diff options
Diffstat (limited to 'dev/LibCompiler/src/Backend')
| -rw-r--r-- | dev/LibCompiler/src/Backend/Assembler32x0.cc | 49 | ||||
| -rw-r--r-- | dev/LibCompiler/src/Backend/Assembler64x0.cc | 873 | ||||
| -rw-r--r-- | dev/LibCompiler/src/Backend/AssemblerAMD64.cc | 1203 | ||||
| -rw-r--r-- | dev/LibCompiler/src/Backend/AssemblerARM64.cc | 585 | ||||
| -rw-r--r-- | dev/LibCompiler/src/Backend/AssemblerPowerPC.cc | 909 |
5 files changed, 3619 insertions, 0 deletions
diff --git a/dev/LibCompiler/src/Backend/Assembler32x0.cc b/dev/LibCompiler/src/Backend/Assembler32x0.cc new file mode 100644 index 0000000..683b866 --- /dev/null +++ b/dev/LibCompiler/src/Backend/Assembler32x0.cc @@ -0,0 +1,49 @@ +/* ------------------------------------------- + + Copyright (C) 2024-2025 Amlal EL Mahrouss, all rights reserved + +------------------------------------------- */ + +/// bugs: 0 + +///////////////////////////////////////////////////////////////////////////////////////// + +// @file 32asm.cxx +// @author EL Mahrouss Amlal +// @brief 32x0 Assembler. + +// REMINDER: when dealing with an undefined symbol use (string +// size):LinkerFindSymbol:(string) so that ld will look for it. + +///////////////////////////////////////////////////////////////////////////////////////// + +#define __ASM_NEED_32x0__ 1 + +#include <LibCompiler/AE.h> +#include <LibCompiler/Backend/32x0.h> +#include <LibCompiler/Frontend.h> +#include <LibCompiler/PEF.h> + +///////////////////// + +// ANSI ESCAPE CODES + +///////////////////// + +#define kBlank "\e[0;30m" +#define kRed "\e[0;31m" +#define kWhite "\e[0;97m" +#define kYellow "\e[0;33m" + +#define kStdOut (std::cout << kWhite) +#define kStdErr (std::cout << kRed) + +///////////////////////////////////////////////////////////////////////////////////////// + +// @brief 32x0 Assembler entrypoint, the program/module starts here. + +///////////////////////////////////////////////////////////////////////////////////////// + +LIBCOMPILER_MODULE(NEAssemblerMain32000) { + return 0; +} diff --git a/dev/LibCompiler/src/Backend/Assembler64x0.cc b/dev/LibCompiler/src/Backend/Assembler64x0.cc new file mode 100644 index 0000000..fb96708 --- /dev/null +++ b/dev/LibCompiler/src/Backend/Assembler64x0.cc @@ -0,0 +1,873 @@ +/* ------------------------------------------- + + Copyright (C) 2024-2025 Amlal EL Mahrouss, all rights reserved + +------------------------------------------- */ + +/// bugs: 0 + +///////////////////////////////////////////////////////////////////////////////////////// + +// @file Assembler64x0.cxx +// @author EL Mahrouss Amlal +// @brief 64x000 Assembler. + +// REMINDER: when dealing with an undefined symbol use (string +// size):LinkerFindSymbol:(string) so that ld will look for it. + +///////////////////////////////////////////////////////////////////////////////////////// + +#define __ASM_NEED_64x0__ 1 + +#include <LibCompiler/AE.h> +#include <LibCompiler/Backend/64x0.h> +#include <LibCompiler/Frontend.h> +#include <LibCompiler/PEF.h> +#include <LibCompiler/Util/LCClUtils.h> +#include <algorithm> +#include <filesystem> +#include <fstream> +#include <iostream> +#include <vector> + +///////////////////// + +// ANSI ESCAPE CODES + +///////////////////// + +static char kOutputArch = LibCompiler::kPefArch64000; + +/// @note The 64x0 is VLSIW, so we need to jump to 4 bytes. +constexpr auto k64x0IPAlignment = 0x4U; + +static std::size_t kCounter = 1UL; + +static std::uintptr_t kOrigin = kPefBaseOrigin; +static std::vector<std::pair<std::string, std::uintptr_t>> kOriginLabel; + +static std::vector<e64k_num_t> kBytes; + +static LibCompiler::AERecordHeader kCurrentRecord{ + .fName = "", .fKind = LibCompiler::kPefCode, .fSize = 0, .fOffset = 0}; + +static std::vector<LibCompiler::AERecordHeader> kRecords; +static std::vector<std::string> kUndefinedSymbols; + +static const std::string kUndefinedSymbol = ":UndefinedSymbol:"; +static const std::string kRelocSymbol = ":RuntimeSymbol:"; + +// \brief forward decl. +static bool asm_read_attributes(std::string line); + +///////////////////////////////////////////////////////////////////////////////////////// + +// @brief 64x0 assembler entrypoint, the program/module starts here. + +///////////////////////////////////////////////////////////////////////////////////////// + +LIBCOMPILER_MODULE(AssemblerMain64x0) { + ::signal(SIGSEGV, Detail::drvi_crash_handler); + + for (size_t i = 1; i < argc; ++i) { + if (argv[i][0] == '-') { + if (strcmp(argv[i], "-64x0-ver") == 0 || strcmp(argv[i], "-64x0-v") == 0) { + kStdOut + << "Assembler64x0: 64x0 Assembler.\nAssembler64x0: v1.10\nAssembler64x0: Copyright (c) " + "Amlal El Mahrouss\n"; + return 0; + } else if (strcmp(argv[i], "-64x0-h") == 0) { + kStdOut << "Assembler64x0: 64x0 Assembler.\nAssembler64x0: Copyright (c) 2024 Mahrouss " + "Logic.\n"; + kStdOut << "--version: Print program version.\n"; + kStdOut << "--verbose: Print verbose output.\n"; + kStdOut << "--binary: Output as flat binary.\n"; + kStdOut << "--64xxx: Compile for a subset of the X64000.\n"; + + return 0; + } else if (strcmp(argv[i], "-64x0-binary") == 0) { + kOutputAsBinary = true; + continue; + } else if (strcmp(argv[i], "-64x0-verbose") == 0) { + kVerbose = true; + continue; + } + + kStdOut << "Assembler64x0: ignore " << argv[i] << "\n"; + continue; + } + + if (!std::filesystem::exists(argv[i])) { + kStdOut << "Assembler64x0: can't open: " << argv[i] << std::endl; + goto asm_fail_exit; + } + + std::string object_output(argv[i]); + + for (auto& ext : kAsmFileExts) { + if (object_output.find(ext) != std::string::npos) { + object_output.erase(object_output.find(ext), std::strlen(ext)); + } + } + + object_output += kOutputAsBinary ? kBinaryFileExt : kObjectFileExt; + + std::ifstream file_ptr(argv[i]); + std::ofstream file_ptr_out(object_output, std::ofstream::binary); + + if (file_ptr_out.bad()) { + if (kVerbose) { + kStdOut << "Assembler64x0: error: " << strerror(errno) << "\n"; + } + } + + std::string line; + + LibCompiler::AEHeader hdr{0}; + + memset(hdr.fPad, kAENullType, kAEPad); + + hdr.fMagic[0] = kAEMag0; + hdr.fMagic[1] = kAEMag1; + hdr.fSize = sizeof(LibCompiler::AEHeader); + hdr.fArch = kOutputArch; + + ///////////////////////////////////////////////////////////////////////////////////////// + + // COMPILATION LOOP + + ///////////////////////////////////////////////////////////////////////////////////////// + + LibCompiler::Encoder64x0 asm64; + + while (std::getline(file_ptr, line)) { + if (auto ln = asm64.CheckLine(line, argv[i]); !ln.empty()) { + Detail::print_error(ln, argv[i]); + continue; + } + + try { + asm_read_attributes(line); + asm64.WriteLine(line, argv[i]); + } catch (const std::exception& e) { + if (kVerbose) { + std::string what = e.what(); + Detail::print_warning("exit because of: " + what, "LibCompiler"); + } + + std::filesystem::remove(object_output); + goto asm_fail_exit; + } + } + + if (!kOutputAsBinary) { + if (kVerbose) { + kStdOut << "Assembler64x0: Writing object file...\n"; + } + + // this is the final step, write everything to the file. + + auto pos = file_ptr_out.tellp(); + + hdr.fCount = kRecords.size() + kUndefinedSymbols.size(); + + file_ptr_out << hdr; + + if (kRecords.empty()) { + kStdErr << "Assembler64x0: At least one record is needed to write an object " + "file.\nAssembler64x0: Make one using `public_segment .code64 foo_bar`.\n"; + + std::filesystem::remove(object_output); + return 1; + } + + kRecords[kRecords.size() - 1].fSize = kBytes.size(); + + std::size_t record_count = 0UL; + + for (auto& rec : kRecords) { + if (kVerbose) kStdOut << "Assembler64x0: Wrote record " << rec.fName << " to file...\n"; + + rec.fFlags |= LibCompiler::kKindRelocationAtRuntime; + rec.fOffset = record_count; + ++record_count; + + file_ptr_out << rec; + } + + // increment once again, so that we won't lie about the kUndefinedSymbols. + ++record_count; + + for (auto& sym : kUndefinedSymbols) { + LibCompiler::AERecordHeader _record_hdr{0}; + + if (kVerbose) kStdOut << "Assembler64x0: Wrote symbol " << sym << " to file...\n"; + + _record_hdr.fKind = kAENullType; + _record_hdr.fSize = sym.size(); + _record_hdr.fOffset = record_count; + + ++record_count; + + memset(_record_hdr.fPad, kAENullType, kAEPad); + memcpy(_record_hdr.fName, sym.c_str(), sym.size()); + + file_ptr_out << _record_hdr; + + ++kCounter; + } + + auto pos_end = file_ptr_out.tellp(); + + file_ptr_out.seekp(pos); + + hdr.fStartCode = pos_end; + hdr.fCodeSize = kBytes.size(); + + file_ptr_out << hdr; + + file_ptr_out.seekp(pos_end); + } else { + if (kVerbose) { + kStdOut << "Assembler64x0: Write raw binary...