阶段 B 步骤 4:compiler 配置驱动化,去除 isa 依赖。

- Project.h 自带 IoBinding;Stb.h/cpp:compiler 侧 .stb 规范(常量/ConstEntry/FNV/StbView/读写/sidecar)
- Codec.h/cpp:指令字 pack/字段/配置驱动 disasm(format 表驱动)
- Codegen:opcode 全查 MachineConfig(E_* 发射器)、常量 tag 查配置类型、初值按基元宽度写、
  映像拼装用 Stb 常量;codegen_project 增加 cfg 参数
- main.cpp:--machine <path>(编译/反汇编必填),--disasm 用 StbView+Codec
- CMake:compiler 不再链接 isa(仅 toml++ 私有);测试补链 isa(自身断言用)
- 验证:20 用例字节不变(hash 0xc3fe4ae74ad45900 相同)、ctest 12/12、缺 --machine 报错退出
This commit is contained in:
2026-08-21 22:14:29 +08:00
parent 44b5ecd46d
commit 447a8d0119
15 changed files with 739 additions and 390 deletions
+4 -2
View File
@@ -7,6 +7,7 @@ project(Compiler
DESCRIPTION "ST / toml 编译器") DESCRIPTION "ST / toml 编译器")
# 词法、语法、符号表、类型检查、codegen、链接都留在 compiler 一个库里 # 词法、语法、符号表、类型检查、codegen、链接都留在 compiler 一个库里
# 指令 opcode / 类型 / FB 布局来自 machine.tomlMachineConfig),不依赖 isa
add_library(compiler STATIC add_library(compiler STATIC
./src/Lexer.cpp ./src/Lexer.cpp
./src/Project.cpp ./src/Project.cpp
@@ -14,11 +15,12 @@ add_library(compiler STATIC
./src/Linker.cpp ./src/Linker.cpp
./src/Typecheck.cpp ./src/Typecheck.cpp
./src/Codegen.cpp ./src/Codegen.cpp
./src/MachineConfig.cpp) ./src/MachineConfig.cpp
./src/Stb.cpp
./src/Codec.cpp)
target_include_directories(compiler PUBLIC ${CMAKE_CURRENT_SOURCE_DIR}/include) target_include_directories(compiler PUBLIC ${CMAKE_CURRENT_SOURCE_DIR}/include)
target_include_directories(compiler PRIVATE ${CMAKE_SOURCE_DIR}/third_party/tomlplusplus/include) target_include_directories(compiler PRIVATE ${CMAKE_SOURCE_DIR}/third_party/tomlplusplus/include)
target_link_libraries(compiler PUBLIC isa)
# 可执行入口:STCompiler <project.toml> -o <name>.stb # 可执行入口:STCompiler <project.toml> -o <name>.stb
add_executable(STCompiler add_executable(STCompiler
+36
View File
@@ -0,0 +1,36 @@
/**
* @file Codec.h
* @brief 指令字编解码 + 配置驱动反汇编(编译器侧)
* @author
* @date 2026-08-21
*
* @details 编译器不依赖 isa:指令字布局([op:8|rd:8|a:8|b:8] 小端)与
* 反汇编输出(按 machine.toml 的 format/参数名)全部由本模块 + MachineConfig 提供,
* 保证与执行器(isa)产生的字节一致。
*/
#pragma once
#include <cstddef>
#include <cstdint>
#include "compiler/MachineConfig.h"
namespace compiler {
// 指令字:[op:8 | rd:8 | a:8 | b:8],小端 u32
typedef uint32_t Instr;
Instr pack(uint8_t opcode, uint8_t rd, uint8_t a, uint8_t b);
uint8_t op_of(Instr w);
uint8_t rd_of(Instr w);
uint8_t a_of(Instr w);
uint8_t b_of(Instr w);
uint16_t imm16_of(Instr w); // a|b 拼 u16const_id / slot / fn_id
int16_t off16_of(Instr w); // a|b 有符号偏移(相对下一条)
// 配置驱动反汇编:按 machine.toml 的 format 输出一行文本。
// 未知操作码 → "??? 0x<hex>"。输出格式与 Doc/isa/指令与映像.md 一致。
void disasm(const MachineConfig& cfg, Instr w, char* out, size_t cap);
}
+4 -2
View File
@@ -12,14 +12,16 @@
#include <vector> #include <vector>
#include "compiler/Linker.h" #include "compiler/Linker.h"
#include "compiler/MachineConfig.h"
#include "compiler/Parser.h" #include "compiler/Parser.h"
#include "compiler/Project.h" #include "compiler/Project.h"
namespace compiler { namespace compiler {
// 编译工程为 .stb 映像字节(在链接 + 类型检查成功后调用)。 // 编译工程为 .stb 映像字节(在链接 + 类型检查成功后调用)。
// 指令 opcode / 类型 tag / FB 布局全部来自 machine.tomlcfg)。
// 失败返回 falseerr 前缀 "codegen error"。 // 失败返回 falseerr 前缀 "codegen error"。
bool codegen_project(const Project& proj, const std::vector<SourceUnit>& units, bool codegen_project(const Project& proj, const std::vector<SourceUnit>& units,
const LinkResult& link, std::vector<uint8_t>* image, const LinkResult& link, const MachineConfig& cfg,
std::string* err); std::vector<uint8_t>* image, std::string* err);
} }
+11 -4
View File
@@ -11,10 +11,17 @@
#include <string> #include <string>
#include <vector> #include <vector>
#include "isa/Image.h"
namespace compiler { namespace compiler {
// I/O 绑定(project.toml [[io.*]];不创造变量,slot 由链接结果解析)
struct IoBinding {
std::string var;
uint32_t slot = 0;
uint32_t channel = 0;
uint32_t bit = 0;
bool is_input = false; // true = [[io.input]]false = [[io.output]]
};
// 与 Doc/初步计划.md 12.1 的 toml 字段表一一对应。 // 与 Doc/初步计划.md 12.1 的 toml 字段表一一对应。
// 字段白名单 / 必填 / io 完整性在 parse_project 内校验。 // 字段白名单 / 必填 / io 完整性在 parse_project 内校验。
struct Project { struct Project {
@@ -24,7 +31,7 @@ namespace compiler {
uint32_t dt_ms = 0; // 必填 > 0 uint32_t dt_ms = 0; // 必填 > 0
std::vector<std::string> files_st; // [files] 必填非空 std::vector<std::string> files_st; // [files] 必填非空
std::string gvl_file; // [gvl] 可选;无则空串 std::string gvl_file; // [gvl] 可选;无则空串
std::vector<isa::IoBinding> io; // [[io.*]] 可选;slot 12.6 才解析,此处填 0 std::vector<IoBinding> io; // [[io.*]] 可选;slot 12.6 才解析,此处填 0
std::string base_dir; // toml 所在目录(解析相对路径用) std::string base_dir; // toml 所在目录(解析相对路径用)
}; };
@@ -37,6 +44,6 @@ namespace compiler {
std::vector<std::string> compile_files(const Project& p); std::vector<std::string> compile_files(const Project& p);
// 校验全部文件存在并计算工程哈希:集合按路径排序,对内容做 FNV-1a 64 增量; // 校验全部文件存在并计算工程哈希:集合按路径排序,对内容做 FNV-1a 64 增量;
// 空集合 = basisisa::kFnvBasis)。缺文件报错前缀 "file missing"。 // 空集合 = basisStb::kFnvBasis)。缺文件报错前缀 "file missing"。
bool compute_project_hash(const Project& p, uint64_t* hash, std::string* err); bool compute_project_hash(const Project& p, uint64_t* hash, std::string* err);
} }
+103
View File
@@ -0,0 +1,103 @@
/**
* @file Stb.h
* @brief 编译器自带的 .stb 映像规范(写侧)+ 只读视图 + FNV-1a
* @author
* @date 2026-08-21
*
* @details 执行器(vm)各有自己的读写实现;格式契约见 Doc/isa/指令与映像.md。
* 12.13 修订(型号标识[32] 头 104 字节 + SHA-256 文件尾)在后续步骤落地。
*/
#pragma once
#include <cstddef>
#include <cstdint>
#include <string>
#include <vector>
#include "compiler/Project.h"
namespace compiler {
// ---- .stb 格式常量(契约)----
static const uint32_t kMagic = 0x43545353u; // "STSC" 小端
static const uint32_t kVersion = 1;
static const size_t kHeaderSize = 72;
static const size_t kConstEntrySize = 12;
static const size_t kFuncRowSize = 12;
static const uint64_t kFnvBasis = 0xcbf29ce484222325ull;
static const uint64_t kFnvPrime = 0x100000001b3ull;
// 常量表一行(tag0=BOOL、1=INT、2=TIME,契约)
struct ConstEntry {
uint32_t tag = 0;
uint64_t value = 0;
};
// FNV-1a 64(工程哈希,非密码学)
uint64_t fnv1a64_update(uint64_t h, const uint8_t* data, size_t len);
uint64_t fnv1a64(const uint8_t* data, size_t len);
// 只读视图(--disasm 用;校验魔数/版本/段边界)
class StbView {
public:
struct FuncRow {
uint32_t nregs = 0;
uint32_t code_offset = 0;
uint32_t code_len = 0;
};
static StbView from(const uint8_t* buf, size_t len);
static StbView from(const std::vector<uint8_t>& buf);
bool ok() const { return ok_; }
const std::string& error() const { return err_; }
uint32_t cycle_limit() const { return cycle_limit_; }
uint32_t dt_ms() const { return dt_ms_; }
uint64_t project_hash() const { return project_hash_; }
uint32_t entry_fn_id() const { return entry_fn_id_; }
uint32_t n_globals() const { return n_globals_; }
uint32_t n_consts() const { return n_consts_; }
uint32_t n_funcs() const { return n_funcs_; }
uint32_t offset_code() const { return offset_code_; }
uint32_t offset_fb() const { return offset_fb_; }
uint32_t offset_data() const { return offset_data_; }
ConstEntry const_entry(size_t i) const;
FuncRow func_row(size_t i) const;
const uint8_t* code_bytes() const;
size_t code_len() const;
const uint8_t* data_bytes() const;
size_t data_len() const;
private:
StbView() : buf_(0), len_(0) {}
uint32_t offs_of_const() const; // offset_constfrom 已校验段起点)
const uint8_t* buf_;
size_t len_;
bool ok_ = false;
std::string err_;
uint32_t cycle_limit_ = 0;
uint32_t dt_ms_ = 0;
uint64_t project_hash_ = 0;
uint32_t entry_fn_id_ = 0;
uint32_t n_globals_ = 0;
uint32_t n_consts_ = 0;
uint32_t n_funcs_ = 0;
uint32_t offset_code_ = 0;
uint32_t offset_fb_ = 0;
uint32_t offset_data_ = 0;
};
// 文件读写(纯字节;校验交给 StbView)
bool read_stb_file(const char* path, std::vector<uint8_t>* out, std::string* err);
bool write_stb_file(const char* path, const std::vector<uint8_t>& img, std::string* err);
// sidecar 生成与写文件(I/O 绑定 var → 槽号 → channel/bit
std::string make_sidecar(const std::vector<IoBinding>& bindings);
bool write_sidecar_file(const char* path, const std::vector<IoBinding>& bindings,
std::string* err);
}
+65
View File
@@ -0,0 +1,65 @@
/**
* @file Codec.cpp
* @brief 指令字编解码 + 配置驱动反汇编(编译器侧)
* @author
* @date 2026-08-21
*/
#include "compiler/Codec.h"
#include <cstdio>
namespace compiler {
Instr pack(uint8_t opcode, uint8_t rd, uint8_t a, uint8_t b) {
return static_cast<uint32_t>(opcode)
| (static_cast<uint32_t>(rd) << 8)
| (static_cast<uint32_t>(a) << 16)
| (static_cast<uint32_t>(b) << 24);
}
uint8_t op_of(Instr w) { return static_cast<uint8_t>(w & 0xFFu); }
uint8_t rd_of(Instr w) { return static_cast<uint8_t>((w >> 8) & 0xFFu); }
uint8_t a_of(Instr w) { return static_cast<uint8_t>((w >> 16) & 0xFFu); }
uint8_t b_of(Instr w) { return static_cast<uint8_t>((w >> 24) & 0xFFu); }
