT_L_SH :// (expr) << (expr)
BINOP(IR_OP_SHL);
break;
case NT_R_SH :// (expr) >> (expr)
BINOP(IR_OP_SHR); // Shift right logical.
// TODO
// BINOP(IR_OP_SAR); // Shift right arithmetic.
break;
case NT_EQ :// (expr) == (expr)
BINOP(IR_OP_EQ);
break;
case NT_NEQ :// (expr) != (expr)
BINOP(IR_OP_NEQ);
break;
case NT_LE :// (expr) <= (expr)
BINOP(IR_OP_LE);
break;
case NT_GE :// (expr) >= (expr)
BINOP(IR_OP_GE);
break;
case NT_LT :// (expr) < (expr)
BINOP(IR_OP_LT);
break;
case NT_GT :// (expr) > (expr)
BINOP(IR_OP_GE);
break;
case NT_AND_AND :// (expr) && (expr)
break;
case NT_OR_OR :// (expr) || (expr)
break;
case NT_NOT :// ! (expr)
ir_node_t* zero = xmalloc(sizeof(ir_node_t));
*zero = (ir_node_t){
.tag = IR_NODE_CONST_INT,
.data.const_int = {
.val = 0,
},
};
*instr = (ir_node_t){
.tag = IR_NODE_OP,
.data.op = {
.op = IR_OP_EQ,
.lhs = zero,
.rhs = lhs,
},
};
ret = instr;
break;
case NT_ASSIGN :// (expr) = (expr)
*instr = (ir_node_t){
.tag = IR_NODE_STORE,
.data.store = {
.target = lhs,
.value = rhs,
},
};
ret = rhs;
break;
// case NT_COND : // (expr) ? (expr) : (expr)
default:
// TODO self error msg
error("Unsupported IR generation for AST node type %d", node->type);
break;
}
return ret;
}
static ir_func_t* new_irfunc(const char* name) {
ir_func_t *func = xmalloc(sizeof(ir_func_t));
vector_init(func->bblocks);
vector_init(func->params);
*func = (ir_func_t) {
.name = name,
// TODO typing system
.type = &type_i32,
};
return func;
}
static void gen_ir_func(ASTNode* node, ir_func_t* func) {
assert(node->type == NT_FUNC);
ir_bblock_t *entry = new_irbblock("entry");
vector_push(func->bblocks, entry);
vector_push(prog.funcs, func);
IRGenContext prev_ctx = ctx;
ctx.cur_func = func;
ctx.cur_block = entry;
ast_node_t* params = node->func.decl->decl_func.params;
for (int i = 0; i < params->params.params.size; i ++) {
ir_node_t* decl = emit_instr(entry);
ast_node_t* param = params->params.params.data[i];
vector_push(func->params, decl);
*decl = (ir_node_t) {
.tag = IR_NODE_ALLOC,
.name = param->decl_val.name->syms.tok.val.str,
.type = &type_i32,
};
param->decl_val.data = decl;
}
gen_ir_from_ast(node->func.body);
ctx = prev_ctx;
}
void gen_ir_from_ast(struct ASTNode* node) {
switch (node->type) {
case NT_ROOT: {
for (int i = 0; i < node->root.children.size; i ++) {
gen_ir_from_ast(node->root.children.data[i]);
}
break;
}
case NT_DECL_FUNC: {
ir_func_t* func = new_irfunc(node->decl_func.name->syms.tok.val.str);