\n"; + } + } + + // byte from byte, we write this. + for (auto& byte : kBytes) { + file_ptr_out.write(reinterpret_cast<const char*>(&byte), sizeof(byte)); + } + + if (kVerbose) kStdOut << "Assembler64x0: Wrote file with program in it.\n"; + + file_ptr_out.flush(); + file_ptr_out.close(); + + if (kVerbose) kStdOut << "Assembler64x0: Exit succeeded.\n"; + + return 0; + } + +asm_fail_exit: + + if (kVerbose) kStdOut << "Assembler64x0: Exit failed.\n"; + + return 1; +} + +///////////////////////////////////////////////////////////////////////////////////////// + +// @brief Check for attributes +// returns true if any was found. + +///////////////////////////////////////////////////////////////////////////////////////// + +static bool asm_read_attributes(std::string line) { + // extern_segment is the opposite of public_segment, it signals to the ld + // that we need this symbol. + if (LibCompiler::find_word(line, "extern_segment")) { + if (kOutputAsBinary) { + Detail::print_error("Invalid extern_segment directive in flat binary mode.", "LibCompiler"); + throw std::runtime_error("invalid_extern_segment_bin"); + } + + auto name = line.substr(line.find("extern_segment") + strlen("extern_segment")); + + /// sanity check to avoid stupid linker errors. + if (name.size() == 0) { + Detail::print_error("Invalid extern_segment", "power-as"); + throw std::runtime_error("invalid_extern_segment"); + } + + std::string result = std::to_string(name.size()); + result += kUndefinedSymbol; + + // mangle this + for (char& j : name) { + if (j == ' ' || j == ',') j = '$'; + } + + result += name; + + if (name.find(".code64") != std::string::npos) { + // data is treated as code. + kCurrentRecord.fKind = LibCompiler::kPefCode; + } else if (name.find(".data64") != std::string::npos) { + // no code will be executed from here. + kCurrentRecord.fKind = LibCompiler::kPefData; + } else if (name.find(".zero64") != std::string::npos) { + // this is a bss section. + kCurrentRecord.fKind = LibCompiler::kPefZero; + } + + // this is a special case for the start stub. + // we want this so that ld can find it. + + if (name == kPefStart) { + kCurrentRecord.fKind = LibCompiler::kPefCode; + } + + // now we can tell the code size of the previous kCurrentRecord. + + if (!kRecords.empty()) kRecords[kRecords.size() - 1].fSize = kBytes.size(); + + memset(kCurrentRecord.fName, 0, kAESymbolLen); + memcpy(kCurrentRecord.fName, result.c_str(), result.size()); + + ++kCounter; + + memset(kCurrentRecord.fPad, kAENullType, kAEPad); + + kRecords.emplace_back(kCurrentRecord); + + return true; + } + // public_segment is a special keyword used by Assembler64x0 to tell the AE output stage to + // mark this section as a header. it currently supports .code64, .data64., + // .zero64 + else if (LibCompiler::find_word(line, "public_segment")) { + if (kOutputAsBinary) { + Detail::print_error("Invalid public_segment directive in flat binary mode.", "LibCompiler"); + throw std::runtime_error("invalid_public_segment_bin"); + } + + auto name = line.substr(line.find("public_segment") + strlen("public_segment")); + + std::string name_copy = name; + + for (char& j : name) { + if (j == ' ') j = '$'; + } + + if (name.find(".code64") != std::string::npos) { + // data is treated as code. + + name_copy.erase(name_copy.find(".code64"), strlen(".code64")); + kCurrentRecord.fKind = LibCompiler::kPefCode; + } else if (name.find(".data64") != std::string::npos) { + // no code will be executed from here. + + name_copy.erase(name_copy.find(".data64"), strlen(".data64")); + kCurrentRecord.fKind = LibCompiler::kPefData; + } else if (name.find(".zero64") != std::string::npos) { + // this is a bss section. + + name_copy.erase(name_copy.find(".zero64"), strlen(".zero64")); + kCurrentRecord.fKind = LibCompiler::kPefZero; + } + + // this is a special case for the start stub. + // we want this so that ld can find it. + + if (name == kPefStart) { + kCurrentRecord.fKind = LibCompiler::kPefCode; + } + + while (name_copy.find(" ") != std::string::npos) name_copy.erase(name_copy.find(" "), 1); + + kOriginLabel.push_back(std::make_pair(name_copy, kOrigin)); + ++kOrigin; + + // now we can tell the code size of the previous kCurrentRecord. + + if (!kRecords.empty()) kRecords[kRecords.size() - 1].fSize = kBytes.size(); + + memset(kCurrentRecord.fName, 0, kAESymbolLen); + memcpy(kCurrentRecord.fName, name.c_str(), name.size()); + + ++kCounter; + + memset(kCurrentRecord.fPad, kAENullType, kAEPad); + + kRecords.emplace_back(kCurrentRecord); + + return true; + } + + return false; +} + +// \brief algorithms and helpers. + +namespace Detail::algorithm { +// \brief authorize a brief set of characters. +static inline bool is_not_alnum_space(char c) { + return !(isalpha(c) || isdigit(c) || (c == ' ') || (c == '\t') || (c == ',') || (c == '(') || + (c == ')') || (c == '"') || (c == '\'') || (c == '[') || (c == ']') || (c == '+') || + (c == '_') || (c == ':') || (c == '@') || (c == '.')); +} + +bool is_valid_64x0(std::string str) { + return std::find_if(str.begin(), str.end(), is_not_alnum_space) == str.end(); +} +} // namespace Detail::algorithm + +///////////////////////////////////////////////////////////////////////////////////////// + +// @brief Check for line (syntax check) + +///////////////////////////////////////////////////////////////////////////////////////// + +std::string LibCompiler::Encoder64x0::CheckLine(std::string line, std::string file) { + std::string err_str; + + if (line.empty() || LibCompiler::find_word(line, "extern_segment") || + LibCompiler::find_word(line, "public_segment") || line.find('#') != std::string::npos || + LibCompiler::find_word(line, ";")) { + if (line.find('#') != std::string::npos) { + line.erase(line.find('#')); + } else if (line.find(';') != std::string::npos) { + line.erase(line.find(';')); + } else { + // now check the line for validity + if (!Detail::algorithm::is_valid_64x0(line)) { + err_str = "Line contains non alphanumeric characters.\nhere -> "; + err_str += line; + } + } + + return err_str; + } + + if (!Detail::algorithm::is_valid_64x0(line)) { + err_str = "Line contains non alphanumeric characters.\nhere -> "; + err_str += line; + + return err_str; + } + + // check for a valid instruction format. + + if (line.find(',') != std::string::npos) { + if (line.find(',') + 1 == line.size()) { + err_str += "\nInstruction lacks right register, here -> "; + err_str += line.substr(line.find(',')); + + return err_str; + } else { + bool nothing_on_right = true; + + if (line.find(',') + 1 > line.size()) { + err_str += "\nInstruction not complete, here -> "; + err_str += line; + + return err_str; + } + + auto substr = line.substr(line.find(',') + 1); + + for (auto& ch : substr) { + if (ch != ' ' && ch != '\t') { + nothing_on_right = false; + } + } + + // this means we found nothing after that ',' . + if (nothing_on_right) { + err_str += "\nInstruction not complete, here -> "; + err_str += line; + + return err_str; + } + } + } + + // these do take an argument. + std::vector<std::string> operands_inst = {"stw", "ldw", "lda", "sta"}; + + // these don't. + std::vector<std::string> filter_inst = {"jlr", "jrl", "int"}; + + for (auto& opcode64x0 : kOpcodes64x0) { + if (line.find(opcode64x0.fName) != std::string::npos) { + if (opcode64x0.fFunct7 == kAsmNoArgs) return err_str; + + for (auto& op : operands_inst) { + // if only the instruction was found. + if (line == op) { + err_str += "\nMalformed "; + err_str += op; + err_str += " instruction, here -> "; + err_str += line; + } + } + + // if it is like that -> addr1, 0x0 + if (auto it = std::find(filter_inst.begin(), filter_inst.end(), opcode64x0.fName); + it == filter_inst.cend()) { + if (LibCompiler::find_word(line, opcode64x0.fName)) { + if (!isspace(line[line.find(opcode64x0.fName) + strlen(opcode64x0.fName)])) { + err_str += "\nMissing space between "; + err_str += opcode64x0.fName; + err_str += " and operands.