uint16_t imm16_of(Instr w) {
return static_cast<uint16_t>(a_of(w) | (static_cast<uint16_t>(b_of(w)) << 8));
}
int16_t off16_of(Instr w) { return static_cast<int16_t>(imm16_of(w)); }
void disasm(const MachineConfig& cfg, Instr w, char* out, size_t cap) {
if (cap == 0) {
return;
}
out[0] = '\0';
const ConfigOp* op = cfg.find_op_by_code(op_of(w));
if (op == nullptr) {
snprintf(out, cap, "??? 0x%08x", static_cast<unsigned>(w));
return;
}
const uint8_t rd = rd_of(w);
const std::string& f = op->format;
if (f == "RR") {
snprintf(out, cap, "%s r%u, r%u", op->name.c_str(), static_cast<unsigned>(rd),
static_cast<unsigned>(a_of(w)));
} else if (f == "RRR") {
snprintf(out, cap, "%s r%u, r%u, r%u", op->name.c_str(), static_cast<unsigned>(rd),
static_cast<unsigned>(a_of(w)), static_cast<unsigned>(b_of(w)));
} else if (f == "IMM" || f == "SLOT") {
snprintf(out, cap, "%s r%u, %u", op->name.c_str(), static_cast<unsigned>(rd),
static_cast<unsigned>(imm16_of(w)));
} else if (f == "JMP") {
snprintf(out, cap, "%s %+d", op->name.c_str(), static_cast<int>(off16_of(w)));
} else if (f == "JC") {
snprintf(out, cap, "%s r%u, %+d", op->name.c_str(), static_cast<unsigned>(rd),
static_cast<int>(off16_of(w)));
} else if (f == "CALL" || f == "CAL") {
snprintf(out, cap, "%s %u", op->name.c_str(), static_cast<unsigned>(imm16_of(w)));
} else { // NONE
snprintf(out, cap, "%s", op->name.c_str());
}
}
} // namespace compiler
+195 -332
View File
@@ -1,38 +1,32 @@
/** /**
* @file Codegen.cpp * @file Codegen.cpp
* @brief 寄存器码生成(12.8,切片 7+ FB 实例、内联展开、CAL_* * @brief 寄存器码生成(配置驱动;compiler 不依赖 isa
* @author * @author
* @date 2026-08-21 * @date 2026-08-21
* *
* @details 设计说明(详见 Doc/compiler/寄存器码.md): * @details 设计说明(详见 Doc/compiler/寄存器码.md 与 Doc/isa/指令配置.md):
* - 切片 1~6:帧/字面量/MOVE/RET;全局数据区;短路 AND/ORCMP 与 IFWHILE 与四则; * - 指令 opcode / 类型 tag / FB 布局全部来自 machine.tomlMachineConfig
* FUNCTION 与 CALL(调用约定 r0/r1..r7/r8+)。 * - 帧/字面量/MOVE/RET;全局数据区;短路 AND/ORCMP 与 IFWHILE 与四则;
* - 切片 7:FB 实例 → 数据区实例块(跨周期持久);字段槽号 = 实例基槽 + 字段序号 * FUNCTION 与 CALL(调用约定 r0/r1..r7/r8+
* (编译期算死);字段读写走 LOAD_GLOBAL / STORE_GLOBAL * - FB 实例 → 数据区实例块(跨周期持久);字段槽号 = 实例基槽 + 字段序号
* 内建 TON/TOF/CTU → 实参写字段后 CAL_* <实例基槽> * - 用户 FB 调用点内联展开;内建 FB → CAL_* <实例基槽>
* 用户 FB → 调用点内联展开(实参写字段 → FB 体以实例字段为变量编译) * - 数据段每槽 8 字节定宽,指令 slot = 槽号
* - 方案 a(12.9 前置):数据段**每槽 8 字节定宽**,指令 slot = 槽号(无映射)
* *
* 函数清单: * 函数清单:
* - put_le32 / put_le64 小端写入映像缓冲 * - Builder::Builder (构造)存工程/源文件/链接结果/机器配置/输出,收集 io 绑定
* - Builder::Builder (构造)存工程/源文件/链接结果/输出,收集 io 绑定分类 * - Builder::run 布局数据区 → 逐 POU 建函数 → 拼映像
* - Builder::run 布局数据区(全局 + 实例)→ 逐 POU 建函数 → 拼映像 * - Builder::fail / opc 错误 / 按名查配置 opcode
* - Builder::fail 组装 "codegen error: <msg>" 返回 false * - E_rr / E_rrr / E_imm / E_slot / E_jmp / E_jc / E_call / E_ret 指令发射(配置 opcode
* - find_pou / find_fn_id 按名查 POU AST / fn_id * - find_pou / find_fn_id 按名查 POU AST / fn_id
* - build_function 编译一个 POU(PROGRAM/FUNCTION/FB 骨架;帧约定分配) * - build_function 编译一个 POU(帧约定分配FB 占位
* - begin_stmt / alloc_temp 临时寄存器:语句内递增、语句结束复用基址 * - begin_stmt / alloc_temp 临时寄存器管理
* - compile_stmt 语句编译(赋值 / IF / WHILE / FB 调用) * - compile_stmt 语句编译(赋值 / IF / WHILE / FB 调用)
* - compile_if / compile_while IF 链 / WHILE 回环 * - compile_if / compile_while / compile_fb_call / compile_fb_inline
* - compile_fb_call FB 调用:内建 CAL_* 或用户内联展开 * - store_target / compile_expr 左值存储 / 表达式编译
* - compile_fb_inline 内联编译用户 FB 体(实例字段为变量) * - arith_name / cmp_name / load_name / store_name 操作码名选择
* - store_target 赋值左值:STORE_* 到全局槽(或内联模式实例字段 * - patch_jump 回填跳转偏移(相对下一条
* - compile_expr 表达式编译到寄存器(含 Field 字段读) * - layout_data / layout_fb_instances / instance_of / init_of
* - cmp_op / arith_op / load_op / store_op 操作码选择 * - const_id 取常量表 id(tag 来自配置类型表)
* - global_offset 槽号恒等(方案 a 无映射)
* - instance_of 实例 → 字段槽号表
* - patch_jump 回填跳转偏移(相对下一条指令)
* - layout_data / layout_fb_instances / init_of 数据区布局(8 字节定宽槽)
* - const_id 取常量表 id(无则追加)
* - assemble_image 拼头 + 常量表 + 函数表 + 字节码 + 数据段 * - assemble_image 拼头 + 常量表 + 函数表 + 字节码 + 数据段
* - codegen_project 对外入口 * - codegen_project 对外入口
*/ */
@@ -45,31 +39,15 @@
#include <string> #include <string>
#include <vector> #include <vector>
#include "isa/Encode.h" #include "compiler/Codec.h"
#include "isa/Image.h" #include "compiler/MachineConfig.h"
#include "isa/Instr.h" #include "compiler/Stb.h"
#include "isa/Types.h"
namespace compiler { namespace compiler {
namespace { namespace {
// ---- 小端写入(Image.cpp 内部实现不可见,这里自带最小版)----
void put_le32(std::vector<uint8_t>& b, size_t off, uint32_t v) {
b[off + 0] = static_cast<uint8_t>(v & 0xFFu);
b[off + 1] = static_cast<uint8_t>((v >> 8) & 0xFFu);
b[off + 2] = static_cast<uint8_t>((v >> 16) & 0xFFu);
b[off + 3] = static_cast<uint8_t>((v >> 24) & 0xFFu);
}
void put_le64(std::vector<uint8_t>& b, size_t off, uint64_t v) {
for (int i = 0; i < 8; ++i) {
b[off + i] = static_cast<uint8_t>((v >> (8 * i)) & 0xFFu);
}
}
/** /**
* @brief 代码生成器(切片 1 * @brief 代码生成器
*/ */
class Builder { class Builder {
public: public:
@@ -78,14 +56,16 @@ namespace {
* @param proj 工程定义(cycle_limit / dt_ms / 哈希用) * @param proj 工程定义(cycle_limit / dt_ms / 哈希用)
* @param units 全部源文件的 AST * @param units 全部源文件的 AST
* @param link 链接结果(POU 顺序 / 符号) * @param link 链接结果(POU 顺序 / 符号)
* @param cfg 机器定义(machine.toml 强校验后;指令 opcode / 类型 tag / FB 布局)
* @param image 输出映像字节 * @param image 输出映像字节
* @param err 错误输出;可为 nullptr(静默) * @param err 错误输出;可为 nullptr(静默)
*/ */
Builder(const Project& proj, const std::vector<SourceUnit>& units, Builder(const Project& proj, const std::vector<SourceUnit>& units,
const LinkResult& link, std::vector<uint8_t>* image, std::string* err) const LinkResult& link, const MachineConfig& cfg,
: proj_(proj), units_(units), link_(link), image_(image), err_(err) { std::vector<uint8_t>* image, std::string* err)
: proj_(proj), units_(units), link_(link), cfg_(cfg), image_(image), err_(err) {
// io 绑定分类(名已折小写;不创造变量,只影响操作码选择) // io 绑定分类(名已折小写;不创造变量,只影响操作码选择)
for (const isa::IoBinding& b : proj_.io) { for (const IoBinding& b : proj_.io) {
std::string key = b.var; std::string key = b.var;
for (char& ch : key) { for (char& ch : key) {
ch = static_cast<char>(std::tolower(static_cast<unsigned char>(ch))); ch = static_cast<char>(std::tolower(static_cast<unsigned char>(ch)));
@@ -100,7 +80,6 @@ namespace {
/** /**
* @brief 布局数据区 → 布局 FB 实例 → 逐 POU 建函数 → 拼映像 * @brief 布局数据区 → 布局 FB 实例 → 逐 POU 建函数 → 拼映像
* @details 偏移先定(函数编译引用 global_offset / instance_of
* @return true 成功;falseerr 已写,前缀 "codegen error" * @return true 成功;falseerr 已写,前缀 "codegen error"
*/ */
bool run() { bool run() {
@@ -127,11 +106,6 @@ namespace {
} }
private: private:
/**
* @brief 组装错误消息并返回 false
* @param msg 错误描述(不含前缀)
* @return 恒 false
*/
bool fail(const std::string& msg) { bool fail(const std::string& msg) {
if (err_) { if (err_) {
*err_ = "codegen error: " + msg; *err_ = "codegen error: " + msg;
@@ -140,10 +114,42 @@ namespace {
} }
/** /**
* @brief 按名查 POU AST * @brief 按名查配置 opcodeMachineConfig 强校验保证存在)
* @param name POU 名(小写 * @param name 指令名(大写,如 "ADD"
* @return POU 指针;未找到返回 nullptr * @return opcode(找不到返回 0,不应发生)
*/ */
uint8_t opc(const char* name) const {
const ConfigOp* op = cfg_.find_op(name);
return op ? static_cast<uint8_t>(op->opcode) : 0;
}
// ---- 指令发射(配置 opcode----
Instr E_rr(const char* name, uint8_t rd, uint8_t rs) {
return pack(opc(name), rd, rs, 0);
}
Instr E_rrr(const char* name, uint8_t rd, uint8_t ra, uint8_t rb) {
return pack(opc(name), rd, ra, rb);
}
Instr E_imm(const char* name, uint8_t rd, uint16_t imm) {
return pack(opc(name), rd, static_cast<uint8_t>(imm & 0xFFu),
static_cast<uint8_t>((imm >> 8) & 0xFFu));
}
Instr E_slot(const char* name, uint8_t rd, uint16_t slot) {
return E_imm(name, rd, slot);
}
Instr E_jmp(int16_t off) {
const uint16_t u = static_cast<uint16_t>(off);
return pack(opc("JMP"), 0, static_cast<uint8_t>(u & 0xFFu),
static_cast<uint8_t>((u >> 8) & 0xFFu));
}
Instr E_jc(const char* name, uint8_t r, int16_t off) {
const uint16_t u = static_cast<uint16_t>(off);
return pack(opc(name), r, static_cast<uint8_t>(u & 0xFFu),
static_cast<uint8_t>((u >> 8) & 0xFFu));
}
Instr E_call(uint16_t fn_id) { return E_imm("CALL", 0, fn_id); }
Instr E_ret() { return pack(opc("RET"), 0, 0, 0); }
const POU* find_pou(const std::string& name) const { const POU* find_pou(const std::string& name) const {
for (const SourceUnit& u : units_) { for (const SourceUnit& u : units_) {
for (const POU& p : u.ast.pous) { for (const POU& p : u.ast.pous) {
@@ -159,7 +165,7 @@ namespace {
struct FuncCtx { struct FuncCtx {
std::string name; std::string name;
std::string pou_name; // 所属 POU 名(实例查找键) std::string pou_name; // 所属 POU 名(实例查找键)
std::vector<isa::Instr> code; // 字节码(函数表 code_offset 相对此段) std::vector<Instr> code; // 字节码