if (node->decl_func.def == NULL) {
ast_node_t* def = new_ast_node();
def->func.body = NULL;
def->func.decl = node;
node->decl_func.def = def;
vector_push(prog.extern_funcs, func);
}
node->decl_func.def->func.data = func;
break;
}
case NT_FUNC: {
gen_ir_func(node, node->func.data);
break;
}
case NT_STMT_RETURN: {
ir_node_t* ret = NULL;
if (node->return_stmt.expr_stmt != NULL) {
ret = gen_ir_expr(node->return_stmt.expr_stmt);
}
ir_node_t* ir = emit_instr(NULL);
*ir = (ir_node_t) {
.tag = IR_NODE_RET,
.data = {
.ret = {
.ret_val = ret,
}
}
};
vector_push(ctx.cur_func->bblocks, new_irbblock(NULL));
break;
}
case NT_STMT_BLOCK: {
gen_ir_from_ast(node->block_stmt.block);
break;
}
case NT_BLOCK: {
for (int i = 0; i < node->block.children.size; i ++) {
gen_ir_from_ast(node->block.children.data[i]);
}
break;
}
case NT_STMT_IF: {
ir_node_t *cond = gen_ir_expr(node->if_stmt.cond);
ir_bblock_t* trueb = new_irbblock("true_block");
ir_bblock_t* falseb = new_irbblock("false_block");
emit_br(cond, trueb, falseb);
vector_push(ctx.cur_func->bblocks, trueb);
ctx.cur_block = trueb;
gen_ir_from_ast(node->if_stmt.if_stmt);
ir_node_t* jmp = emit_instr(NULL);
if (node->if_stmt.else_stmt != NULL) {
vector_push(ctx.cur_func->bblocks, falseb);
ctx.cur_block = falseb;
gen_ir_from_ast(node->if_stmt.else_stmt);
ir_node_t* jmp = emit_instr(NULL);
ctx.cur_block = new_irbblock("jmp_block");
vector_push(ctx.cur_func->bblocks, ctx.cur_block);
*jmp = (ir_node_t) {
.tag = IR_NODE_JUMP,
.data.jump = {
.target_bblock = ctx.cur_block,
},
};
} else {
ctx.cur_block = falseb;
}
*jmp = (ir_node_t) {
.tag = IR_NODE_JUMP,
.data.jump = {
.target_bblock = ctx.cur_block,
},
};
break;
}
case NT_STMT_WHILE: {
node->while_stmt.cond;
node->while_stmt.body;
break;
}
case NT_STMT_DOWHILE: {
node->do_while_stmt.cond;
node->do_while_stmt.body;
break;
}
case NT_STMT_FOR: {
node->for_stmt.init;
node->for_stmt.cond;
node->for_stmt.iter;
node->for_stmt.body;
break;
}
case NT_DECL_VAR: {
ir_node_t* ret_node = emit_instr(NULL);
*ret_node = (ir_node_t) {
.tag = IR_NODE_ALLOC,
.name = node->decl_val.name->syms.tok.val.str,
.type = &type_i32,
};
node->decl_val.data = ret_node;
if (node->decl_val.expr_stmt != NULL) {
gen_ir_from_ast(node->decl_val.expr_stmt);
}
break;
}
case NT_STMT_EXPR: {
gen_ir_expr(node->expr_stmt.expr_stmt);
break;
}
case NT_STMT_EMPTY: {
break;
}
default:
// TODO: 错误处理
error("unknown node type");
break;
}
}