\nhere -> "; + err_str += line; + } + } + } + + return err_str; + } + } + + err_str += "Unrecognized instruction: " + line; + + return err_str; +} + +bool LibCompiler::Encoder64x0::WriteNumber(const std::size_t& pos, std::string& jump_label) { + if (!isdigit(jump_label[pos])) return false; + + switch (jump_label[pos + 1]) { + case 'x': { + if (auto res = strtol(jump_label.substr(pos + 2).c_str(), nullptr, 16); !res) { + if (errno != 0) { + Detail::print_error("invalid hex number: " + jump_label, "LibCompiler"); + throw std::runtime_error("invalid_hex_number"); + } + } + + LibCompiler::NumberCast64 num(strtol(jump_label.substr(pos + 2).c_str(), nullptr, 16)); + + for (char& i : num.number) { + kBytes.push_back(i); + } + + if (kVerbose) { + kStdOut << "Assembler64x0: found a base 16 number here: " << jump_label.substr(pos) << "\n"; + } + + return true; + } + case 'b': { + if (auto res = strtol(jump_label.substr(pos + 2).c_str(), nullptr, 2); !res) { + if (errno != 0) { + Detail::print_error("invalid binary number: " + jump_label, "LibCompiler"); + throw std::runtime_error("invalid_bin"); + } + } + + LibCompiler::NumberCast64 num(strtol(jump_label.substr(pos + 2).c_str(), nullptr, 2)); + + if (kVerbose) { + kStdOut << "Assembler64x0: found a base 2 number here: " << jump_label.substr(pos) << "\n"; + } + + for (char& i : num.number) { + kBytes.push_back(i); + } + + return true; + } + case 'o': { + if (auto res = strtol(jump_label.substr(pos + 2).c_str(), nullptr, 7); !res) { + if (errno != 0) { + Detail::print_error("invalid octal number: " + jump_label, "LibCompiler"); + throw std::runtime_error("invalid_octal"); + } + } + + LibCompiler::NumberCast64 num(strtol(jump_label.substr(pos + 2).c_str(), nullptr, 7)); + + if (kVerbose) { + kStdOut << "Assembler64x0: found a base 8 number here: " << jump_label.substr(pos) << "\n"; + } + + for (char& i : num.number) { + kBytes.push_back(i); + } + + return true; + } + default: { + break; + } + } + + /* check for errno and stuff like that */ + if (auto res = strtol(jump_label.substr(pos).c_str(), nullptr, 10); !res) { + if (errno != 0) { + return false; + } + } + + LibCompiler::NumberCast64 num(strtol(jump_label.substr(pos).c_str(), nullptr, 10)); + + for (char& i : num.number) { + kBytes.push_back(i); + } + + if (kVerbose) { + kStdOut << "Assembler64x0: found a base 10 number here: " << jump_label.substr(pos) << "\n"; + } + + return true; +} + +///////////////////////////////////////////////////////////////////////////////////////// + +// @brief Read and write an instruction to the output array. + +///////////////////////////////////////////////////////////////////////////////////////// + +bool LibCompiler::Encoder64x0::WriteLine(std::string line, std::string file) { + if (LibCompiler::find_word(line, "public_segment ")) return true; + + for (auto& opcode64x0 : kOpcodes64x0) { + // strict check here + if (LibCompiler::find_word(line, opcode64x0.fName) && Detail::algorithm::is_valid_64x0(line)) { + std::string name(opcode64x0.fName); + std::string jump_label, cpy_jump_label; + + kBytes.emplace_back(opcode64x0.fOpcode); + kBytes.emplace_back(opcode64x0.fFunct3); + kBytes.emplace_back(opcode64x0.fFunct7); + + // check funct7 type. + switch (opcode64x0.fFunct7) { + // reg to reg means register to register transfer operation. + case kAsmRegToReg: + case kAsmImmediate: { + // \brief how many registers we found. + std::size_t found_some = 0UL; + + for (size_t line_index = 0UL; line_index < line.size(); line_index++) { + if (line[line_index] == kAsmRegisterPrefix[0] && isdigit(line[line_index + 1])) { + std::string register_syntax = kAsmRegisterPrefix; + register_syntax += line[line_index + 1]; + + if (isdigit(line[line_index + 2])) register_syntax += line[line_index + 2]; + + std::string reg_str; + reg_str += line[line_index + 1]; + + if (isdigit(line[line_index + 2])) reg_str += line[line_index + 2]; + + // it ranges from r0 to r19 + // something like r190 doesn't exist in the instruction set. + if (kOutputArch == LibCompiler::kPefArch64000) { + if (isdigit(line[line_index + 3]) && isdigit(line[line_index + 2])) { + reg_str += line[line_index + 3]; + Detail::print_error("invalid register index, r" + reg_str + + "\nnote: The 64x0 accepts registers from r0 to r20.", + file); + throw std::runtime_error("invalid_register_index"); + } + } + + // finally cast to a size_t + std::size_t reg_index = strtol(reg_str.c_str(), nullptr, 10); + + if (reg_index > kAsmRegisterLimit) { + Detail::print_error("invalid register index, r" + reg_str, file); + throw std::runtime_error("invalid_register_index"); + } + + kBytes.emplace_back(reg_index); + ++found_some; + + if (kVerbose) { + kStdOut << "Assembler64x0: Register found: " << register_syntax << "\n"; + kStdOut << "Assembler64x0: Register amount in instruction: " << found_some << "\n"; + } + } + } + + // we're not in immediate addressing, reg to reg. + if (opcode64x0.fFunct7 != kAsmImmediate) { + // remember! register to register! + if (found_some == 1) { + Detail::print_error( + "Too few registers.\ntip: each Assembler64x0 register " + "starts with 'r'.\nline: " + + line, + file); + throw std::runtime_error("not_a_register"); + } + } + + if (found_some < 1 && name != "ldw" && name != "lda" && name != "stw") { + Detail::print_error("invalid combination of opcode and registers.\nline: " + line, + file); + throw std::runtime_error("invalid_comb_op_reg"); + } else if (found_some == 1 && name == "add") { + Detail::print_error("invalid combination of opcode and registers.\nline: " + line, + file); + throw std::runtime_error("invalid_comb_op_reg"); + } else if (found_some == 1 && name == "sub") { + Detail::print_error("invalid combination of opcode and registers.\nline: " + line, + file); + throw std::runtime_error("invalid_comb_op_reg"); + } + + if (found_some > 0 && name == "pop") { + Detail::print_error( + "invalid combination for opcode 'pop'.\ntip: it expects " + "nothing.\nline: " + + line, + file); + throw std::runtime_error("invalid_comb_op_pop"); + } + } + default: + break; + } + + // try to fetch a number from the name + if (name == "stw" || name == "ldw" || name == "lda" || name == "sta") { + auto where_string = name; + + // if we load something, we'd need it's symbol/literal + if (name == "stw" || name == "sta" || name == "ldw" || name == "lda" || name == "sta") + where_string = ","; + + jump_label = line; + + auto found_sym = false; + + while (jump_label.find(where_string) != std::string::npos) { + jump_label = jump_label.substr(jump_label.find(where_string) + where_string.size()); + + while (jump_label.find(" ") != std::string::npos) { + jump_label.erase(jump_label.find(" "), 1); + } + + if (jump_label[0] != kAsmRegisterPrefix[0] && !isdigit(jump_label[1])) { + if (found_sym) { + Detail::print_error( + "invalid combination of opcode and operands.\nhere -> " + jump_label, file); + throw std::runtime_error("invalid_comb_op_ops"); + } else { + // death trap installed. + found_sym = true; + } + } + } + + cpy_jump_label = jump_label; + + // replace any spaces with $ + if (jump_label[0] == ' ') { + while (jump_label.find(' ') != std::string::npos) { + if (isalnum(jump_label[0]) || isdigit(jump_label[0])) break; + + jump_label.erase(jump_label.find(' '), 1); + } + } + + if (!this->WriteNumber(0, jump_label)) { + // sta expects this: sta 0x000000, r0 + if (name == "sta") { + Detail::print_error("invalid combination of opcode and operands.