std::map<std::string, uint8_t> regs; // 变量名 → 帧寄存器 std::map<std::string, uint8_t> regs; // 变量名 → 帧寄存器
uint8_t nlocals = 0; // 变量区终点 = 临时寄存器起始 uint8_t nlocals = 0; // 变量区终点 = 临时寄存器起始
uint8_t nregs = 0; // 峰值(变量 + 临时) uint8_t nregs = 0; // 峰值(变量 + 临时)
@@ -168,24 +174,15 @@ namespace {
std::string result_name; // FUNCTION 名(结果寄存器映射) std::string result_name; // FUNCTION 名(结果寄存器映射)
}; };
// FB 实例:字段名 → 数据区字节地址 // FB 实例:字段名 → 数据区槽号
struct InstFields { struct InstFields {
std::map<std::string, uint32_t> field_addr; std::map<std::string, uint32_t> field_addr;
uint32_t base = 0; uint32_t base = 0;
std::string type_name; // 实例的 FB 类型名(内建 / 用户) std::string type_name; // 实例的 FB 类型名(内建 / 用户)
}; };
/**
* @brief 语句开始:临时寄存器基址复用
* @param f 当前函数
*/
void begin_stmt(FuncCtx& f) { f.temp_used = 0; } void begin_stmt(FuncCtx& f) { f.temp_used = 0; }
/**
* @brief 分配一个临时寄存器(语句内递增)
* @param f 当前函数
* @return 临时寄存器号(可能更新 nregs 峰值)
*/
uint8_t alloc_temp(FuncCtx& f) { uint8_t alloc_temp(FuncCtx& f) {
const uint8_t r = f.nlocals + f.temp_used; const uint8_t r = f.nlocals + f.temp_used;
++f.temp_used; ++f.temp_used;
@@ -201,16 +198,12 @@ namespace {
* FB 不生成业务字节码(调用点内联),只出空函数占位保持 fn_id 一致; * FB 不生成业务字节码(调用点内联),只出空函数占位保持 fn_id 一致;
* 帧约定(见 Doc/compiler/寄存器码.md):结果 r0、输入 r1..r7(只读)、 * 帧约定(见 Doc/compiler/寄存器码.md):结果 r0、输入 r1..r7(只读)、
* 变量与临时全部从 r8 起 * 变量与临时全部从 r8 起
* @param pou POU AST
* @param f 输出函数编译态
* @return true 成功;falseerr 已写)
*/ */
bool build_function(const POU& pou, FuncCtx* f) { bool build_function(const POU& pou, FuncCtx* f) {
if (pou.kind == PouKind::FunctionBlock) { if (pou.kind == PouKind::FunctionBlock) {
// FB 内联展开(见 compile_fb_call),此处只占 fn_id
f->nlocals = 8; f->nlocals = 8;
f->nregs = 8; f->nregs = 8;
f->code.push_back(isa::enc_ret()); f->code.push_back(E_ret());
return true; return true;
} }
f->is_function = pou.kind == PouKind::Function; f->is_function = pou.kind == PouKind::Function;
@@ -229,16 +222,13 @@ namespace {
} }
uint8_t r = 8; // 变量/临时基址(调用约定区 r0..r7 不占用) uint8_t r = 8; // 变量/临时基址(调用约定区 r0..r7 不占用)
for (const VarBlock& b : pou.blocks) { for (const VarBlock& b : pou.blocks) {
// External / Global 走数据区槽,不占帧寄存器
if (b.section == VarSection::External || b.section == VarSection::Global) { if (b.section == VarSection::External || b.section == VarSection::Global) {
continue; continue;
} }
// FUNCTION 输入已分配(r1..r7
if (f->is_function && b.section == VarSection::Input) { if (f->is_function && b.section == VarSection::Input) {
continue; continue;
} }
for (const VarDecl& d : b.vars) { for (const VarDecl& d : b.vars) {
// FB 实例走数据区(layout_fb_instances),不占帧寄存器
if (d.type.kind == TypeKind::FbUser || d.type.kind == TypeKind::FbBuiltin) { if (d.type.kind == TypeKind::FbUser || d.type.kind == TypeKind::FbBuiltin) {
continue; continue;
} }
@@ -253,16 +243,10 @@ namespace {
return false; return false;
} }
} }
f->code.push_back(isa::enc_ret()); f->code.push_back(E_ret());
return true; return true;
} }
/**
* @brief 编译一条语句(切片 4:赋值 / IF)
* @param f 当前函数
* @param st 语句 AST
* @return true 成功;falseerr 已写)
*/
bool compile_stmt(FuncCtx& f, const Stmt& st) { bool compile_stmt(FuncCtx& f, const Stmt& st) {
if (st.kind == StmtKind::If) { if (st.kind == StmtKind::If) {
return compile_if(f, st); return compile_if(f, st);
@@ -274,7 +258,7 @@ namespace {
return compile_fb_call(f, st); return compile_fb_call(f, st);
} }
if (st.kind != StmtKind::Assign) { if (st.kind != StmtKind::Assign) {
return fail("statement not supported in slice 7"); return fail("statement not supported");
} }
// 内联 FB 体内:左值可能是实例字段(STORE_GLOBAL // 内联 FB 体内:左值可能是实例字段(STORE_GLOBAL
if (inline_fields_) { if (inline_fields_) {
@@ -284,32 +268,20 @@ namespace {
if (!compile_expr(f, *st.value, t)) { if (!compile_expr(f, *st.value, t)) {
return false; return false;
} }
f.code.push_back(isa::enc_slot(isa::Op::STORE_GLOBAL, t, fit->second)); f.code.push_back(E_slot("STORE_GLOBAL", t, fit->second));
return true; return true;
} }
} }
const auto it = f.regs.find(st.target); const auto it = f.regs.find(st.target);
if (it == f.regs.end()) { if (it == f.regs.end()) {
// 全局 / 外部左值 → STORE_* 到数据区
return store_target(f, st.target, *st.value); return store_target(f, st.target, *st.value);
} }
// 函数输入只读(调用约定,见 Doc/compiler/寄存器码.md
if (f.is_function && it->second >= 1 && it->second <= 7) { if (f.is_function && it->second >= 1 && it->second <= 7) {
return fail("cannot write function input '" + st.target + "'"); return fail("cannot write function input '" + st.target + "'");
} }
const uint8_t rd = it->second; return compile_expr(f, *st.value, it->second);
return compile_expr(f, *st.value, rd);
} }
/**
* @brief 编译 FB 调用
* @details 实参求值到临时 → STORE_GLOBAL 到实例字段;
* 内建 TON/TOF/CTU → CAL_* <实例偏移>
* 用户 FB → compile_fb_inline 内联展开
* @param f 当前函数
* @param st FbCall 语句 AST
* @return true 成功;falseerr 已写)
*/
bool compile_fb_call(FuncCtx& f, const Stmt& st) { bool compile_fb_call(FuncCtx& f, const Stmt& st) {
const InstFields* inst = instance_of(f.pou_name, st.instance); const InstFields* inst = instance_of(f.pou_name, st.instance);
if (inst == nullptr) { if (inst == nullptr) {
@@ -324,22 +296,15 @@ namespace {
if (!compile_expr(f, *a.value, t)) { if (!compile_expr(f, *a.value, t)) {
return false; return false;
} }
f.code.push_back(isa::enc_slot(isa::Op::STORE_GLOBAL, t, it->second)); f.code.push_back(E_slot("STORE_GLOBAL", t, it->second));
} }
if (inst->type_name == "ton" || inst->type_name == "tof" || // 内建 FB:配置里必须有对应 op 行(强校验保证)→ 直接查配置 opcode
inst->type_name == "tp" || inst->type_name == "ctu" || const std::string& tn = inst->type_name;
inst->type_name == "ctd" || inst->type_name == "ctud" || if (cfg_.find_op("CAL_" + uppercase_of(tn)) != nullptr &&
inst->type_name == "r_trig" || inst->type_name == "f_trig") { (tn == "ton" || tn == "tof" || tn == "tp" || tn == "ctu" ||
const isa::Op op = tn == "ctd" || tn == "ctud" || tn == "r_trig" || tn == "f_trig")) {
inst->type_name == "ton" ? isa::Op::CAL_TON const std::string opname = "CAL_" + uppercase_of(tn);
: inst->type_name == "tof" ? isa::Op::CAL_TOF f.code.push_back(E_slot(opname.c_str(), 0, inst->base));
: inst->type_name == "tp" ? isa::Op::CAL_TP
: inst->type_name == "ctu" ? isa::Op::CAL_CTU
: inst->type_name == "ctd" ? isa::Op::CAL_CTD
: inst->type_name == "ctud" ? isa::Op::CAL_CTUD
: inst->type_name == "r_trig" ? isa::Op::CAL_R_TRIG
: isa::Op::CAL_F_TRIG;
f.code.push_back(isa::enc_slot(op, 0, inst->base));
return true; return true;
} }
const POU* fb = find_pou(inst->type_name); const POU* fb = find_pou(inst->type_name);
@@ -349,15 +314,14 @@ namespace {
return compile_fb_inline(f, *fb, *inst); return compile_fb_inline(f, *fb, *inst);
} }
/** static std::string uppercase_of(const std::string& s) {
* @brief 内联编译用户 FB 体 std::string out = s;
* @details 切到实例字段上下文(VarRef/赋值左值映射到字段地址); for (char& ch : out) {
* 表达式临时与调用方共用 r8+;字段读写全走 LOAD/STORE_GLOBAL ch = static_cast<char>(std::toupper(static_cast<unsigned char>(ch)));
* @param f 当前函数 }
* @param fb FB 类型 POU return out;
* @param inst 本实例字段表 }
* @return true 成功;falseerr 已写)
*/
bool compile_fb_inline(FuncCtx& f, const POU& fb, const InstFields& inst) { bool compile_fb_inline(FuncCtx& f, const POU& fb, const InstFields& inst) {
const InstFields* saved = inline_fields_; const InstFields* saved = inline_fields_;
inline_fields_ = &inst; inline_fields_ = &inst;
@@ -371,14 +335,6 @@ namespace {
return true; return true;
} }
/**
* @brief 编译 WHILE
* @details L_loop: <cond→t> JF t, L_end <body> JMP L_loop L_end:
* 条件 JF 与回环 JMP 双回填
* @param f 当前函数
* @param st WHILE 语句 AST
* @return true 成功;falseerr 已写)
*/
bool compile_while(FuncCtx& f, const Stmt& st) { bool compile_while(FuncCtx& f, const Stmt& st) {
const size_t loop = f.code.size(); const size_t loop = f.code.size();
const uint8_t t = alloc_temp(f); const uint8_t t = alloc_temp(f);
@@ -386,7 +342,7 @@ namespace {
return false; return false;
} }
const size_t jf_idx = f.code.size(); const size_t jf_idx = f.code.size();
f.code.push_back(isa::enc_jc(isa::Op::JF, t, 0)); // 假 → L_end f.code.push_back(E_jc("JF", t, 0));
for (const Stmt& s : st.body) { for (const Stmt& s : st.body) {
begin_stmt(f); begin_stmt(f);
if (!compile_stmt(f, s)) { if (!compile_stmt(f, s)) {
@@ -394,29 +350,20 @@ namespace {
} }
} }
const size_t jmp_idx = f.code.size(); const size_t jmp_idx = f.code.size();
f.code.push_back(isa::enc_jmp(0)); // 回环 f.code.push_back(E_jmp(0));
patch_jump(f, jmp_idx, loop); patch_jump(f, jmp_idx, loop);
patch_jump(f, jf_idx, f.code.size()); patch_jump(f, jf_idx, f.code.size());
return true; return true;
} }
/**
* @brief 编译 IF / ELSIF / ELSE
* @details 每分支:<cond→t> JF t, L_next → <body> → JMP L_end
* 最后一个分支无 else 时 JF 直接落 L_end(不补 JMP
* @param f 当前函数
* @param st IF 语句 AST