View File

@ -57,22 +57,25 @@ typedef struct {
vector_header(extern_funcs, ir_func_t*);
} ir_prog_t;
typedef enum ir_node_tag {
IR_NODE_NULL,
IR_NODE_CONST_INT,
IR_NODE_ALLOC,
IR_NODE_LOAD,
IR_NODE_STORE,
IR_NODE_GET_PTR,
IR_NODE_OP,
IR_NODE_BRANCH,
IR_NODE_JUMP,
IR_NODE_CALL,
IR_NODE_RET,
} ir_node_tag_t;
struct ir_node {
const ir_type_t* type;
const char* name;
vector_header(used_by, ir_node_t*);
enum {
IR_NODE_CONST_INT,
IR_NODE_ALLOC,
IR_NODE_LOAD,
IR_NODE_STORE,
IR_NODE_GET_PTR,
IR_NODE_OP,
IR_NODE_BRANCH,
IR_NODE_JUMP,
IR_NODE_CALL,
IR_NODE_RET,
} tag;
ir_node_tag_t tag;
union {
struct {
int32_t val;

View File

@ -0,0 +1,439 @@
#include "ir.h"
#include "ir_lib.h"
#include "ir_type.h"
#include "../frontend/frontend.h"
// 上下文结构,记录生成过程中的状态
typedef struct {
ir_func_t* cur_func; // 当前处理的函数
ir_bblock_t* cur_block; // 当前基本块
} IRGenContext;
IRGenContext ctx;
ir_prog_t prog;
static void emit_instr(ir_bblock_t* block, ir_node_t* node) {
if (block == NULL) block = ctx.cur_block;
vector_push(block->instrs, node);
// return &(vector_at(block->instrs, block->instrs.size - 1));
}
static ir_node_t* emit_br(ir_node_t* cond, ir_bblock_t* trueb, ir_bblock_t* falseb) {
ir_node_t* br = new_ir_node(NULL, IR_NODE_BRANCH);
emit_instr(NULL, br);
br->data.branch.cond = cond;
br->data.branch.true_bblock = trueb;
br->data.branch.false_bblock = falseb;
return br;
}
static ir_node_t* gen_ir_expr(ast_node_t* node);
static ir_node_t* gen_ir_term(ast_node_t* node) {
switch (node->type) {
case NT_TERM_VAL: {
ir_node_t* ir = new_ir_node(NULL, IR_NODE_CONST_INT);
ir->data.const_int.val = node->syms.tok.val.i;
return ir;
}
case NT_TERM_IDENT: {
ir_node_t* decl = node->syms.decl_node->decl_val.data;
return decl;
}
case NT_TERM_CALL: {
ir_node_t* call = new_ir_node(NULL, IR_NODE_CALL);
call->data.call.callee = node->call.func_decl->decl_func.def->func.data;
for (int i = 0; i < node->call.params->params.params.size; i++) {
ast_node_t* param = vector_at(node->call.params->params.params, i);
ir_node_t *tmp = gen_ir_expr(param);
vector_push(call->data.call.args, tmp);
}
emit_instr(NULL, call);
return call;
}
default: {
assert(0);
}
}
}
static ir_node_t* gen_ir_expr(ast_node_t* node) {
// term node
switch (node->type) {
case NT_TERM_VAL:
case NT_TERM_IDENT:
case NT_TERM_CALL:
return gen_ir_term(node);
default:
break;
}
ir_node_t* lhs = gen_ir_expr(node->expr.left);
ir_node_t* rhs = node->expr.right ? gen_ir_expr(node->expr.right) : NULL;
if (node->type == NT_COMMA) {
return rhs;
}
ir_node_t* instr = NULL;
vector_push(lhs->used_by, instr);
if (rhs) { vector_push(rhs->used_by, instr); }
ir_node_t* ret;
#define BINOP(operand) do { \
instr = new_ir_node(NULL, IR_NODE_OP); \
instr->data.op.op = operand; \
instr->data.op.lhs = lhs; \
instr->data.op.rhs = rhs; \
ret = instr; \
} while (0)
switch (node->type) {
case NT_ADD: {
// (expr) + (expr)
BINOP(IR_OP_ADD); break;
}
case NT_SUB: {
// (expr) - (expr)
BINOP(IR_OP_SUB); break;
}
case NT_MUL: {
// (expr) * (expr)
BINOP(IR_OP_MUL); break;
}
case NT_DIV: {
// (expr) / (expr)
BINOP(IR_OP_DIV); break;
}
case NT_MOD: {
// (expr) % (expr)
BINOP(IR_OP_MOD); break;
}
case NT_AND: {
// (expr) & (expr)
BINOP(IR_OP_AND); break;
}
case NT_OR: {
// (expr) | (expr)
BINOP(IR_OP_OR); break;
}
case NT_XOR: {
// (expr) ^ (expr)