\nHere ->" + line, + file); + throw std::runtime_error("invalid_comb_op_ops"); + } + } else { + if (name == "sta" && cpy_jump_label.find("extern_segment ") != std::string::npos) { + Detail::print_error("invalid usage extern_segment on 'sta', here: " + line, file); + throw std::runtime_error("invalid_sta_usage"); + } + } + + goto asm_write_label; + } + + // This is the case where we jump to a label, it is also used as a goto. + if (name == "lda" || name == "sta") { + asm_write_label: + if (cpy_jump_label.find('\n') != std::string::npos) + cpy_jump_label.erase(cpy_jump_label.find('\n'), 1); + + if (cpy_jump_label.find("extern_segment") != std::string::npos) { + cpy_jump_label.erase(cpy_jump_label.find("extern_segment"), strlen("extern_segment")); + + if (name == "sta") { + Detail::print_error("extern_segment is not allowed on a sta operation.", file); + throw std::runtime_error("extern_segment_sta_op"); + } else { + goto asm_end_label_cpy; + } + } + + if (name == "lda" || name == "sta") { + for (auto& label : kOriginLabel) { + if (cpy_jump_label == label.first) { + if (kVerbose) { + kStdOut << "Assembler64x0: Replace label " << cpy_jump_label + << " to address: " << label.second << std::endl; + } + + LibCompiler::NumberCast64 num(label.second); + + for (auto& num : num.number) { + kBytes.push_back(num); + } + + goto asm_end_label_cpy; + } + } + + if (cpy_jump_label[0] == '0') { + switch (cpy_jump_label[1]) { + case 'x': + case 'o': + case 'b': + if (this->WriteNumber(0, cpy_jump_label)) goto asm_end_label_cpy; + + break; + default: + break; + } + + if (isdigit(cpy_jump_label[0])) { + if (this->WriteNumber(0, cpy_jump_label)) goto asm_end_label_cpy; + + break; + } + } + } + + if (cpy_jump_label.size() < 1) { + Detail::print_error("label is empty, can't jump on it.", file); + throw std::runtime_error("label_empty"); + } + + /// don't go any further if: + /// load word (ldw) or store word. (stw) + + if (name == "ldw" || name == "stw") break; + + auto mld_reloc_str = std::to_string(cpy_jump_label.size()); + mld_reloc_str += kUndefinedSymbol; + mld_reloc_str += cpy_jump_label; + + bool ignore_back_slash = false; + + for (auto& reloc_chr : mld_reloc_str) { + if (reloc_chr == '\\') { + ignore_back_slash = true; + continue; + } + + if (ignore_back_slash) { + ignore_back_slash = false; + continue; + } + + kBytes.push_back(reloc_chr); + } + + kBytes.push_back('\0'); + goto asm_end_label_cpy; + } + + asm_end_label_cpy: + kOrigin += k64x0IPAlignment; + + break; + } + } + + return true; +} + +// Last rev 13-1-24 diff --git a/dev/LibCompiler/src/Backend/AssemblerAMD64.cc b/dev/LibCompiler/src/Backend/AssemblerAMD64.cc new file mode 100644 index 0000000..e9f3ad5 --- /dev/null +++ b/dev/LibCompiler/src/Backend/AssemblerAMD64.cc @@ -0,0 +1,1203 @@ +/* ------------------------------------------- + + Copyright (C) 2024-2025 Amlal EL Mahrouss, all rights reserved + +------------------------------------------- */ + +///////////////////////////////////////////////////////////////////////////////////////// + +/// @file AssemblerAMD64.cc +/// @author EL Mahrouss Amlal +/// @brief AMD64 Assembler. +/// REMINDER: when dealing with an undefined symbol use (string +/// size):LinkerFindSymbol:(string) so that ld will look for it. + +///////////////////////////////////////////////////////////////////////////////////////// + +/// bugs: 0 + +/// feature request: 1 +/// Encode registers in mov, add, xor... + +///////////////////////////////////////////////////////////////////////////////////////// + +#define __ASM_NEED_AMD64__ 1 + +#define kAssemblerPragmaSymStr "#" +#define kAssemblerPragmaSym '#' + +#include <LibCompiler/AE.h> +#include <LibCompiler/Backend/X64.h> +#include <LibCompiler/Frontend.h> +#include <LibCompiler/PEF.h> +#include <algorithm> +#include <cstdlib> +#include <filesystem> +#include <fstream> +#include <iostream> +#include <vector> + +///////////////////// + +// ANSI ESCAPE CODES + +///////////////////// + +#define kBlank "\e[0;30m" +#define kRed "\e[0;31m" +#define kWhite "\e[0;97m" +#define kYellow "\e[0;33m" + +static char kOutputArch = LibCompiler::kPefArchAMD64; + +constexpr auto kIPAlignement = 0x1U; + +static std::size_t kCounter = 1UL; + +static std::uintptr_t kOrigin = kPefBaseOrigin; +static std::vector<std::pair<std::string, std::uintptr_t>> kOriginLabel; + +/// @brief keep it simple by default. +static std::int32_t kRegisterBitWidth = 16U; + +static std::vector<i64_byte_t> kAppBytes; + +static LibCompiler::AERecordHeader kCurrentRecord{ + .fName = "", .fKind = LibCompiler::kPefCode, .fSize = 0, .fOffset = 0}; + +static std::vector<LibCompiler::AERecordHeader> kRecords; +static std::vector<std::string> kDefinedSymbols; +static std::vector<std::string> kUndefinedSymbols; + +static const std::string kUndefinedSymbol = ":UndefinedSymbol:"; + +// \brief forward decl. +static bool asm_read_attributes(std::string line); + +#include <LibCompiler/Util/LCAsmUtils.h> + +///////////////////////////////////////////////////////////////////////////////////////// + +// @brief AMD64 assembler entrypoint, the program/module starts here. + +///////////////////////////////////////////////////////////////////////////////////////// + +LIBCOMPILER_MODULE(AssemblerMainAMD64) { + //////////////// CPU OPCODES BEGIN //////////////// + + ::signal(SIGSEGV, Detail::drvi_crash_handler); + + std::string opcodes_jump[kJumpLimit] = {"ja", "jae", "jb", "jbe", "jc", "je", "jg", "jge", + "jl", "jle", "jna", "jnae", "jnb", "jnbe", "jnc", "jne", + "jng", "jnge", "jnl", "jnle", "jno", "jnp", "jns", "jnz", + "jo", "jp", "jpe", "jpo", "js", "jz"}; + + for (i64_hword_t i = 0; i < kJumpLimit; i++) { + CpuOpcodeAMD64 code{.fName = opcodes_jump[i], + .fOpcode = static_cast<i64_hword_t>(kAsmJumpOpcode + i)}; + kOpcodesAMD64.push_back(code); + } + + CpuOpcodeAMD64 code{.fName = "jcxz", .fOpcode = 0xE3}; + kOpcodesAMD64.push_back(code); + + for (i64_hword_t i = kJumpLimitStandard; i < kJumpLimitStandardLimit; i++) { + CpuOpcodeAMD64 code{.fName = "jmp", .fOpcode = i}; + kOpcodesAMD64.push_back(code); + } + + CpuOpcodeAMD64 lahf{.fName = "lahf", .fOpcode = 0x9F}; + kOpcodesAMD64.push_back(lahf); + + CpuOpcodeAMD64 lds{.fName = "lds", .fOpcode = 0xC5}; + kOpcodesAMD64.push_back(lds); + + CpuOpcodeAMD64 lea{.fName = "lea", .fOpcode = 0x8D}; + kOpcodesAMD64.push_back(lea); + + CpuOpcodeAMD64 nop{.fName = "nop", .fOpcode = 0x90}; + kOpcodesAMD64.push_back(nop); + + //////////////// CPU OPCODES END //////////////// + + for (size_t i = 1; i < argc; ++i) { + if (argv[i][0] == '-') { + if (strcmp(argv[i], "--amd64:ver") == 0 || strcmp(argv[i], "--amd64:v") == 0) { + kStdOut << "AssemblerAMD64: AMD64 Assembler Driver.\nAssemblerAMD64: " + "v1.10\nAssemblerAMD64: Copyright " + "(c) Amlal El Mahrouss\n"; + return 0; + } else if (strcmp(argv[i], "--amd64:h") == 0) { + kStdOut << "AssemblerAMD64: AMD64 Assembler Driver.\nAssemblerAMD64: Copyright (c) 2024 " + "Amlal El Mahrouss\n"; + kStdOut << "--version: Print program version.\n"; + kStdOut << "--verbose: Print verbose output.\n"; + kStdOut << "--binary: Output as flat binary.\n"; + + return 0; + } else if (strcmp(argv[i], "--amd64:binary") == 0) { + kOutputAsBinary = true; + continue; + } else if (strcmp(argv[i], "--amd64:verbose") == 0) { + kVerbose = true; + continue; + } + + kStdOut << "AssemblerAMD64: ignore " << argv[i] << "\n"; + continue; + } + + if (!std::filesystem::exists(argv[i])) { + kStdOut << "AssemblerAMD64: can't open: " << argv[i] << std::endl; + goto asm_fail_exit; + } + + std::string object_output(argv[i]); + std::string asm_input(argv[i]); + + for (auto& ext : kAsmFileExts) { + if (object_output.ends_with(ext)) { + object_output.erase(object_output.find(ext), std::strlen(ext)); + break; + } + } + + object_output += kOutputAsBinary ? kBinaryFileExt : kObjectFileExt; + + std::ifstream file_ptr(argv[i]); + std::ofstream file_ptr_out(object_output, std::ofstream::binary); + + kStdOut << "AssemblerAMD64: Assembling: " << argv[i] << "\n"; + + if (file_ptr_out.bad()) { + if (kVerbose) { + kStdOut << "AssemblerAMD64: error: " << strerror(errno) << "\n"; + } + + return 1; + } + + std::string line; + + LibCompiler::AEHeader hdr{0}; + + memset(hdr.fPad, kAENullType, kAEPad); + + hdr.fMagic[0] = kAEMag0; + hdr.fMagic[1] = kAEMag1; + hdr.fSize = sizeof(LibCompiler::AEHeader); + hdr.fArch = kOutputArch; + + ///////////////////////////////////////////////////////////////////////////////////////// + + // COMPILATION LOOP + + ///////////////////////////////////////////////////////////////////////////////////////// + + LibCompiler::EncoderAMD64 asm64; + + if (kVerbose) { + kStdOut << "Compiling: " + asm_input << "\n"; + kStdOut << "From: " + line << "\n"; + } + + while (std::getline(file_ptr, line)) { + if (auto ln = asm64.CheckLine(line, argv[i]); !ln.empty()) { + Detail::print_error(ln, argv[i]); + continue; + } + + try { + asm_read_attributes(line); + asm64.WriteLine(line, argv[i]); + } catch (const std::exception& e) { + if (kVerbose) { + std::string what = e.what(); + Detail::print_warning("exit because of: " + what, "LibCompiler"); + } + + try { + std::filesystem::remove(object_output); + } catch (...) { + } + + goto asm_fail_exit; + } + } + + if (!kOutputAsBinary) { + if (kVerbose) { + kStdOut << "AssemblerAMD64: Writing object file...