* @return true 成功;falseerr 已写)
*/
bool compile_if(FuncCtx& f, const Stmt& st) { bool compile_if(FuncCtx& f, const Stmt& st) {
// 条件链:IF 体 + ELSIF 各体;ELSE 体可选
std::vector<std::pair<const Expr*, const std::vector<Stmt>*>> branches; std::vector<std::pair<const Expr*, const std::vector<Stmt>*>> branches;
branches.push_back({st.cond.get(), &st.body}); branches.push_back({st.cond.get(), &st.body});
for (const IfBranch& b : st.elsifs) { for (const IfBranch& b : st.elsifs) {
branches.push_back({b.cond.get(), &b.body}); branches.push_back({b.cond.get(), &b.body});
} }
const bool has_else = !st.else_body.empty(); const bool has_else = !st.else_body.empty();
std::vector<size_t> end_jmps; // 各分支尾部 JMP(回填到 L_end) std::vector<size_t> end_jmps;
for (size_t i = 0; i < branches.size(); ++i) { for (size_t i = 0; i < branches.size(); ++i) {
const uint8_t t = alloc_temp(f); const uint8_t t = alloc_temp(f);
@@ -424,7 +371,7 @@ namespace {
return false; return false;
} }
const size_t jf_idx = f.code.size(); const size_t jf_idx = f.code.size();
f.code.push_back(isa::enc_jc(isa::Op::JF, t, 0)); // 假 → 下一分支 f.code.push_back(E_jc("JF", t, 0));
for (const Stmt& s : *branches[i].second) { for (const Stmt& s : *branches[i].second) {
begin_stmt(f); begin_stmt(f);
if (!compile_stmt(f, s)) { if (!compile_stmt(f, s)) {
@@ -433,10 +380,10 @@ namespace {
} }
if (i + 1 < branches.size() || has_else) { if (i + 1 < branches.size() || has_else) {
const size_t jmp_idx = f.code.size(); const size_t jmp_idx = f.code.size();
f.code.push_back(isa::enc_jmp(0)); // 分支尾部 → L_end f.code.push_back(E_jmp(0));
end_jmps.push_back(jmp_idx); end_jmps.push_back(jmp_idx);
} }
patch_jump(f, jf_idx, f.code.size()); // 回填 JF 到下一分支起点 patch_jump(f, jf_idx, f.code.size());
} }
if (has_else) { if (has_else) {
for (const Stmt& s : st.else_body) { for (const Stmt& s : st.else_body) {
@@ -453,13 +400,6 @@ namespace {
return true; return true;
} }
/**
* @brief 赋值左值:帧寄存器直写或 STORE_* 到全局槽
* @param f 当前函数
* @param target 左值名(全局 / 外部)
* @param value 右值表达式
* @return true 成功;falseerr 已写)
*/
bool store_target(FuncCtx& f, const std::string& target, const Expr& value) { bool store_target(FuncCtx& f, const std::string& target, const Expr& value) {
const auto git = link_.global_index.find(target); const auto git = link_.global_index.find(target);
if (git == link_.global_index.end()) { if (git == link_.global_index.end()) {
@@ -472,49 +412,36 @@ namespace {
if (!compile_expr(f, value, tmp)) { if (!compile_expr(f, value, tmp)) {
return false; return false;
} }
const uint16_t slot = global_offset(git->second); f.code.push_back(E_slot(store_name(target), tmp, static_cast<uint16_t>(git->second)));
f.code.push_back(isa::enc_slot(store_op(target), tmp, slot));
return true; return true;
} }
/**
* @brief 表达式编译到目标寄存器
* @details 字面量 → LOADK;帧变量 → MOVE;全局/外部 → LOAD_*io.input 用 LOAD_I);
* NOT → 求值后 NOT rd, tAND/OR → 短路跳转(JF/JT 跳过右侧)
* @param f 当前函数
* @param e 表达式 AST
* @param rd 目标寄存器
* @return true 成功;falseerr 已写)
*/
bool compile_expr(FuncCtx& f, const Expr& e, uint8_t rd) { bool compile_expr(FuncCtx& f, const Expr& e, uint8_t rd) {
if (e.kind == ExprKind::LitBool || e.kind == ExprKind::LitInt || if (e.kind == ExprKind::LitBool || e.kind == ExprKind::LitInt ||
e.kind == ExprKind::LitTime) { e.kind == ExprKind::LitTime) {
const isa::types::TypeTag tag = const char* type_name = e.kind == ExprKind::LitBool ? "BOOL"
e.kind == ExprKind::LitBool ? isa::types::Bool : e.kind == ExprKind::LitInt ? "INT"
: e.kind == ExprKind::LitInt ? isa::types::Int : "TIME";
: isa::types::Time; f.code.push_back(E_imm("LOADK", rd, const_id(type_name, e.int_value)));
f.code.push_back(
isa::enc_imm(isa::Op::LOADK, rd, const_id(tag, e.int_value)));
return true; return true;
} }
if (e.kind == ExprKind::VarRef) { if (e.kind == ExprKind::VarRef) {
// 内联 FB 体内:字段优先(实例字段为变量)
if (inline_fields_) { if (inline_fields_) {
const auto fit = inline_fields_->field_addr.find(e.name); const auto fit = inline_fields_->field_addr.find(e.name);
if (fit != inline_fields_->field_addr.end()) { if (fit != inline_fields_->field_addr.end()) {
f.code.push_back(isa::enc_slot(isa::Op::LOAD_GLOBAL, rd, fit->second)); f.code.push_back(E_slot("LOAD_GLOBAL", rd, fit->second));
return true; return true;
} }
} }
const auto sit = f.regs.find(e.name); const auto sit = f.regs.find(e.name);
if (sit != f.regs.end()) { if (sit != f.regs.end()) {
f.code.push_back(isa::enc_rr(isa::Op::MOVE, rd, sit->second)); f.code.push_back(E_rr("MOVE", rd, sit->second));
return true; return true;
} }
const auto git = link_.global_index.find(e.name); const auto git = link_.global_index.find(e.name);
if (git != link_.global_index.end()) { if (git != link_.global_index.end()) {
const uint16_t slot = global_offset(git->second); f.code.push_back(
f.code.push_back(isa::enc_slot(load_op(e.name), rd, slot)); E_slot(load_name(e.name), rd, static_cast<uint16_t>(git->second)));
return true; return true;
} }
return fail("no register or slot for '" + e.name + "'"); return fail("no register or slot for '" + e.name + "'");
@@ -528,7 +455,7 @@ namespace {
if (fit == inst->field_addr.end()) { if (fit == inst->field_addr.end()) {
return fail("unknown field '" + e.field + "' for '" + e.name + "'"); return fail("unknown field '" + e.field + "' for '" + e.name + "'");
} }
f.code.push_back(isa::enc_slot(isa::Op::LOAD_GLOBAL, rd, fit->second)); f.code.push_back(E_slot("LOAD_GLOBAL", rd, fit->second));
return true; return true;
} }
if (e.kind == ExprKind::Add || e.kind == ExprKind::Sub || if (e.kind == ExprKind::Add || e.kind == ExprKind::Sub ||
@@ -541,7 +468,7 @@ namespace {
if (!compile_expr(f, *e.rhs, r)) { if (!compile_expr(f, *e.rhs, r)) {
return false; return false;
} }
f.code.push_back(isa::enc_rrr(arith_op(e.kind), rd, l, r)); f.code.push_back(E_rrr(arith_name(e.kind), rd, l, r));
return true; return true;
} }
if (e.kind == ExprKind::Cmp) { if (e.kind == ExprKind::Cmp) {
@@ -553,7 +480,7 @@ namespace {
if (!compile_expr(f, *e.rhs, r)) { if (!compile_expr(f, *e.rhs, r)) {
return false; return false;
} }
f.code.push_back(isa::enc_rrr(cmp_op(e.op), rd, l, r)); f.code.push_back(E_rrr(cmp_name(e.op), rd, l, r));
return true; return true;
} }
if (e.kind == ExprKind::Not) { if (e.kind == ExprKind::Not) {
@@ -561,27 +488,25 @@ namespace {
if (!compile_expr(f, *e.operand, t)) { if (!compile_expr(f, *e.operand, t)) {
return false; return false;
} }
f.code.push_back(isa::enc_rr(isa::Op::NOT, rd, t)); f.code.push_back(E_rr("NOT", rd, t));
return true; return true;
} }
if (e.kind == ExprKind::And || e.kind == ExprKind::Or) { if (e.kind == ExprKind::And || e.kind == ExprKind::Or) {
// 短路:左侧结果在 rd;AND 为假 / OR 为真时跳过右侧
if (!compile_expr(f, *e.lhs, rd)) { if (!compile_expr(f, *e.lhs, rd)) {
return false; return false;
} }
const isa::Op jop = (e.kind == ExprKind::And) ? isa::Op::JF : isa::Op::JT; const char* jname = (e.kind == ExprKind::And) ? "JF" : "JT";
const size_t jmp_idx = f.code.size(); const size_t jmp_idx = f.code.size();
f.code.push_back(isa::enc_jc(jop, rd, 0)); // 偏移留洞,稍后回填 f.code.push_back(E_jc(jname, rd, 0));
const uint8_t t = alloc_temp(f); const uint8_t t = alloc_temp(f);
if (!compile_expr(f, *e.rhs, t)) { if (!compile_expr(f, *e.rhs, t)) {
return false; return false;
} }
f.code.push_back(isa::enc_rr(isa::Op::MOVE, rd, t)); f.code.push_back(E_rr("MOVE", rd, t));
patch_jump(f, jmp_idx, f.code.size()); patch_jump(f, jmp_idx, f.code.size());
return true; return true;
} }
if (e.kind == ExprKind::Call) { if (e.kind == ExprKind::Call) {
// 调用约定:实参求值到 r8+ 临时 → MOVE 到 r1..r7 → CALL → 结果 r0
if (e.args.size() > 7) { if (e.args.size() > 7) {
return fail("too many arguments (max 7)"); return fail("too many arguments (max 7)");
} }
@@ -598,21 +523,15 @@ namespace {
return fail("no fn_id for '" + e.name + "'"); return fail("no fn_id for '" + e.name + "'");
} }
for (size_t i = 0; i < args.size(); ++i) { for (size_t i = 0; i < args.size(); ++i) {
f.code.push_back( f.code.push_back(E_rr("MOVE", static_cast<uint8_t>(1 + i), args[i]));
isa::enc_rr(isa::Op::MOVE, static_cast<uint8_t>(1 + i), args[i]));
} }
f.code.push_back(isa::enc_call(static_cast<uint16_t>(fn_id))); f.code.push_back(E_call(static_cast<uint16_t>(fn_id)));
f.code.push_back(isa::enc_rr(isa::Op::MOVE, rd, 0)); f.code.push_back(E_rr("MOVE", rd, 0));
return true; return true;
} }
return fail("expression not supported in slice 6"); return fail("expression not supported");
} }
/**
* @brief 按名查 fn_idPOU 收集序下标)
* @param name POU 名(小写)
* @return fn_id;未找到返回 -1
*/
int find_fn_id(const std::string& name) const { int find_fn_id(const std::string& name) const {
for (size_t i = 0; i < link_.scopes.size(); ++i) { for (size_t i = 0; i < link_.scopes.size(); ++i) {