BINOP(IR_OP_XOR); break;
}
case NT_BIT_NOT: {
// ~ (expr)
// TODO
// BINOP(IR_OP_NOT);
break;
}
case NT_L_SH: {
// (expr) << (expr)
BINOP(IR_OP_SHL);
break;
}
case NT_R_SH: {
// (expr) >> (expr)
BINOP(IR_OP_SHR); // Shift right logical.
// TODO
// BINOP(IR_OP_SAR); // Shift right arithmetic.
break;
}
case NT_EQ: {
// (expr) == (expr)
BINOP(IR_OP_EQ); break;
}
case NT_NEQ: {
// (expr) != (expr)
BINOP(IR_OP_NEQ); break;
}
case NT_LE: {
// (expr) <= (expr)
BINOP(IR_OP_LE); break;
}
case NT_GE: {
// (expr) >= (expr)
BINOP(IR_OP_GE); break;
}
case NT_LT: {
// (expr) < (expr)
BINOP(IR_OP_LT); break;
}
case NT_GT: {
// (expr) > (expr)
BINOP(IR_OP_GE); break;
}
case NT_AND_AND:// (expr) && (expr)
error("unimpliment");
break;
case NT_OR_OR:// (expr) || (expr)
error("unimpliment");
break;
case NT_NOT: {
// ! (expr)
instr = new_ir_node(NULL, IR_NODE_OP);
instr->data.op.op = IR_OP_EQ,
instr->data.op.lhs = &node_zero,
instr->data.op.rhs = lhs,
ret = instr;
break;
}
case NT_ASSIGN: {
// (expr) = (expr)
instr = new_ir_node(NULL, IR_NODE_STORE);
instr->data.store.target = lhs;
instr->data.store.value = rhs;
ret = rhs;
break;
}
// case NT_COND: // (expr) ? (expr) : (expr)
default: {
// TODO self error msg
error("Unsupported IR generation for AST node type %d", node->type);
break;
}
}
emit_instr(NULL, instr);
return ret;
}
static void gen_ir_func(ast_node_t* node, ir_func_t* func) {
assert(node->type == NT_FUNC);
ir_bblock_t *entry = new_ir_bblock("entry");
vector_push(func->bblocks, entry);
vector_push(prog.funcs, func);
IRGenContext prev_ctx = ctx;
ctx.cur_func = func;
ctx.cur_block = entry;
ast_node_t* params = node->func.decl->decl_func.params;
for (int i = 0; i < params->params.params.size; i ++) {
ast_node_t* param = params->params.params.data[i];
ir_node_t* decl = new_ir_node(param->decl_val.name->syms.tok.val.str, IR_NODE_ALLOC);
emit_instr(entry, decl);
vector_push(func->params, decl);
// TODO Typing system
decl->type = &type_i32;
param->decl_val.data = decl;
}
gen_ir_from_ast(node->func.body);
ctx = prev_ctx;
}
void gen_ir_jmp(ast_node_t* node) {
ir_bblock_t *bblocks[3];
for (int i = 0; i < sizeof(bblocks)/sizeof(bblocks[0]); i++) {
bblocks[i] = new_ir_bblock(NULL);
vector_push(ctx.cur_func->bblocks, bblocks[i]);
}
#define NEW_IR_JMP(name, block) do { \
name = new_ir_node(NULL, IR_NODE_JUMP); \
name->data.jump.target_bblock = block; \
} while (0)
switch (node->type) {
case NT_STMT_IF: {
ir_bblock_t* trueb = bblocks[0];
ir_bblock_t* falseb = bblocks[1];
ir_bblock_t* endb = bblocks[2];
ir_node_t* jmp;
// cond
ir_node_t *cond = gen_ir_expr(node->if_stmt.cond);
emit_br(cond, trueb, falseb);
// true block
vector_push(ctx.cur_func->bblocks, trueb);
ctx.cur_block = trueb;
gen_ir_from_ast(node->if_stmt.if_stmt);
// else block
if (node->if_stmt.else_stmt != NULL) {
vector_push(ctx.cur_func->bblocks, falseb);
ctx.cur_block = falseb;
gen_ir_from_ast(node->if_stmt.else_stmt);
ir_node_t* jmp;
ctx.cur_block = endb;
vector_push(ctx.cur_func->bblocks, ctx.cur_block);
NEW_IR_JMP(jmp, ctx.cur_block);
emit_instr(falseb, jmp);
} else {
ctx.cur_block = falseb;
}
NEW_IR_JMP(jmp, ctx.cur_block);
emit_instr(trueb, jmp);
break;
}
case NT_STMT_WHILE: {
ir_bblock_t* entryb = bblocks[0];
ir_bblock_t* bodyb = bblocks[1];
ir_bblock_t* endb = bblocks[2];