\n"; + } + + // this is the final step, write everything to the file. + + auto pos = file_ptr_out.tellp(); + + hdr.fCount = kRecords.size() + kUndefinedSymbols.size(); + + file_ptr_out << hdr; + + if (kRecords.empty()) { + kStdErr << "AssemblerAMD64: At least one record is needed to write an object " + "file.\nAssemblerAMD64: Make one using `public_segment .code64 foo_bar`.\n"; + + std::filesystem::remove(object_output); + return 1; + } + + kRecords[kRecords.size() - 1].fSize = kAppBytes.size(); + + std::size_t record_count = 0UL; + + for (auto& rec : kRecords) { + if (kVerbose) kStdOut << "AssemblerAMD64: Wrote record " << rec.fName << " to file...\n"; + + rec.fFlags |= LibCompiler::kKindRelocationAtRuntime; + rec.fOffset = record_count; + ++record_count; + + file_ptr_out << rec; + } + + // increment once again, so that we won't lie about the kUndefinedSymbols. + ++record_count; + + for (auto& sym : kUndefinedSymbols) { + LibCompiler::AERecordHeader _record_hdr{0}; + + if (kVerbose) kStdOut << "AssemblerAMD64: Wrote symbol " << sym << " to file...\n"; + + _record_hdr.fKind = kAENullType; + _record_hdr.fSize = sym.size(); + _record_hdr.fOffset = record_count; + + ++record_count; + + memset(_record_hdr.fPad, kAENullType, kAEPad); + memcpy(_record_hdr.fName, sym.c_str(), sym.size()); + + file_ptr_out << _record_hdr; + + ++kCounter; + } + + auto pos_end = file_ptr_out.tellp(); + + file_ptr_out.seekp(pos); + + hdr.fStartCode = pos_end; + hdr.fCodeSize = kAppBytes.size(); + + file_ptr_out << hdr; + + file_ptr_out.seekp(pos_end); + } else { + if (kVerbose) { + kStdOut << "AssemblerAMD64: Write raw binary...\n"; + } + } + + // byte from byte, we write this. + for (auto& byte : kAppBytes) { + if (byte == 0) continue; + + if (byte == 0xFF) { + byte = 0; + } + + file_ptr_out << reinterpret_cast<const char*>(&byte)[0]; + } + + if (kVerbose) kStdOut << "AssemblerAMD64: Wrote file with program in it.\n"; + + file_ptr_out.flush(); + file_ptr_out.close(); + + if (kVerbose) kStdOut << "AssemblerAMD64: Exit succeeded.\n"; + + return 0; + } + +asm_fail_exit: + + if (kVerbose) kStdOut << "AssemblerAMD64: Exit failed.\n"; + + return 1; +} + +///////////////////////////////////////////////////////////////////////////////////////// + +// @brief Check for attributes +// returns true if any was found. + +///////////////////////////////////////////////////////////////////////////////////////// + +static bool asm_read_attributes(std::string line) { + // extern_segment is the opposite of public_segment, it signals to the ld + // that we need this symbol. + if (LibCompiler::find_word(line, "extern_segment")) { + if (kOutputAsBinary) { + Detail::print_error("Invalid directive in flat binary mode.", "LibCompiler"); + throw std::runtime_error("invalid_extern_segment_bin"); + } + + auto name = line.substr(line.find("extern_segment") + strlen("extern_segment") + 1); + + if (name.size() == 0) { + Detail::print_error("Invalid extern_segment", "power-as"); + throw std::runtime_error("invalid_extern_segment"); + } + + std::string result = std::to_string(name.size()); + result += kUndefinedSymbol; + + // mangle this + for (char& j : name) { + if (j == ' ' || j == ',') j = '$'; + } + + result += name; + + if (name.find(kPefCode64) != std::string::npos) { + // data is treated as code. + kCurrentRecord.fKind = LibCompiler::kPefCode; + } else if (name.find(kPefData64) != std::string::npos) { + // no code will be executed from here. + kCurrentRecord.fKind = LibCompiler::kPefData; + } else if (name.find(kPefZero64) != std::string::npos) { + // this is a bss section. + kCurrentRecord.fKind = LibCompiler::kPefZero; + } + + // this is a special case for the start stub. + // we want this so that ld can find it. + + if (name == kPefStart) { + kCurrentRecord.fKind = LibCompiler::kPefCode; + } + + // now we can tell the code size of the previous kCurrentRecord. + + if (!kRecords.empty()) kRecords[kRecords.size() - 1].fSize = kAppBytes.size(); + + memset(kCurrentRecord.fName, 0, kAESymbolLen); + memcpy(kCurrentRecord.fName, result.c_str(), result.size()); + + ++kCounter; + + memset(kCurrentRecord.fPad, kAENullType, kAEPad); + + kRecords.emplace_back(kCurrentRecord); + + return true; + } + // public_segment is a special keyword used by AssemblerAMD64 to tell the AE output stage to + // mark this section as a header. it currently supports .code64, .data64 and + // .zero64. + else if (LibCompiler::find_word(line, "public_segment")) { + if (kOutputAsBinary) { + Detail::print_error("Invalid directive in flat binary mode.", "LibCompiler"); + throw std::runtime_error("invalid_public_segment_bin"); + } + + auto name = line.substr(line.find("public_segment") + strlen("public_segment") + 1); + + std::string name_copy = name; + + for (char& j : name) { + if (j == ' ') j = '$'; + } + + if (std::find(kDefinedSymbols.begin(), kDefinedSymbols.end(), name) != kDefinedSymbols.end()) { + Detail::print_error("Symbol already defined.", "LibCompiler"); + throw std::runtime_error("invalid_public_segment_bin"); + } + + kDefinedSymbols.push_back(name); + + if (name.find(".code64") != std::string::npos) { + // data is treated as code. + + name_copy.erase(name_copy.find(".code64"), strlen(".code64")); + kCurrentRecord.fKind = LibCompiler::kPefCode; + } else if (name.find(".data64") != std::string::npos) { + // no code will be executed from here. + + name_copy.erase(name_copy.find(".data64"), strlen(".data64")); + kCurrentRecord.fKind = LibCompiler::kPefData; + } else if (name.find(".zero64") != std::string::npos) { + // this is a bss section. + + name_copy.erase(name_copy.find(".zero64"), strlen(".zero64")); + kCurrentRecord.fKind = LibCompiler::kPefZero; + } + + // this is a special case for the start stub. + // we want this so that ld can find it. + + if (name == kPefStart) { + kCurrentRecord.fKind = LibCompiler::kPefCode; + } + + while (name_copy.find(" ") != std::string::npos) name_copy.erase(name_copy.find(" "), 1); + + kOriginLabel.push_back(std::make_pair(name_copy, kOrigin)); + ++kOrigin; + + // now we can tell the code size of the previous kCurrentRecord. + + if (!kRecords.empty()) kRecords[kRecords.size() - 1].fSize = kAppBytes.size(); + + memset(kCurrentRecord.fName, 0, kAESymbolLen); + memcpy(kCurrentRecord.fName, name.c_str(), name.size()); + + ++kCounter; + + memset(kCurrentRecord.fPad, kAENullType, kAEPad); + + kRecords.emplace_back(kCurrentRecord); + + return true; + } + + return false; +} + +// \brief algorithms and helpers. + +namespace Detail::algorithm { +// \brief authorize a brief set of characters. +static inline bool is_not_valid(char c) { + if ((isalpha(c) || isdigit(c)) || + ((c == ' ') || (c == '\t') || (c == ',') || (c == '(') || (c == ')') || (c == '"') || + (c == '*') || (c == '\'') || (c == '[') || (c == ']') || (c == '+') || (c == '_') || + (c == ':') || (c == '@') || (c == '.') || (c == '#') || (c == ';'))) + return false; + + return true; +} + +bool is_valid_amd64(std::string str) { + return std::find_if(str.begin(), str.end(), is_not_valid) == str.end(); +} +} // namespace Detail::algorithm + +///////////////////////////////////////////////////////////////////////////////////////// + +// @brief Check for line (syntax check) + +///////////////////////////////////////////////////////////////////////////////////////// + +std::string LibCompiler::EncoderAMD64::CheckLine(std::string line, std::string file) { + std::string err_str; + + if (line.empty() || LibCompiler::find_word(line, "extern_segment") || + LibCompiler::find_word(line, "public_segment") || + LibCompiler::find_word(line, kAssemblerPragmaSymStr) || LibCompiler::find_word(line, ";") || + line[0] == kAssemblerPragmaSym) { + if (line.find(';') != std::string::npos) { + line.erase(line.find(';')); + } else { + // now check the line for validity + if (!Detail::algorithm::is_valid_amd64(line)) { + err_str = "Line contains non valid characters.