if (link_.scopes[i].name == name) { if (link_.scopes[i].name == name) {
@@ -622,87 +541,43 @@ namespace {
return -1; return -1;
} }
/**
* @brief 回填跳转偏移(相对下一条指令)
* @param f 当前函数
* @param idx 跳转指令下标(占位)
* @param target_idx 目标指令下标
*/
void patch_jump(FuncCtx& f, size_t idx, size_t target_idx) { void patch_jump(FuncCtx& f, size_t idx, size_t target_idx) {
const int16_t off = static_cast<int16_t>( const int16_t off = static_cast<int16_t>(
static_cast<int64_t>(target_idx) - (static_cast<int64_t>(idx) + 1)); static_cast<int64_t>(target_idx) - (static_cast<int64_t>(idx) + 1));
const isa::Instr w = f.code[idx]; const Instr w = f.code[idx];
f.code[idx] = isa::pack(isa::op(w), isa::rd(w), const uint16_t u = static_cast<uint16_t>(off);
static_cast<uint8_t>(static_cast<uint16_t>(off) & 0xFFu), f.code[idx] = pack(op_of(w), rd_of(w), static_cast<uint8_t>(u & 0xFFu),
static_cast<uint8_t>((static_cast<uint16_t>(off) >> 8) & 0xFFu)); static_cast<uint8_t>((u >> 8) & 0xFFu));
} }
/** // ---- 操作码名(MachineConfig 已强校验存在)----
* @brief 算术运算 → 操作码 static const char* arith_name(ExprKind k) {
* @param k 表达式种类(Add/Sub/Mul/Div
* @return 对应操作码
*/
isa::Op arith_op(ExprKind k) const {
switch (k) { switch (k) {
case ExprKind::Add: return isa::Op::ADD; case ExprKind::Add: return "ADD";
case ExprKind::Sub: return isa::Op::SUB; case ExprKind::Sub: return "SUB";
case ExprKind::Mul: return isa::Op::MUL; case ExprKind::Mul: return "MUL";
case ExprKind::Div: return isa::Op::DIV; case ExprKind::Div: return "DIV";
default: return isa::Op::ADD; default: return "ADD";
} }
} }
static const char* cmp_name(BinOp op) {
/**
* @brief 比较符 → CMP_xx 操作码
* @param op 比较运算
* @return 对应操作码(CMP_EQ / CMP_NE / CMP_LT / CMP_LE / CMP_GT / CMP_GE
*/
isa::Op cmp_op(BinOp op) const {
switch (op) { switch (op) {
case BinOp::Eq: return isa::Op::CMP_EQ; case BinOp::Eq: return "CMP_EQ";
case BinOp::Ne: return isa::Op::CMP_NE; case BinOp::Ne: return "CMP_NE";
case BinOp::Lt: return isa::Op::CMP_LT; case BinOp::Lt: return "CMP_LT";
case BinOp::Le: return isa::Op::CMP_LE; case BinOp::Le: return "CMP_LE";
case BinOp::Gt: return isa::Op::CMP_GT; case BinOp::Gt: return "CMP_GT";
case BinOp::Ge: return isa::Op::CMP_GE; case BinOp::Ge: return "CMP_GE";
default: return isa::Op::CMP_EQ; // Add/Sub/Mul/Div 不经此函数 default: return "CMP_EQ";
} }
} }
const char* load_name(const std::string& name) const {
return io_input_.count(name) ? "LOAD_I" : "LOAD_GLOBAL";
}
const char* store_name(const std::string& name) const {
return io_output_.count(name) ? "STORE_Q" : "STORE_GLOBAL";
}
/**
* @brief 按 io 绑定选读取操作码
* @param name 变量名(小写)
* @return io.input 绑定 → LOAD_I;否则 LOAD_GLOBAL
*/
isa::Op load_op(const std::string& name) const {
return io_input_.count(name) ? isa::Op::LOAD_I : isa::Op::LOAD_GLOBAL;
}
/**
* @brief 按 io 绑定选写入操作码
* @param name 变量名(小写)
* @return io.output 绑定 → STORE_Q;否则 STORE_GLOBAL
*/
isa::Op store_op(const std::string& name) const {
return io_output_.count(name) ? isa::Op::STORE_Q : isa::Op::STORE_GLOBAL;
}
/**
* @brief 逻辑槽号(符号表 address)即指令 slot(方案 a:8 字节定宽,无映射)
* @param slot 全局逻辑槽号
* @return 同值(u16 可容,layout_data 已校验)
*/
uint16_t global_offset(uint32_t slot) const {
return static_cast<uint16_t>(slot);
}
/**
* @brief 布局全局数据区:每槽 8 字节定宽(方案 a)
* @details 槽号 = 符号表 address(声明序,12.6);
* BOOL 用低 1 字节、INT 用低 2 字节(小端)、TIME 全 8 字节;
* 初值取自 GVL 声明的 has_init(无则 0
* @return true 成功;false(槽号越界,err 已写)
*/
bool layout_data() { bool layout_data() {
for (const Symbol& s : link_.globals) { for (const Symbol& s : link_.globals) {
if (s.address > 0xFFFF) { if (s.address > 0xFFFF) {
@@ -714,30 +589,33 @@ namespace {
const int64_t v = has_init ? init : 0; const int64_t v = has_init ? init : 0;
const size_t off = static_cast<size_t>(s.address) * 8; const size_t off = static_cast<size_t>(s.address) * 8;
data_.resize(off + 8, 0); data_.resize(off + 8, 0);
if (s.type_name == "int") { // 初值按基元宽度写(类型元数据来自配置别名 → base)
const ConfigType* ct = cfg_.find_type(uppercase_of(s.type_name));
if (ct == nullptr) {
return fail("no type in machine.toml for '" + s.type_name + "'");
}
const PrimType* prim = MachineConfig::find_prim(ct->base);
if (prim == nullptr) {
return fail("no prim '" + ct->base + "' for type '" + ct->name + "'");
}
if (prim->width == 2 && !prim->is_float) {
const uint16_t iv = static_cast<uint16_t>(v); const uint16_t iv = static_cast<uint16_t>(v);
data_[off] = static_cast<uint8_t>(iv & 0xFFu); data_[off] = static_cast<uint8_t>(iv & 0xFFu);
data_[off + 1] = static_cast<uint8_t>((iv >> 8) & 0xFFu); data_[off + 1] = static_cast<uint8_t>((iv >> 8) & 0xFFu);
} else if (s.type_name == "time") { } else if (prim->width >= 8 && !prim->is_float) {
const uint64_t tv = static_cast<uint64_t>(v); const uint64_t tv = static_cast<uint64_t>(v);
for (int i = 0; i < 8; ++i) { for (int i = 0; i < 8; ++i) {
data_[off + i] = static_cast<uint8_t>((tv >> (8 * i)) & 0xFFu); data_[off + i] = static_cast<uint8_t>((tv >> (8 * i)) & 0xFFu);
} }
} else { // bool } else {
data_[off] = v ? 1 : 0; data_[off] = v ? 1 : 0;
} }
} }
return true; return true;
} }
/**
* @brief 布局 FB 实例块:数据区全局之后,按 POU 收集序 / 实例声明序
* @details 每实例:字段按布局段序各占一槽(8 字节定宽),
* 字段地址 = 实例基槽 + 字段序号;实例基槽 = 累计槽数
* @return true 成功;falseerr 已写)
*/
bool layout_fb_instances() { bool layout_fb_instances() {
uint32_t cur = static_cast<uint32_t>(data_.size() / 8); // 槽号起点 uint32_t cur = static_cast<uint32_t>(data_.size() / 8);
bool any = false; bool any = false;
for (const LinkResult::PouScope& sc : link_.scopes) { for (const LinkResult::PouScope& sc : link_.scopes) {
for (const Symbol& s : sc.syms) { for (const Symbol& s : sc.syms) {
@@ -749,7 +627,7 @@ namespace {
return fail("no layout for instance '" + s.name + "'"); return fail("no layout for instance '" + s.name + "'");
} }
InstFields inst; InstFields inst;
inst.base = cur; // 实例基槽(CAL_* 操作数) inst.base = cur;
inst.type_name = s.type_name; inst.type_name = s.type_name;
for (size_t i = 0; i < lay->second.fields.size(); ++i) { for (size_t i = 0; i < lay->second.fields.size(); ++i) {
inst.field_addr[lay->second.fields[i].name] = cur + i; inst.field_addr[lay->second.fields[i].name] = cur + i;
@@ -759,31 +637,18 @@ namespace {
any = true; any = true;
} }
} }
// 仅在有实例时把数据区补到实例区终点(无实例则保持全局区原样)
if (any) { if (any) {
data_.resize(static_cast<size_t>(cur) * 8, 0); data_.resize(static_cast<size_t>(cur) * 8, 0);
} }
return true; return true;
} }
/**
* @brief 按 POU + 实例名查实例字段表
* @param pou POU 名
* @param name 实例名
* @return 实例字段表指针;未找到返回 nullptr
*/
const InstFields* instance_of(const std::string& pou, const InstFields* instance_of(const std::string& pou,
const std::string& name) const { const std::string& name) const {
const auto it = instances_.find(pou + "/" + name); const auto it = instances_.find(pou + "/" + name);
return it == instances_.end() ? nullptr : &it->second; return it == instances_.end() ? nullptr : &it->second;
} }
/**
* @brief 查全局声明的初值(AST;gvl 文件顶层段,声明序 = 槽号序)
* @param name 全局名(小写)
* @param has_init 输出是否有初值
* @param init 输出初值
*/
void init_of(const std::string& name, bool* has_init, int64_t* init) const { void init_of(const std::string& name, bool* has_init, int64_t* init) const {
for (const SourceUnit& u : units_) { for (const SourceUnit& u : units_) {
for (const VarBlock& b : u.ast.globals) { for (const VarBlock& b : u.ast.globals) {
@@ -802,11 +667,13 @@ namespace {
/** /**
* @brief 取常量表 id(无则追加) * @brief 取常量表 id(无则追加)
* @param tag 类型标记(BOOL/INT/TIME * @param type_name 语言类型名("BOOL"/"INT"/"TIME")→ 配置 tag(契约
* @param value 常量值 * @param value 常量值
* @return const_idu16 * @return const_idu16
*/ */
uint16_t const_id(isa::types::TypeTag tag, int64_t value) { uint16_t const_id(const char* type_name, int64_t value) {
const ConfigType* ct = cfg_.find_type(type_name);
const uint32_t tag = ct ? ct->tag : 0;
for (size_t i = 0; i < consts_.size(); ++i) { for (size_t i = 0; i < consts_.size(); ++i) {
if (consts_[i].tag == tag && consts_[i].value == static_cast<uint64_t>(value)) { if (consts_[i].tag == tag && consts_[i].value == static_cast<uint64_t>(value)) {
return static_cast<uint16_t>(i); return static_cast<uint16_t>(i);
@@ -816,18 +683,12 @@ namespace {
return static_cast<uint16_t>(consts_.size() - 1); return static_cast<uint16_t>(consts_.size() - 1);
} }
/**
* @brief 拼映像:头 + 常量表 + 函数表 + 字节码 + 数据段
* @details 段序:const → funcs → code → fb(空)→ data
* 数据段放全局初值(layout_data 已生成字节)
* @return true 成功;falseerr 已写)
*/
bool assemble_image() { bool assemble_image() {
const uint32_t off_const = isa::kHeaderSize; const uint32_t off_const = static_cast<uint32_t>(kHeaderSize);
const uint32_t off_funcs = off_const + const uint32_t off_funcs = off_const +