ir_node_t* entry;
NEW_IR_JMP(entry, entryb);
emit_instr(NULL, entry);
// Entry:
ctx.cur_block = entryb;
ir_node_t *cond = gen_ir_expr(node->while_stmt.cond);
emit_br(cond, bodyb, endb);
// Body:
ir_node_t* jmp;
ctx.cur_block = bodyb;
gen_ir_from_ast(node->while_stmt.body);
NEW_IR_JMP(jmp, entryb);
emit_instr(NULL, jmp);
// End:
ctx.cur_block = endb;
break;
}
case NT_STMT_DOWHILE: {
ir_bblock_t* entryb = bblocks[0];
ir_bblock_t* bodyb = bblocks[1];
ir_bblock_t* endb = bblocks[2];
ir_node_t* entry;
NEW_IR_JMP(entry, bodyb);
emit_instr(NULL, entry);
// Body:
ctx.cur_block = bodyb;
gen_ir_from_ast(node->do_while_stmt.body);
ir_node_t* jmp;
NEW_IR_JMP(jmp, entryb);
emit_instr(NULL, jmp);
// Entry:
ctx.cur_block = entryb;
ir_node_t *cond = gen_ir_expr(node->do_while_stmt.cond);
emit_br(cond, bodyb, endb);
// End:
ctx.cur_block = endb;
break;
}
case NT_STMT_FOR: {
ir_bblock_t* entryb = bblocks[0];
ir_bblock_t* bodyb = bblocks[1];
ir_bblock_t* endb = bblocks[2];
if (node->for_stmt.init) {
gen_ir_from_ast(node->for_stmt.init);
}
ir_node_t* entry;
NEW_IR_JMP(entry, entryb);
emit_instr(NULL, entry);
// Entry:
ctx.cur_block = entryb;
if (node->for_stmt.cond) {
ir_node_t *cond = gen_ir_expr(node->for_stmt.cond);
emit_br(cond, bodyb, endb);
} else {
ir_node_t* jmp;
NEW_IR_JMP(jmp, bodyb);
}
// Body:
ctx.cur_block = bodyb;
gen_ir_from_ast(node->for_stmt.body);
if (node->for_stmt.iter) {
gen_ir_expr(node->for_stmt.iter);
}
ir_node_t* jmp;
NEW_IR_JMP(jmp, entryb);
emit_instr(NULL, jmp);
// End:
ctx.cur_block = endb;
break;
}
default:
error("ir jmp can't hit here");
}
}
void gen_ir_from_ast(ast_node_t* node) {
switch (node->type) {
case NT_ROOT: {
for (int i = 0; i < node->root.children.size; i ++) {
gen_ir_from_ast(node->root.children.data[i]);
}
break;
}
case NT_DECL_FUNC: {
ir_func_t* func = new_ir_func(node->decl_func.name->syms.tok.val.str, &type_i32);
if (node->decl_func.def == NULL) {
ast_node_t* def = new_ast_node();
def->func.body = NULL;
def->func.decl = node;
node->decl_func.def = def;
vector_push(prog.extern_funcs, func);
}
node->decl_func.def->func.data = func;
break;
}
case NT_FUNC: {
gen_ir_func(node, node->func.data);
break;
}
case NT_STMT_RETURN: {
ir_node_t* ret = NULL;
if (node->return_stmt.expr_stmt != NULL) {
ret = gen_ir_expr(node->return_stmt.expr_stmt);
}
ir_node_t* ir = new_ir_node(NULL, IR_NODE_RET);
ir->data.ret.ret_val = ret;
emit_instr(NULL, ir);
ir_bblock_t* block = new_ir_bblock(NULL);
ctx.cur_block = block;
vector_push(ctx.cur_func->bblocks, block);
break;
}
case NT_STMT_BLOCK: {
gen_ir_from_ast(node->block_stmt.block);
break;
}
case NT_BLOCK: {
for (int i = 0; i < node->block.children.size; i ++) {
gen_ir_from_ast(node->block.children.data[i]);
}
break;
}
case NT_STMT_IF:
case NT_STMT_WHILE:
case NT_STMT_DOWHILE:
case NT_STMT_FOR:
gen_ir_jmp(node);
break;
case NT_DECL_VAR: {
ir_node_t* ir = new_ir_node(node->decl_val.name->syms.tok.val.str, IR_NODE_ALLOC);
emit_instr(NULL, ir);
// TODO Typing system
ir->type = &type_i32;
node->decl_val.data = ir;
if (node->decl_val.expr_stmt != NULL) {
gen_ir_from_ast(node->decl_val.expr_stmt);
}
break;
}
case NT_STMT_EXPR: {
gen_ir_expr(node->expr_stmt.expr_stmt);
break;
}
case NT_STMT_EMPTY: {
break;
}
default:
// TODO: 错误处理
error("unknown node type");
break;
}
}