\nhere -> "; + err_str += line; + } + } + + return err_str; + } + + // check for a valid instruction format. + + if (line.find(',') != std::string::npos) { + if (line.find(',') + 1 == line.size()) { + err_str += "\nInstruction lacks right register, here -> "; + err_str += line.substr(line.find(',')); + + return err_str; + } else { + bool nothing_on_right = true; + + if (line.find(',') + 1 > line.size()) { + err_str += "\nInstruction not complete, here -> "; + err_str += line; + + return err_str; + } + + auto substr = line.substr(line.find(',') + 1); + + for (auto& ch : substr) { + if (ch != ' ' && ch != '\t') { + nothing_on_right = false; + } + } + + // this means we found nothing after that ',' . + if (nothing_on_right) { + err_str += "\nInstruction not complete, here -> "; + err_str += line; + + return err_str; + } + } + } + for (auto& opcodeAMD64 : kOpcodesAMD64) { + if (LibCompiler::find_word(line, opcodeAMD64.fName)) { + return err_str; + } + } + + err_str += "\nUnrecognized instruction -> " + line; + + return err_str; +} + +bool LibCompiler::EncoderAMD64::WriteNumber(const std::size_t& pos, std::string& jump_label) { + if (!isdigit(jump_label[pos])) return false; + + switch (jump_label[pos + 1]) { + case 'x': { + if (auto res = strtol(jump_label.substr(pos + 2).c_str(), nullptr, 16); !res) { + if (errno != 0) { + Detail::print_error("invalid hex number: " + jump_label, "LibCompiler"); + throw std::runtime_error("invalid_hex"); + } + } + + LibCompiler::NumberCast64 num = + LibCompiler::NumberCast64(strtol(jump_label.substr(pos + 2).c_str(), nullptr, 16)); + + for (char& i : num.number) { + if (i == 0) i = 0xFF; + + kAppBytes.push_back(i); + } + + if (kVerbose) { + kStdOut << "AssemblerAMD64: Found a base 16 number here: " << jump_label.substr(pos) + << "\n"; + } + + return true; + } + case 'b': { + if (auto res = strtol(jump_label.substr(pos + 2).c_str(), nullptr, 2); !res) { + if (errno != 0) { + Detail::print_error("invalid binary number: " + jump_label, "LibCompiler"); + throw std::runtime_error("invalid_bin"); + } + } + + LibCompiler::NumberCast64 num = + LibCompiler::NumberCast64(strtol(jump_label.substr(pos + 2).c_str(), nullptr, 2)); + + if (kVerbose) { + kStdOut << "AssemblerAMD64: Found a base 2 number here: " << jump_label.substr(pos) << "\n"; + } + + for (char& i : num.number) { + if (i == 0) i = 0xFF; + + kAppBytes.push_back(i); + } + + return true; + } + case 'o': { + if (auto res = strtol(jump_label.substr(pos + 2).c_str(), nullptr, 7); !res) { + if (errno != 0) { + Detail::print_error("invalid octal number: " + jump_label, "LibCompiler"); + throw std::runtime_error("invalid_octal"); + } + } + + LibCompiler::NumberCast64 num = + LibCompiler::NumberCast64(strtol(jump_label.substr(pos + 2).c_str(), nullptr, 7)); + + if (kVerbose) { + kStdOut << "AssemblerAMD64: Found a base 8 number here: " << jump_label.substr(pos) << "\n"; + } + + for (char& i : num.number) { + if (i == 0) i = 0xFF; + + kAppBytes.push_back(i); + } + + return true; + } + default: { + break; + } + } + + /* check for errno and stuff like that */ + if (auto res = strtol(jump_label.substr(pos).c_str(), nullptr, 10); !res) { + if (errno != 0) { + return false; + } + } + + LibCompiler::NumberCast64 num = + LibCompiler::NumberCast64(strtol(jump_label.substr(pos).c_str(), nullptr, 10)); + + for (char& i : num.number) { + if (i == 0) i = 0xFF; + + kAppBytes.push_back(i); + } + + if (kVerbose) { + kStdOut << "AssemblerAMD64: Found a base 10 number here: " << jump_label.substr(pos) << "\n"; + } + + return true; +} + +bool LibCompiler::EncoderAMD64::WriteNumber32(const std::size_t& pos, std::string& jump_label) { + if (!isdigit(jump_label[pos])) return false; + + switch (jump_label[pos + 1]) { + case 'x': { + auto res = strtol(jump_label.substr(pos + 2).c_str(), nullptr, 16); + res += kOrigin; + + if (errno != 0) { + return false; + } + + LibCompiler::NumberCast32 num = LibCompiler::NumberCast32(res); + + for (char& i : num.number) { + if (i == 0) i = 0xFF; + + kAppBytes.push_back(i); + } + + if (kVerbose) { + kStdOut << "AssemblerAMD64: Found a base 16 number here: " << jump_label.substr(pos) + << "\n"; + } + + return true; + } + case 'b': { + auto res = strtol(jump_label.substr(pos + 2).c_str(), nullptr, 2); + res += kOrigin; + + if (errno != 0) { + return false; + } + + LibCompiler::NumberCast32 num = LibCompiler::NumberCast32(res); + + if (kVerbose) { + kStdOut << "AssemblerAMD64: Found a base 2 number here: " << jump_label.substr(pos) << "\n"; + } + + for (char& i : num.number) { + if (i == 0) i = 0xFF; + + kAppBytes.push_back(i); + } + + return true; + } + case 'o': { + auto res = strtol(jump_label.substr(pos + 2).c_str(), nullptr, 7); + res += kOrigin; + + if (errno != 0) { + return false; + } + + LibCompiler::NumberCast32 num = LibCompiler::NumberCast32(res); + + if (kVerbose) { + kStdOut << "AssemblerAMD64: Found a base 8 number here: " << jump_label.substr(pos) << "\n"; + } + + for (char& i : num.number) { + if (i == 0) i = 0xFF; + + kAppBytes.push_back(i); + } + + return true; + } + default: { + break; + } + } + + auto res = strtol(jump_label.substr(pos + 2).c_str(), nullptr, 10); + res += kOrigin; + + if (errno != 0) { + return false; + } + + LibCompiler::NumberCast32 num = LibCompiler::NumberCast32(res); + + for (char& i : num.number) { + if (i == 0) i = 0xFF; + + kAppBytes.push_back(i); + } + + if (kVerbose) { + kStdOut << "AssemblerAMD64: Found a base 10 number here: " << jump_label.substr(pos) << "\n"; + } + + return true; +} + +bool LibCompiler::EncoderAMD64::WriteNumber16(const std::size_t& pos, std::string& jump_label) { + if (!isdigit(jump_label[pos])) return false; + + switch (jump_label[pos + 1]) { + case 'x': { + if (auto res = strtol(jump_label.substr(pos + 2).c_str(), nullptr, 16); !res) { + if (errno != 0) { + Detail::print_error("invalid hex number: " + jump_label, "LibCompiler"); + throw std::runtime_error("invalid_hex"); + } + } + + LibCompiler::NumberCast16 num = + LibCompiler::NumberCast16(strtol(jump_label.substr(pos + 2).c_str(), nullptr, 16)); + + for (char& i : num.number) { + if (i == 0) i = 0xFF; + + kAppBytes.push_back(i); + } + + if (kVerbose) { + kStdOut << "AssemblerAMD64: Found a base 16 number here: " << jump_label.substr(pos) + << "\n"; + } + + return true; + } + case 'b': { + if (auto res = strtol(jump_label.substr(pos + 2).c_str(), nullptr, 2); !res) { + if (errno != 0) { + Detail::print_error("invalid binary number: " + jump_label, "LibCompiler"); + throw std::runtime_error("invalid_bin"); + } + } + + LibCompiler::NumberCast16 num = + LibCompiler::NumberCast16(strtol(jump_label.substr(pos + 2).c_str(), nullptr, 2)); + + if (kVerbose) { + kStdOut << "AssemblerAMD64: Found a base 2 number here: " << jump_label.substr(pos) << "\n"; + } + + for (char& i : num.number) { + if (i == 0) i = 0xFF; + + kAppBytes.push_back(i); + } + + return true; + } + case 'o': { + if (auto res = strtol(jump_label.substr(pos + 2).c_str(), nullptr, 7); !res) { + if (errno != 0) { + Detail::print_error("invalid octal number: " + jump_label, "LibCompiler"); + throw std::runtime_error("invalid_octal"); + } + } + + LibCompiler::NumberCast16 num = + LibCompiler::NumberCast16(strtol(jump_label.substr(pos + 2).c_str(), nullptr, 7)); + + if (kVerbose) { + kStdOut << "AssemblerAMD64: Found a base 8 number here: " << jump_label.substr(pos) << "\n"; + } + + for (char& i : num.number) { + if (i == 0) i = 0xFF; + + kAppBytes.push_back(i); + } + + return true; + } + default: { + break; + } + } + + /* check for errno and stuff like that */ + if (auto res = strtol(jump_label.substr(pos).c_str(), nullptr, 10); !res) { + if (errno != 0) { + return false; + } + } + + LibCompiler::NumberCast16 num = + LibCompiler::NumberCast16(strtol(jump_label.substr(pos).c_str(), nullptr, 10)); + + for (char& i : num.number) { + if (i == 0) i = 0xFF; + + kAppBytes.push_back(i); + } + + if (kVerbose) { + kStdOut << "AssemblerAMD64: Found a base 10 number here: " << jump_label.substr(pos) << "\n"; + } + + return true; +} + +bool LibCompiler::EncoderAMD64::WriteNumber8(const std::size_t& pos, std::string& jump_label) { + if (!isdigit(jump_label[pos])) return false; + + switch (jump_label[pos + 1]) { + case 'x': { + if (auto res = strtol(jump_label.substr(pos + 2).c_str(), nullptr, 16); !res) { + if (errno != 0) { + Detail::print_error("invalid hex number: " + jump_label, "LibCompiler"); + throw std::runtime_error("invalid_hex"); + } + } + + LibCompiler::NumberCast8 num = + LibCompiler::NumberCast8(strtol(jump_label.substr(pos + 2).c_str(), nullptr, 16)); + + kAppBytes.push_back(num.number); + + if (kVerbose) { + kStdOut << "AssemblerAMD64: Found a base 16 number here: " << jump_label.substr(pos) + << "\n"; + } + + return true; + } + case 'b': { + if (auto res = strtol(jump_label.substr(pos + 2).c_str(), nullptr, 2); !res) { + if (errno != 0) { + Detail::print_error("invalid binary number: " + jump_label, "LibCompiler"); + throw