static_cast<uint32_t>(consts_.size()) * isa::kConstEntrySize; static_cast<uint32_t>(consts_.size()) * static_cast<uint32_t>(kConstEntrySize);
uint32_t off_code = off_funcs + uint32_t off_code = off_funcs +
static_cast<uint32_t>(funcs_.size()) * isa::kFuncRowSize; static_cast<uint32_t>(funcs_.size()) * static_cast<uint32_t>(kFuncRowSize);
uint32_t code_total = 0; uint32_t code_total = 0;
for (const FuncCtx& f : funcs_) { for (const FuncCtx& f : funcs_) {
code_total += static_cast<uint32_t>(f.code.size()) * 4; code_total += static_cast<uint32_t>(f.code.size()) * 4;
@@ -837,12 +698,12 @@ namespace {
std::vector<uint8_t>& b = *image_; std::vector<uint8_t>& b = *image_;
b.assign(off_end, 0); b.assign(off_end, 0);
put_le32(b, 0, isa::kMagic); put_le32(b, 0, kMagic);
put_le32(b, 4, isa::kVersion); put_le32(b, 4, kVersion);
put_le32(b, 8, proj_.cycle_limit); put_le32(b, 8, proj_.cycle_limit);
put_le32(b, 12, proj_.dt_ms); put_le32(b, 12, proj_.dt_ms);
uint64_t hash = isa::kFnvBasis; uint64_t hash = kFnvBasis;
if (!compute_project_hash(proj_, &hash, err_)) { if (!compute_project_hash(proj_, &hash, err_)) {
return false; return false;
} }
@@ -855,80 +716,82 @@ namespace {
} }
} }
put_le32(b, 24, entry); put_le32(b, 24, entry);
put_le32(b, 28, static_cast<uint32_t>(link_.globals.size())); // n_globals put_le32(b, 28, static_cast<uint32_t>(link_.globals.size()));
put_le32(b, 32, 0); // n_i put_le32(b, 32, 0);
put_le32(b, 36, 0); // n_q put_le32(b, 36, 0);
put_le32(b, 40, 0); // n_m put_le32(b, 40, 0);
put_le32(b, 44, static_cast<uint32_t>(consts_.size())); put_le32(b, 44, static_cast<uint32_t>(consts_.size()));
put_le32(b, 48, static_cast<uint32_t>(funcs_.size())); put_le32(b, 48, static_cast<uint32_t>(funcs_.size()));
put_le32(b, 52, off_const); put_le32(b, 52, off_const);
put_le32(b, 56, off_funcs); put_le32(b, 56, off_funcs);
put_le32(b, 60, off_code); put_le32(b, 60, off_code);
put_le32(b, 64, off_data); // offset_fb(空,与数据段起点相同) put_le32(b, 64, off_data);
put_le32(b, 68, off_data); // offset_data put_le32(b, 68, off_data);
for (size_t i = 0; i < consts_.size(); ++i) { for (size_t i = 0; i < consts_.size(); ++i) {
const size_t o = off_const + i * isa::kConstEntrySize; const size_t o = off_const + i * kConstEntrySize;
put_le32(b, o, static_cast<uint32_t>(consts_[i].tag)); put_le32(b, o, consts_[i].tag);
put_le64(b, o + 4, consts_[i].value); put_le64(b, o + 4, consts_[i].value);
} }
uint32_t c = off_code; uint32_t c = off_code;
for (size_t i = 0; i < funcs_.size(); ++i) { for (size_t i = 0; i < funcs_.size(); ++i) {
const FuncCtx& f = funcs_[i]; const FuncCtx& f = funcs_[i];
const size_t o = off_funcs + i * isa::kFuncRowSize; const size_t o = off_funcs + i * kFuncRowSize;
put_le32(b, o, f.nregs); put_le32(b, o, f.nregs);
put_le32(b, o + 8, static_cast<uint32_t>(f.code.size())); put_le32(b, o + 8, static_cast<uint32_t>(f.code.size()));
for (const isa::Instr in : f.code) { for (const Instr in : f.code) {
put_le32(b, c, in); put_le32(b, c, in);
c += 4; c += 4;
} }
} }
// code_offset 回填(相对字节码段起点)
uint32_t acc = 0; uint32_t acc = 0;
for (size_t i = 0; i < funcs_.size(); ++i) { for (size_t i = 0; i < funcs_.size(); ++i) {
const size_t o = off_funcs + i * isa::kFuncRowSize; const size_t o = off_funcs + i * kFuncRowSize;
put_le32(b, o + 4, acc); put_le32(b, o + 4, acc);
acc += static_cast<uint32_t>(funcs_[i].code.size()) * 4; acc += static_cast<uint32_t>(funcs_[i].code.size()) * 4;
} }
// 数据段:全局初值字节
for (size_t i = 0; i < data_.size(); ++i) { for (size_t i = 0; i < data_.size(); ++i) {
b[off_data + i] = data_[i]; b[off_data + i] = data_[i];
} }
return true; return true;
} }
static void put_le32(std::vector<uint8_t>& b, size_t off, uint32_t v) {
b[off + 0] = static_cast<uint8_t>(v & 0xFFu);
b[off + 1] = static_cast<uint8_t>((v >> 8) & 0xFFu);
b[off + 2] = static_cast<uint8_t>((v >> 16) & 0xFFu);
b[off + 3] = static_cast<uint8_t>((v >> 24) & 0xFFu);
}
static void put_le64(std::vector<uint8_t>& b, size_t off, uint64_t v) {
for (int i = 0; i < 8; ++i) {
b[off + i] = static_cast<uint8_t>((v >> (8 * i)) & 0xFFu);
}
}
// ---- 成员 ---- // ---- 成员 ----
const Project& proj_; const Project& proj_;
const std::vector<SourceUnit>& units_; const std::vector<SourceUnit>& units_;
const LinkResult& link_; const LinkResult& link_;
const MachineConfig& cfg_;
std::vector<uint8_t>* image_; std::vector<uint8_t>* image_;
std::string* err_; std::string* err_;
std::vector<FuncCtx> funcs_; std::vector<FuncCtx> funcs_;
std::vector<isa::ConstEntry> consts_; std::vector<ConstEntry> consts_;
std::map<std::string, bool> io_input_; // io.input 绑定名(小写) std::map<std::string, bool> io_input_;
std::map<std::string, bool> io_output_; // io.output 绑定名(小写) std::map<std::string, bool> io_output_;
std::vector<uint8_t> data_; // 数据区(8 字节定宽槽:全局 + FB 实例块) std::vector<uint8_t> data_;
std::map<std::string, InstFields> instances_; // "POU/实例名" → 字段槽号 std::map<std::string, InstFields> instances_;
const InstFields* inline_fields_ = nullptr; // 内联 FB 体字段上下文(可空) const InstFields* inline_fields_ = nullptr;
}; };
} // namespace } // namespace
/**
* @brief 编译工程为 .stb 映像字节(对外入口)
* @param proj 工程定义
* @param units 全部源文件的 AST
* @param link 链接结果(在类型检查成功后调用)
* @param image 输出映像字节
* @param err 错误输出;可为 nullptr(静默)
* @return true 成功;false 失败(err 前缀 "codegen error"
*/
bool codegen_project(const Project& proj, const std::vector<SourceUnit>& units, bool codegen_project(const Project& proj, const std::vector<SourceUnit>& units,
const LinkResult& link, std::vector<uint8_t>* image, const LinkResult& link, const MachineConfig& cfg,
std::string* err) { std::vector<uint8_t>* image, std::string* err) {
Builder b(proj, units, link, image, err); Builder b(proj, units, link, cfg, image, err);
return b.run(); return b.run();
} }
+1 -1
View File
@@ -662,7 +662,7 @@ namespace {
* @return true 全部绑定合法;falseerr 已写) * @return true 全部绑定合法;falseerr 已写)
*/ */
bool check_io() { bool check_io() {
for (const isa::IoBinding& b : proj_.io) { for (const IoBinding& b : proj_.io) {
std::string key = b.var; std::string key = b.var;
for (char& ch : key) { for (char& ch : key) {
ch = static_cast<char>(std::tolower(static_cast<unsigned char>(ch))); ch = static_cast<char>(std::tolower(static_cast<unsigned char>(ch)));
+4 -4
View File
@@ -16,7 +16,7 @@
#include <string> #include <string>
#include <vector> #include <vector>
#include "isa/Image.h" #include "compiler/Stb.h"
namespace compiler { namespace compiler {
namespace { namespace {
@@ -241,7 +241,7 @@ namespace {
if (!check_keys(sec, is_input ? "io.input" : "io.output", allowed, 3, err)) { if (!check_keys(sec, is_input ? "io.input" : "io.output", allowed, 3, err)) {
return false; return false;
} }
isa::IoBinding b; IoBinding b;
b.is_input = is_input; b.is_input = is_input;
b.slot = 0; // 12.6 链接阶段再解析 b.slot = 0; // 12.6 链接阶段再解析
if (!req_string(sec, is_input ? "io.input" : "io.output", "var", &b.var, err)) { if (!req_string(sec, is_input ? "io.input" : "io.output", "var", &b.var, err)) {
@@ -353,7 +353,7 @@ bool compute_project_hash(const Project& p, uint64_t* hash, std::string* err) {
std::sort(sorted.begin(), sorted.end()); std::sort(sorted.begin(), sorted.end());
std::filesystem::path base(p.base_dir); std::filesystem::path base(p.base_dir);
uint64_t h = isa::kFnvBasis; uint64_t h = kFnvBasis;
for (const std::string& f : sorted) { for (const std::string& f : sorted) {
std::ifstream in(base / f, std::ios::binary); std::ifstream in(base / f, std::ios::binary);
if (!in) { if (!in) {
@@ -365,7 +365,7 @@ bool compute_project_hash(const Project& p, uint64_t* hash, std::string* err) {
in.read(buf, sizeof buf); in.read(buf, sizeof buf);
const std::streamsize n = in.gcount(); const std::streamsize n = in.gcount();
if (n > 0) { if (n > 0) {
h = isa::fnv1a64_update(h, reinterpret_cast<const uint8_t*>(buf), h = fnv1a64_update(h, reinterpret_cast<const uint8_t*>(buf),
static_cast<size_t>(n)); static_cast<size_t>(n));
} }
} }
+225
View File
@@ -0,0 +1,225 @@
/**
* @file Stb.cpp
* @brief 编译器自带的 .stb 映像规范(写侧)+ 只读视图 + FNV-1a + sidecar
* @author
* @date 2026-08-21
*/
#include "compiler/Stb.h"
#include <cstdio>
#include <cstring>
#include <fstream>
#include "compiler/Project.h"
namespace compiler {
uint64_t fnv1a64_update(uint64_t h, const uint8_t* data, size_t len) {
for (size_t i = 0; i < len; ++i) {
h ^= data[i];
h *= kFnvPrime;
}
return h;
}
uint64_t fnv1a64(const uint8_t* data, size_t len) {
return fnv1a64_update(kFnvBasis, data, len);
}
namespace {
uint32_t get_le32(const uint8_t* p) {
return static_cast<uint32_t>(p[0])
| (static_cast<uint32_t>(p[1]) << 8)
| (static_cast<uint32_t>(p[2]) << 16)
| (static_cast<uint32_t>(p[3]) << 24);
}
uint64_t get_le64(const uint8_t* p) {
uint64_t v = 0;
for (int i = 0; i < 8; ++i) {