View File

View File

@ -0,0 +1,122 @@
#include "ir.h"
// FIXME using stdlib.h
#include <stdlib.h>
static int total_alloc = 0;
typedef union ir_alloc_item {
ir_node_t node;
ir_bblock_t bblock;
ir_func_t func;
ir_prog_t prog;
} ir_alloc_item_t;
ir_alloc_item_t* alloc_item() {
return malloc(sizeof(ir_alloc_item_t));
}
void free_item(ir_alloc_item_t* item) {
return free(item);
}
ir_node_t* new_ir_node(const char* name, ir_node_tag_t tag) {
ir_node_t* node = (ir_node_t*)alloc_item();
node->name = name;
node->type = NULL;
node->tag = tag;
switch (tag) {
case IR_NODE_ALLOC: {
node->type = NULL;
break;
}
case IR_NODE_BRANCH: {
node->data.branch.cond = NULL;
node->data.branch.true_bblock = NULL;
node->data.branch.false_bblock = NULL;
break;
}
case IR_NODE_CALL: {
vector_init(node->data.call.args);
node->data.call.callee = NULL;
break;
}
case IR_NODE_CONST_INT: {
node->data.const_int.val = 0;
break;
}
case IR_NODE_JUMP: {
node->data.jump.target_bblock = NULL;
break;
}
case IR_NODE_LOAD: {
node->data.load.target = NULL;
break;
}
case IR_NODE_STORE: {
node->data.store.target = NULL;
node->data.store.value = NULL;
break;
}
case IR_NODE_OP: {
node->data.op.op = 0;
node->data.op.lhs = NULL;
node->data.op.rhs = NULL;
break;
}
case IR_NODE_RET: {
node->data.ret.ret_val = NULL;
break;
}
case IR_NODE_GET_PTR: {
}
default: {
exit(0);
}
}
vector_init(node->used_by);
return node;
}
void dump_ir_node(ir_node_t* node) {
}
void free_irnode() {
}
ir_bblock_t* new_ir_bblock(const char* name) {
ir_bblock_t* block = (ir_bblock_t*)alloc_item();
block->label = name;
vector_init(block->instrs);
return block;
}
void free_irbblock() {
}
ir_func_t* new_ir_func(const char* name, ir_type_t* type) {
ir_func_t* func = (ir_func_t*)alloc_item();
func->name = name;
func->type = type;
vector_init(func->params);
vector_init(func->bblocks);
return func;
}
void free_irfunc() {
}
ir_prog_t* new_ir_prog() {
ir_prog_t* prog = (ir_prog_t*)alloc_item();
vector_init(prog->global);
vector_init(prog->funcs);
vector_init(prog->extern_funcs);
return prog;
}
void free_irprog() {
}

View File

@ -0,0 +1,9 @@
#ifndef __IR_LIB_H__
#define __IR_LIB_H__
#include "ir.h"
ir_node_t* new_ir_node(const char* name, ir_node_tag_t tag);
ir_bblock_t* new_ir_bblock(const char* name);
ir_func_t* new_ir_func(const char* name, ir_type_t* type);
#endif

View File

@ -0,0 +1,12 @@
#include "ir.h"
ir_type_t type_i32 = {
.tag = IR_TYPE_INT32,
};
ir_node_t node_zero = {
.tag = IR_NODE_CONST_INT,
.data.const_int = {
.val = 0,
},
};

View File

@ -0,0 +1,8 @@
#ifndef __IR_TYPE_H__
#define __IR_TYPE_H__
#include "ir.h"
extern ir_type_t type_i32;
extern ir_node_t node_zero;
#endif