std::runtime_error("invalid_bin"); + } + } + + LibCompiler::NumberCast8 num = + LibCompiler::NumberCast8(strtol(jump_label.substr(pos + 2).c_str(), nullptr, 2)); + + if (kVerbose) { + kStdOut << "AssemblerAMD64: Found a base 2 number here: " << jump_label.substr(pos) << "\n"; + } + + kAppBytes.push_back(num.number); + + return true; + } + case 'o': { + if (auto res = strtol(jump_label.substr(pos + 2).c_str(), nullptr, 7); !res) { + if (errno != 0) { + Detail::print_error("invalid octal number: " + jump_label, "LibCompiler"); + throw std::runtime_error("invalid_octal"); + } + } + + LibCompiler::NumberCast8 num = + LibCompiler::NumberCast8(strtol(jump_label.substr(pos + 2).c_str(), nullptr, 7)); + + if (kVerbose) { + kStdOut << "AssemblerAMD64: Found a base 8 number here: " << jump_label.substr(pos) << "\n"; + } + + kAppBytes.push_back(num.number); + + return true; + } + default: { + break; + } + } + + /* check for errno and stuff like that */ + if (auto res = strtol(jump_label.substr(pos).c_str(), nullptr, 10); !res) { + if (errno != 0) { + return false; + } + } + + LibCompiler::NumberCast8 num = + LibCompiler::NumberCast8(strtol(jump_label.substr(pos).c_str(), nullptr, 10)); + + kAppBytes.push_back(num.number); + + if (kVerbose) { + kStdOut << "AssemblerAMD64: Found a base 10 number here: " << jump_label.substr(pos) << "\n"; + } + + return true; +} + +///////////////////////////////////////////////////////////////////////////////////////// + +// @brief Read and write an instruction to the output array. + +///////////////////////////////////////////////////////////////////////////////////////// + +bool LibCompiler::EncoderAMD64::WriteLine(std::string line, std::string file) { + if (LibCompiler::find_word(line, "public_segment ")) return true; + + struct RegMapAMD64 { + std::string fName; + i64_byte_t fModRM; + }; + + std::vector<RegMapAMD64> kRegisterList{ + {.fName = "ax", .fModRM = 0x0}, {.fName = "cx", .fModRM = 1}, + {.fName = "dx", .fModRM = 0x2}, {.fName = "bx", .fModRM = 3}, + {.fName = "sp", .fModRM = 0x4}, {.fName = "bp", .fModRM = 5}, + {.fName = "si", .fModRM = 0x6}, {.fName = "di", .fModRM = 7}, + }; + + BOOL foundInstruction = false; + + for (auto& opcodeAMD64 : kOpcodesAMD64) { + // strict check here + if (LibCompiler::find_word(line, opcodeAMD64.fName) && + Detail::algorithm::is_valid_amd64(line)) { + foundInstruction = true; + std::string name(opcodeAMD64.fName); + + /// Move instruction handler. + if (line.find(name) != std::string::npos) { + if (name == "mov" || name == "xor") { + std::string substr = line.substr(line.find(name) + name.size()); + + uint64_t bits = kRegisterBitWidth; + + if (substr.find(",") == std::string::npos) { + Detail::print_error("Syntax error: missing right operand.", "LibCompiler"); + throw std::runtime_error("syntax_err"); + } + + bool onlyOneReg = true; + + std::vector<RegMapAMD64> currentRegList; + + for (auto& reg : kRegisterList) { + std::vector<char> regExt = {'e', 'r'}; + + for (auto& ext : regExt) { + std::string registerName; + + if (bits > 16) registerName.push_back(ext); + + registerName += reg.fName; + + while (line.find(registerName) != std::string::npos) { + line.erase(line.find(registerName), registerName.size()); + + if (bits == 16) { + if (registerName[0] == 'r') { + Detail::print_error("invalid size for register, current bit width is: " + + std::to_string(kRegisterBitWidth), + file); + throw std::runtime_error("invalid_reg_size"); + } + } + + currentRegList.push_back({.fName = registerName, .fModRM = reg.fModRM}); + } + } + } + + if (currentRegList.size() > 1) onlyOneReg = false; + + bool hasRBasedRegs = false; + + if (!onlyOneReg) { + /// very tricky to understand. + /// but this checks for a r8 through r15 register. + if (currentRegList[0].fName[0] == 'r' || currentRegList[1].fName[0] == 'r') { + if (isdigit(currentRegList[0].fName[1]) && isdigit(currentRegList[1].fName[1])) { + kAppBytes.emplace_back(0x4d); + hasRBasedRegs = true; + } else if (isdigit(currentRegList[0].fName[1]) || + isdigit(currentRegList[1].fName[1])) { + kAppBytes.emplace_back(0x4c); + hasRBasedRegs = true; + } + } + } + + if (name == "mov") { + if (bits == 64 || bits == 32) { + if (!hasRBasedRegs && bits >= 32) { + kAppBytes.emplace_back(opcodeAMD64.fOpcode); + } + + if (!onlyOneReg) kAppBytes.emplace_back(0x89); + } else if (bits == 16) { + if (hasRBasedRegs) { + Detail::print_error("Invalid combination of operands and registers.", + "LibCompiler"); + throw std::runtime_error("comb_op_reg"); + } else { + kAppBytes.emplace_back(0x66); + kAppBytes.emplace_back(0x89); + } + } + } else { + if (!hasRBasedRegs && bits >= 32) { + kAppBytes.emplace_back(opcodeAMD64.fOpcode); + } + + kAppBytes.emplace_back(0x31); + } + + if (onlyOneReg) { + auto num = GetNumber32(line, ","); + + for (auto& num_idx : num.number) { + if (num_idx == 0) num_idx = 0xFF; + } + + auto modrm = (0x3 << 6 | currentRegList[0].fModRM); + + kAppBytes.emplace_back(0xC7); // prefixed before placing the modrm and then the number. + kAppBytes.emplace_back(modrm); + + if (name != "xor") { + kAppBytes.emplace_back(num.number[0]); + kAppBytes.emplace_back(num.number[1]); + kAppBytes.emplace_back(num.number[2]); + kAppBytes.emplace_back(num.number[3]); + } + + break; + } + + if (currentRegList[1].fName[0] == 'r' && currentRegList[0].fName[0] == 'e') { + Detail::print_error("Invalid combination of operands and registers.", "LibCompiler"); + throw std::runtime_error("comb_op_reg"); + } + + if (currentRegList[0].fName[0] == 'r' && currentRegList[1].fName[0] == 'e') { + Detail::print_error("Invalid combination of operands and registers.", "LibCompiler"); + throw std::runtime_error("comb_op_reg"); + } + + if (bits == 16) { + if (currentRegList[0].fName[0] == 'r' || currentRegList[0].fName[0] == 'e') { + Detail::print_error("Invalid combination of operands and registers.", "LibCompiler"); + throw std::runtime_error("comb_op_reg"); + } + + if (currentRegList[1].fName[0] == 'r' || currentRegList[1].fName[0] == 'e') { + Detail::print_error("Invalid combination of operands and registers.", "LibCompiler"); + throw std::runtime_error("comb_op_reg"); + } + } else { + if (currentRegList[0].fName[0] != 'r' || currentRegList[0].fName[0] == 'e') { + Detail::print_error("Invalid combination of operands and registers.", "LibCompiler"); + throw std::runtime_error("comb_op_reg"); + } + + if (currentRegList[1].fName[0] != 'r' || currentRegList[1].fName[0] == 'e') { + Detail::print_error("Invalid combination of operands and registers.", "LibCompiler"); + throw std::runtime_error("comb_op_reg"); + } + } + + /// encode register using the modrm encoding. + + auto modrm = (0x3 << 6 | currentRegList[1].fModRM << 3 | currentRegList[0].fModRM); + + kAppBytes.emplace_back(modrm); + + break; + } + } + + if (name == "int" || name == "into" || name == "intd") { + kAppBytes.emplace_back(opcodeAMD64.fOpcode); + this->WriteNumber8(line.find(name) + name.size() + 1, line); + + break; + } else if (name == "jmp" || name == "call") { + kAppBytes.emplace_back(opcodeAMD64.fOpcode); + + if (!this->WriteNumber32(line.find(name) + name.size() + 1, line)) { + throw std::runtime_error("BUG: WriteNumber32"); + } + + break; + } else if (name == "syscall") { + kAppBytes.emplace_back(opcodeAMD64.fOpcode); + kAppBytes.emplace_back(0x05); + + break; + } else { + kAppBytes.emplace_back(opcodeAMD64.fOpcode); + + break; + } + } + } + + if (line[0] == kAssemblerPragmaSym) { + if (foundInstruction) { + Detail::print_error("Syntax error: " + line, file); + throw std::runtime_error("syntax_err"); + } + + if (line.find("bits 64") != std::string::npos) { + kRegisterBitWidth = 64U; + } else if (line.find("bits 32") != std::string::npos) { + kRegisterBitWidth = 32U; + } else if (line.find("bits 16") != std::string::npos) { + kRegisterBitWidth = 16U; + } else if (line.find("org") != std::string::npos) { + size_t base[] = {10, 16, 2, 7}; + + for (size_t i = 0; i < 4; i++) { + if (kOrigin = strtol((line.substr(line.find("org") + strlen("org") + 1)).c_str(), nullptr, + base[i]); + kOrigin) { + if (errno != 0) { + continue; + } else { + if (kVerbose) { + kStdOut << "AssemblerAMD64: Origin Set: " << kOrigin << std::endl; + } + + break; + } + } + } + } + } + /// write a dword + else if (line.find(".dword") != std::string::npos) { + this->WriteNumber32(line.find(".dword") + strlen(".dword") + 