v |= static_cast<uint64_t>(p[i]) << (8 * i);
}
return v;
}
} // namespace
StbView StbView::from(const uint8_t* buf, size_t len) {
StbView v;
v.buf_ = buf;
v.len_ = len;
if (buf == nullptr) {
v.err_ = "null buffer";
return v;
}
if (len < kHeaderSize) {
v.err_ = "image too short";
return v;
}
if (get_le32(buf + 0) != kMagic) {
v.err_ = "bad magic";
return v;
}
if (get_le32(buf + 4) != kVersion) {
v.err_ = "bad version";
return v;
}
v.cycle_limit_ = get_le32(buf + 8);
v.dt_ms_ = get_le32(buf + 12);
v.project_hash_ = get_le64(buf + 16);
v.entry_fn_id_ = get_le32(buf + 24);
v.n_globals_ = get_le32(buf + 28);
v.n_consts_ = get_le32(buf + 44);
v.n_funcs_ = get_le32(buf + 48);
v.offset_code_ = get_le32(buf + 60);
v.offset_fb_ = get_le32(buf + 64);
v.offset_data_ = get_le32(buf + 68);
// 段校验
const uint64_t offs[5] = {get_le32(buf + 52), get_le32(buf + 56), v.offset_code_,
v.offset_fb_, v.offset_data_};
for (int i = 0; i < 5; ++i) {
if (offs[i] < kHeaderSize || offs[i] > len) {
v.err_ = "segment offset out of range";
return v;
}
if (i > 0 && offs[i] < offs[i - 1]) {
v.err_ = "segment offsets not monotonic";
return v;
}
}
if (offs[1] - offs[0] != static_cast<uint64_t>(v.n_consts_) * kConstEntrySize) {
v.err_ = "const table size mismatch";
return v;
}
if (offs[2] - offs[1] != static_cast<uint64_t>(v.n_funcs_) * kFuncRowSize) {
v.err_ = "function table size mismatch";
return v;
}
if ((v.offset_fb_ - v.offset_code_) % 4 != 0) {
v.err_ = "code segment not 4-byte aligned";
return v;
}
if (v.entry_fn_id_ >= v.n_funcs_ && v.n_funcs_ != 0) {
v.err_ = "entry fn_id out of range";
return v;
}
v.ok_ = true;
v.err_.clear();
return v;
}
StbView StbView::from(const std::vector<uint8_t>& buf) {
return from(buf.data(), buf.size());
}
ConstEntry StbView::const_entry(size_t i) const {
ConstEntry e;
if (ok_ && i < n_consts_) {
const uint8_t* p = buf_ + offs_of_const() + i * kConstEntrySize;
e.tag = get_le32(p);
e.value = get_le64(p + 4);
}
return e;
}
uint32_t StbView::offs_of_const() const {
// offset_const = offs[0],由 from() 已校验的段起点
return static_cast<uint32_t>(get_le32(buf_ + 52));
}
StbView::FuncRow StbView::func_row(size_t i) const {
FuncRow r;
if (ok_ && i < n_funcs_) {
const uint8_t* p = buf_ + get_le32(buf_ + 56) + i * kFuncRowSize;
r.nregs = get_le32(p + 0);
r.code_offset = get_le32(p + 4);
r.code_len = get_le32(p + 8);
}
return r;
}
const uint8_t* StbView::code_bytes() const {
return ok_ ? buf_ + offset_code_ : nullptr;
}
size_t StbView::code_len() const {
return ok_ ? offset_fb_ - offset_code_ : 0;
}
const uint8_t* StbView::data_bytes() const {
return ok_ ? buf_ + offset_data_ : nullptr;
}
size_t StbView::data_len() const {
return ok_ ? len_ - offset_data_ : 0;
}
bool read_stb_file(const char* path, std::vector<uint8_t>* out, std::string* err) {
std::ifstream in(path, std::ios::binary);
if (!in) {
if (err) {
*err = "cannot open for read: " + std::string(path);
}
return false;
}
out->assign(std::istreambuf_iterator<char>(in), std::istreambuf_iterator<char>());
return true;
}
bool write_stb_file(const char* path, const std::vector<uint8_t>& img, std::string* err) {
std::FILE* f = std::fopen(path, "wb");
if (f == nullptr) {
if (err) {
*err = "cannot open for write: " + std::string(path);
}
return false;
}
const bool ok = img.empty() || std::fwrite(&img[0], 1, img.size(), f) == img.size();
std::fclose(f);
if (!ok && err) {
*err = "write failed: " + std::string(path);
}
return ok;
}
std::string make_sidecar(const std::vector<IoBinding>& bindings) {
std::string out;
for (const IoBinding& b : bindings) {
char buf[64];
out += b.is_input ? "[[io.input]]\n" : "[[io.output]]\n";
out += "var = \"";
out += b.var;
out += "\"\n";
std::snprintf(buf, sizeof buf, "slot = %u\n", static_cast<unsigned>(b.slot));
out += buf;
std::snprintf(buf, sizeof buf, "channel = %u\n", static_cast<unsigned>(b.channel));
out += buf;
std::snprintf(buf, sizeof buf, "bit = %u\n", static_cast<unsigned>(b.bit));
out += buf;
out += "\n";
}
return out;
}
bool write_sidecar_file(const char* path, const std::vector<IoBinding>& bindings,
std::string* err) {
std::FILE* f = std::fopen(path, "wb");
if (f == nullptr) {
if (err) {
*err = "cannot open for write: " + std::string(path);
}
return false;
}
const std::string s = make_sidecar(bindings);
const bool ok = s.empty() || std::fwrite(s.data(), 1, s.size(), f) == s.size();
std::fclose(f);
if (!ok && err) {
*err = "write failed: " + std::string(path);
}
return ok;
}
} // namespace compiler
+57 -35
View File
@@ -12,67 +12,66 @@
#include <string> #include <string>
#include <vector> #include <vector>
#include "compiler/Codec.h"
#include "compiler/Codegen.h" #include "compiler/Codegen.h"
#include "compiler/Linker.h" #include "compiler/Linker.h"
#include "compiler/MachineConfig.h"
#include "compiler/Project.h" #include "compiler/Project.h"
#include "compiler/Stb.h"
#include "compiler/Typecheck.h" #include "compiler/Typecheck.h"
#include "isa/Encode.h"
#include "isa/Image.h"
namespace { namespace {
void usage() { void usage() {
std::printf("usage: STCompiler <project.toml> [-o <name>.stb]\n" std::printf("usage: STCompiler <project.toml> [-o <name>.stb] --machine <machine.toml>\n"
" STCompiler <name>.stb --disasm\n" " STCompiler <name>.stb --disasm --machine <machine.toml>\n"
" 解析工程并编译(词法 → 语法 → 链接 → 类型 → 寄存器码)\n" " 解析工程并编译(词法 → 语法 → 链接 → 类型 → 寄存器码)\n"
" -o <name>.stb 编译并写出映像文件\n" " -o <name>.stb 编译并写出映像文件\n"
" 无 -o 只打印文件集合与工程哈希(12.3 阶段\n" " --machine <path> 机器定义(machine.toml,编译路径必填\n"
" --disasm 反汇编一个已编译的 .stb 映像\n" " --disasm 反汇编一个已编译的 .stb 映像(需要 --machine\n"
" --help 打印本帮助\n"); " --help 打印本帮助\n");
} }
// 反汇编一个已编译映像(函数表逐条 + 常量表 + 数据段摘要)。 // 反汇编一个已编译映像(函数表逐条 + 常量表 + 数据段摘要)。
// 映像无符号表,按 fn_id / 槽号显示;指令偏移为文件绝对字节偏移。 // 指令偏移为文件绝对字节偏移;反汇编按 machine.toml 的 format 输出
int dump_image(const char* path) { int dump_image(const char* path, const compiler::MachineConfig& cfg) {
std::vector<uint8_t> bytes; std::vector<uint8_t> bytes;
std::string err; std::string err;
if (!isa::read_stb_file(path, &bytes, &err)) { if (!compiler::read_stb_file(path, &bytes, &err)) {
std::fprintf(stderr, "error: %s\n", err.c_str()); std::fprintf(stderr, "error: %s\n", err.c_str());
return 1; return 1;
} }
const isa::ImageView v = isa::ImageView::from(bytes); const compiler::StbView v = compiler::StbView::from(bytes);
if (!v.ok()) { if (!v.ok()) {
std::fprintf(stderr, "error: %s\n", v.error().c_str()); std::fprintf(stderr, "error: %s\n", v.error().c_str());
return 1; return 1;
} }
const isa::ImageHeader& h = v.header();
std::printf("image: %s (%zu bytes, %u functions, %u globals, entry fn %u)\n", std::printf("image: %s (%zu bytes, %u functions, %u globals, entry fn %u)\n",
path, bytes.size(), h.n_funcs, h.n_globals, h.entry_fn_id); path, bytes.size(), v.n_funcs(), v.n_globals(), v.entry_fn_id());
std::printf(" dt_ms=%u cycle_limit=%u hash=0x%016llx\n", h.dt_ms, std::printf(" dt_ms=%u cycle_limit=%u hash=0x%016llx\n", v.dt_ms(),
h.cycle_limit, static_cast<unsigned long long>(h.project_hash)); v.cycle_limit(), static_cast<unsigned long long>(v.project_hash()));
if (h.n_consts) { if (v.n_consts()) {
std::printf("constants:\n"); std::printf("constants:\n");
for (uint32_t i = 0; i < h.n_consts; ++i) { for (uint32_t i = 0; i < v.n_consts(); ++i) {
const isa::ConstEntry c = v.const_entry(i); const compiler::ConstEntry c = v.const_entry(i);
const char* tag = c.tag == isa::types::Bool ? "BOOL" const char* tag = c.tag == 0 ? "BOOL" : c.tag == 1 ? "INT" : "TIME";
: c.tag == isa::types::Int ? "INT"
: "TIME";
std::printf(" [%u] %s %llu\n", i, tag, std::printf(" [%u] %s %llu\n", i, tag,
static_cast<unsigned long long>(c.value)); static_cast<unsigned long long>(c.value));
} }
} }
std::printf("functions:\n"); std::printf("functions:\n");
for (uint32_t i = 0; i < h.n_funcs; ++i) { for (uint32_t i = 0; i < v.n_funcs(); ++i) {
const isa::FuncRow r = v.func_row(i); const compiler::StbView::FuncRow r = v.func_row(i);
std::printf(" fn %u: nregs=%u, offset=%u, len=%u\n", i, r.nregs, std::printf(" fn %u: nregs=%u, offset=%u, len=%u\n", i, r.nregs,
r.code_offset, r.code_len); r.code_offset, r.code_len);
const uint8_t* base = v.code_bytes() + r.code_offset; const uint8_t* base = v.code_bytes() + r.code_offset;
for (uint32_t j = 0; j < r.code_len; ++j) { for (uint32_t j = 0; j < r.code_len; ++j) {
char buf[64]; char buf[64];
isa::disasm(reinterpret_cast<const uint32_t*>(base)[j], buf, sizeof buf); compiler::disasm(cfg, reinterpret_cast<const uint32_t*>(base)[j], buf,
sizeof buf);
std::printf(" 0x%04x %s\n", std::printf(" 0x%04x %s\n",
h.offset_code + r.code_offset + j * 4, buf); v.offset_code() + r.code_offset + j * 4, buf);
} }
} }
@@ -102,13 +101,37 @@ int main(int argc, char** argv) {
return 0; return 0;
} }
// --disasm:第一参数是 .stb 路径,不重新编译 // 参数收集
std::string machine_path;
bool disasm_mode = false;
for (int i = 2; i + 1 < argc; ++i) {
if (std::strcmp(argv[i], "--machine") == 0) {
machine_path = argv[i + 1];
}
}
for (int i = 1; i < argc; ++i) { for (int i = 1; i < argc; ++i) {
if (std::strcmp(argv[i], "--disasm") == 0) { if (std::strcmp(argv[i], "--disasm") == 0) {
return dump_image(argv[1]); disasm_mode = true;
} }
} }
// 机器定义(编译与反汇编都需要;加载/校验失败报错退出)
if (machine_path.empty()) {
std::fprintf(stderr, "error: missing --machine <machine.toml>\n");
usage();
return 1;
}
compiler::MachineConfig cfg;
std::string err;
if (!cfg.load(machine_path, &err)) {
std::fprintf(stderr, "error: %s\n", err.c_str());
return 1;
}
if (disasm_mode) {
return dump_image(argv[1], cfg);
}
const std::string toml_path = argv[1]; const std::string toml_path = argv[1];
std::string out_path; std::string out_path;
for (int i = 2; i + 1 < argc; ++i) { for (int i = 2; i + 1 < argc; ++i) {
@@ -118,7 +141,6 @@ int main(int argc, char** argv) {
} }
compiler::Project proj; compiler::Project proj;
std::string err;
if (!compiler::parse_project(toml_path, &proj, &err)) { if (!compiler::parse_project(toml_path, &proj, &err)) {
std::fprintf(stderr, "error: %s\n", err.c_str()); std::fprintf(stderr, "error: %s\n", err.c_str());
return 1; return 1;
@@ -160,19 +182,19 @@ int main(int argc, char** argv) {
return 1; return 1;
} }
std::vector<uint8_t> image; std::vector<uint8_t> image;
if (!compiler::codegen_project(proj, units, link, &image, &err)) { if (!compiler::codegen_project(proj, units, link, cfg, &image, &err)) {
std::fprintf(stderr, "error: %s\n", err.c_str()); std::fprintf(stderr, "error: %s\n", err.c_str());
return 1; return 1;
} }
if (!isa::write_stb_file(out_path.c_str(), image, &err)) { if (!compiler::write_stb_file(out_path.c_str(), image, &err)) {
std::fprintf(stderr, "error: %s\n", err.c_str()); std::fprintf(stderr, "error: %s\n", err.c_str());
return 1; return 1;
} }
// sidecarI/O 绑定 var → 槽号 → channel/bit(不进映像,执行器采样用) // sidecarI/O 绑定 var → 槽号 → channel/bit(不进映像,执行器采样用)
std::vector<isa::IoBinding> bindings; std::vector<compiler::IoBinding> bindings;
for (const isa::IoBinding& b : proj.io) { for (const compiler::IoBinding& b : proj.io) {
isa::IoBinding out_b = b; compiler::IoBinding out_b = b;
std::string key = b.var; std::string key = b.var;
for (char& ch : key) { for (char& ch : key) {
ch = static_cast<char>(std::tolower(static_cast<unsigned char>(ch))); ch = static_cast<char>(std::tolower(static_cast<unsigned char>(ch)));
@@ -185,13 +207,13 @@ int main(int argc, char** argv) {
} }
std::filesystem::path sidecar = std::filesystem::path(out_path); std::filesystem::path sidecar = std::filesystem::path(out_path);
sidecar.replace_extension(".runtime.toml"); sidecar.replace_extension(".runtime.toml");
if (!isa::write_sidecar_file(sidecar.string().c_str(), bindings, &err)) { if (!compiler::write_sidecar_file(sidecar.string().c_str(), bindings, &err)) {
std::fprintf(stderr, "error: %s\n", err.c_str()); std::fprintf(stderr, "error: %s\n", err.c_str());
return 1; return 1;
} }
const isa::ImageView v = isa::ImageView::from(image); const compiler::StbView v = compiler::StbView::from(image);
std::printf("compiled: %s (%zu bytes, %u functions, %u globals)\n", std::printf("compiled: %s (%zu bytes, %u functions, %u globals)\n",
out_path.c_str(), image.size(), v.header().n_funcs, v.header().n_globals); out_path.c_str(), image.size(), v.n_funcs(), v.n_globals());
return 0; return 0;
} }
+1 -1
View File
@@ -72,7 +72,7 @@ add_test(NAME typecheck_types
add_executable(codegen_test add_executable(codegen_test
./src/codegen_test.cpp) ./src/codegen_test.cpp)
target_link_libraries(codegen_test PRIVATE compiler) target_link_libraries(codegen_test PRIVATE compiler isa)
target_compile_definitions(codegen_test PRIVATE target_compile_definitions(codegen_test PRIVATE
REPO_ROOT="${CMAKE_SOURCE_DIR}") REPO_ROOT="${CMAKE_SOURCE_DIR}")
+6 -1
View File
@@ -119,8 +119,13 @@ namespace {
std::printf("FAIL %s type: %s\n", dir, err.c_str()); std::printf("FAIL %s type: %s\n", dir, err.c_str());
return false; return false;
} }
MachineConfig cfg;
if (!cfg.load(std::string(REPO_ROOT) + "/compiler/machine.toml", &err)) {
std::printf("FAIL %s machine load\n", dir);
return false;
}
std::vector<uint8_t> img; std::vector<uint8_t> img;
if (!codegen_project(p, units, link, &img, &err)) { if (!codegen_project(p, units, link, cfg, &img, &err)) {
if (want == Outcome::CompileError) { if (want == Outcome::CompileError) {
++g_checks; ++g_checks;
return true; return true;
+18 -3
View File
@@ -34,8 +34,19 @@ static int g_checks = 0;
} while (0) } while (0)
// 从用例目录走完整编译管线(解析 → 链接 → 类型检查 → 代码生成) // 从用例目录走完整编译管线(解析 → 链接 → 类型检查 → 代码生成)
static bool load_cfg(compiler::MachineConfig* cfg) {
std::string err;
if (!cfg->load(std::string(REPO_ROOT) + "/compiler/machine.toml", &err)) {
std::printf("FAIL load machine.toml: %s\n", err.c_str());
return false;
}
return true;
}
static bool compile_case(const char* dir, std::vector<uint8_t>* image, std::string* err) { static bool compile_case(const char* dir, std::vector<uint8_t>* image, std::string* err) {
using namespace compiler; using namespace compiler;
compiler::MachineConfig cfg;
if (!load_cfg(&cfg)) return false;
const std::string toml = std::string(REPO_ROOT) + "/tests/cases/" + dir + "/project.toml"; const std::string toml = std::string(REPO_ROOT) + "/tests/cases/" + dir + "/project.toml";
Project p; Project p;
if (!parse_project(toml, &p, err)) { if (!parse_project(toml, &p, err)) {
@@ -56,7 +67,7 @@ static bool compile_case(const char* dir, std::vector<uint8_t>* image, std::stri
if (!check_project(p, units, link, err)) { if (!check_project(p, units, link, err)) {
return false; return false;
} }
return codegen_project(p, units, link, image, err); return codegen_project(p, units, link, cfg, image, err);
} }
// 从临时工程(globals.st + main.st + 可含 io 绑定)走完整管线 // 从临时工程(globals.st + main.st + 可含 io 绑定)走完整管线
@@ -64,6 +75,8 @@ static bool compile_src(const char* name, const char* io_extra, const char* glob
const char* main_st, std::vector<uint8_t>* image, const char* main_st, std::vector<uint8_t>* image,
std::string* err) { std::string* err) {
using namespace compiler; using namespace compiler;
compiler::MachineConfig cfg;
if (!load_cfg(&cfg)) return false;
const std::string dir = std::string(REPO_ROOT) + "/build/cg_tmp_" + name; const std::string dir = std::string(REPO_ROOT) + "/build/cg_tmp_" + name;
std::filesystem::remove_all(dir); std::filesystem::remove_all(dir);
std::filesystem::create_directories(dir); std::filesystem::create_directories(dir);
@@ -100,7 +113,7 @@ static bool compile_src(const char* name, const char* io_extra, const char* glob
if (!check_project(p, units, link, err)) { if (!check_project(p, units, link, err)) {
return false; return false;
} }
return codegen_project(p, units, link, image, err); return codegen_project(p, units, link, cfg, image, err);
} }
// ---- 1. 用例 01:空 MAIN + RET ---- // ---- 1. 用例 01:空 MAIN + RET ----
@@ -653,7 +666,9 @@ static bool test_line1() {
LinkResult link; LinkResult link;
CHECK(link_project(p, units, &link, &err)); CHECK(link_project(p, units, &link, &err));
CHECK(check_project(p, units, link, &err)); CHECK(check_project(p, units, link, &err));
CHECK(codegen_project(p, units, link, &img, &err)); compiler::MachineConfig cfg;
CHECK(load_cfg(&cfg));
CHECK(codegen_project(p, units, link, cfg, &img, &err));
const isa::ImageView v = isa::ImageView::from(img); const isa::ImageView v = isa::ImageView::from(img);
CHECK(v.ok()); CHECK(v.ok());
+5 -1
View File
@@ -55,6 +55,10 @@ static void wslot(vm::Machine& m, int s, int64_t v) {
static bool make_machine(const char* dir, vm::Machine* m, std::string* err) { static bool make_machine(const char* dir, vm::Machine* m, std::string* err) {
using namespace compiler; using namespace compiler;
MachineConfig cfg;
if (!cfg.load(std::string(REPO_ROOT) + "/compiler/machine.toml", err)) {
return false;
}
const std::string toml = std::string(REPO_ROOT) + "/" + dir + "/project.toml"; const std::string toml = std::string(REPO_ROOT) + "/" + dir + "/project.toml";
Project p; Project p;
if (!parse_project(toml, &p, err)) { if (!parse_project(toml, &p, err)) {
@@ -76,7 +80,7 @@ static bool make_machine(const char* dir, vm::Machine* m, std::string* err) {
return false; return false;
} }
std::vector<uint8_t> img; std::vector<uint8_t> img;
if (!codegen_project(p, units, link, &img, err)) { if (!codegen_project(p, units, link, cfg, &img, err)) {
return false; return false;
} }
return vm::Machine::create(img, m, err); return vm::Machine::create(img, m, err);