1, line); + } + /// write a long + else if (line.find(".long") != std::string::npos) { + this->WriteNumber(line.find(".long") + strlen(".long") + 1, line); + } + /// write a 16-bit number + else if (line.find(".word") != std::string::npos) { + this->WriteNumber16(line.find(".word") + strlen(".word") + 1, line); + } + + kOrigin += kIPAlignement; + + return true; +} + +// Last rev 13-1-24 diff --git a/dev/LibCompiler/src/Backend/AssemblerARM64.cc b/dev/LibCompiler/src/Backend/AssemblerARM64.cc new file mode 100644 index 0000000..8fad6b5 --- /dev/null +++ b/dev/LibCompiler/src/Backend/AssemblerARM64.cc @@ -0,0 +1,585 @@ +/* ------------------------------------------- + + Copyright (C) 2024-2025 Amlal EL Mahrouss, all rights reserved + +------------------------------------------- */ + +///////////////////////////////////////////////////////////////////////////////////////// + +/// @file AssemblerARM64.cxx +/// @author EL Mahrouss Amlal +/// @brief 'ACORN' Assembler. + +/// REMINDER: when dealing with an undefined symbol use (string +/// size):LinkerFindSymbol:(string) so that li will look for it. + +///////////////////////////////////////////////////////////////////////////////////////// + +#define __ASM_NEED_ARM64__ 1 + +#include <LibCompiler/AE.h> +#include <LibCompiler/Backend/Aarch64.h> +#include <LibCompiler/ErrorID.h> +#include <LibCompiler/Frontend.h> +#include <LibCompiler/PEF.h> +#include <LibCompiler/Util/LCAsmUtils.h> +#include <LibCompiler/Version.h> +#include <algorithm> +#include <filesystem> +#include <fstream> +#include <iostream> +#include <vector> + +///////////////////// + +// ANSI ESCAPE CODES + +///////////////////// + +#define kBlank "\e[0;30m" +#define kRed "\e[0;31m" +#define kWhite "\e[0;97m" +#define kYellow "\e[0;33m" + +constexpr auto cPowerIPAlignment = 0x1U; + +static CharType kOutputArch = LibCompiler::kPefArchARM64; + +static std::size_t kCounter = 1UL; + +static std::uintptr_t kOrigin = kPefBaseOrigin; +static std::vector<std::pair<std::string, std::uintptr_t>> kOriginLabel; + +static std::vector<uint8_t> kBytes; + +static LibCompiler::AERecordHeader kCurrentRecord{ + .fName = "", .fKind = LibCompiler::kPefCode, .fSize = 0, .fOffset = 0}; + +static std::vector<LibCompiler::AERecordHeader> kRecords; +static std::vector<std::string> kUndefinedSymbols; + +static const std::string kUndefinedSymbol = ":UndefinedSymbol:"; +static const std::string kRelocSymbol = ":RuntimeSymbol:"; + +// \brief forward decl. +static bool asm_read_attributes(std::string line); + +///////////////////////////////////////////////////////////////////////////////////////// + +/// @brief POWER assembler entrypoint, the program/module starts here. + +///////////////////////////////////////////////////////////////////////////////////////// + +LIBCOMPILER_MODULE(AssemblerMainARM64) { + ::signal(SIGSEGV, Detail::drvi_crash_handler); + + for (size_t i = 1; i < argc; ++i) { + if (argv[i][0] == '-') { + if (strcmp(argv[i], "--ver") == 0 || strcmp(argv[i], "--v") == 0) { + kStdOut << "AssemblerPower: AARCH64 Assembler Driver.\nAssemblerPower: " << kDistVersion + << "\nAssemblerPower: " + "Copyright (c) " + "Amlal El Mahrouss\n"; + return 0; + } else if (strcmp(argv[i], "--h") == 0) { + kStdOut << "AssemblerPower: AARCH64 Assembler Driver.\nAssemblerPower: Copyright (c) 2024 " + "Amlal El Mahrouss\n"; + kStdOut << "--version,/v: print program version.\n"; + kStdOut << "--verbose: print verbose output.\n"; + kStdOut << "--binary: output as flat binary.\n"; + + return 0; + } else if (strcmp(argv[i], "--binary") == 0) { + kOutputAsBinary = true; + continue; + } else if (strcmp(argv[i], "--verbose") == 0) { + kVerbose = true; + continue; + } + + kStdOut << "AssemblerPower: ignore " << argv[i] << "\n"; + continue; + } + + if (!std::filesystem::exists(argv[i])) { + kStdOut << "AssemblerPower: can't open: " << argv[i] << std::endl; + goto asm_fail_exit; + } + + std::string object_output(argv[i]); + + for (auto& ext : kAsmFileExts) { + if (object_output.find(ext) != std::string::npos) { + object_output.erase(object_output.find(ext), std::strlen(ext)); + } + } + + object_output += kOutputAsBinary ? kBinaryFileExt : kObjectFileExt; + + std::ifstream file_ptr(argv[i]); + std::ofstream file_ptr_out(object_output, std::ofstream::binary); + + if (file_ptr_out.bad()) { + if (kVerbose) { + kStdOut << "AssemblerPower: error: " << strerror(errno) << "\n"; + } + } + + std::string line; + + LibCompiler::AEHeader hdr{0}; + + memset(hdr.fPad, kAENullType, kAEPad); + + hdr.fMagic[0] = kAEMag0; + hdr.fMagic[1] = kAEMag1; + hdr.fSize = sizeof(LibCompiler::AEHeader); + hdr.fArch = kOutputArch; + + ///////////////////////////////////////////////////////////////////////////////////////// + + // COMPILATION LOOP + + ///////////////////////////////////////////////////////////////////////////////////////// + + LibCompiler::EncoderARM64 asm64; + + while (std::getline(file_ptr, line)) { + if (auto ln = asm64.CheckLine(line, argv[i]); !ln.empty()) { + Detail::print_error(ln, argv[i]); + continue; + } + + try { + asm_read_attributes(line); + asm64.WriteLine(line, argv[i]); + } catch (const std::exception& e) { + if (kVerbose) { + std::string what = e.what(); + Detail::print_warning("exit because of: " + what, "LibCompiler"); + } + + std::filesystem::remove(object_output); + goto asm_fail_exit; + } + } + + if (!kOutputAsBinary) { + if (kVerbose) { + kStdOut << "AssemblerARM64: Writing object file...\n"; + } + + // this is the final step, write everything to the file. + + auto pos = file_ptr_out.tellp(); + + hdr.fCount = kRecords.size() + kUndefinedSymbols.size(); + + file_ptr_out << hdr; + + if (kRecords.empty()) { + kStdErr << "AssemblerARM64: At least one record is needed to write an object " + "file.\nAssemblerARM64: Make one using `public_segment .code64 foo_bar`.\n"; + + std::filesystem::remove(object_output); + return 1; + } + + kRecords[kRecords.size() - 1].fSize = kBytes.size(); + + std::size_t record_count = 0UL; + + for (auto& record_hdr : kRecords) { + record_hdr.fFlags |= LibCompiler::kKindRelocationAtRuntime; + record_hdr.fOffset = record_count; + ++record_count; + + file_ptr_out << record_hdr; + + if (kVerbose) kStdOut << "AssemblerARM64: Wrote record " << record_hdr.fName << "...\n"; + } + + // increment once again, so that we won't lie about the kUndefinedSymbols. + ++record_count; + + for (auto& sym : kUndefinedSymbols) { + LibCompiler::AERecordHeader undefined_sym{0}; + + if (kVerbose) kStdOut << "AssemblerARM64: Wrote symbol " << sym << " to file...\n"; + + undefined_sym.fKind = kAENullType; + undefined_sym.fSize = sym.size(); + undefined_sym.fOffset = record_count; + + ++record_count; + + memset(undefined_sym.fPad, kAENullType, kAEPad); + memcpy(undefined_sym.fName, sym.c_str(), sym.size()); + + file_ptr_out << undefined_sym; + + ++kCounter; + } + + auto pos_end = file_ptr_out.tellp(); + + file_ptr_out.seekp(pos); + + hdr.fStartCode = pos_end; + hdr.fCodeSize = kBytes.size(); + + file_ptr_out << hdr; + + file_ptr_out.seekp(pos_end); + } else { + if (kVerbose) { + kStdOut << "AssemblerARM64: Write raw binary...\n"; + } + } + + // byte from byte, we write this. + for (auto& byte : kBytes) { + file_ptr_out.write(reinterpret_cast<const char*>(&byte), sizeof(byte)); + } + + if (kVerbose) kStdOut << "AssemblerARM64: Wrote file with program in it.\n"; + + file_ptr_out.flush(); + file_ptr_out.close(); + + if (kVerbose) kStdOut << "AssemblerARM64: Exit succeeded.\n"; + + return 0; + } + +asm_fail_exit: + + if (kVerbose) kStdOut << "AssemblerARM64: Exit failed.\n"; + + return LIBCOMPILER_EXEC_ERROR; +} + +///////////////////////////////////////////////////////////////////////////////////////// + +// @brief Check for attributes +// returns true if any was found. + +///////////////////////////////////////////////////////////////////////////////////////// + +static bool asm_read_attributes(std::string line) { + // extern_segment is the opposite of public_segment, it signals to the li + // that we need this symbol. + if (LibCompiler::find_word(line, "extern_segment")) { + if (kOutputAsBinary) { + Detail::print_error("Invalid extern_segment directive in flat binary mode.", "LibCompiler"); + throw std::runtime_error("invalid_extern_segment_bin"); + } + + auto name = line.substr(line.find("extern_segment") + strlen("extern_segment") + 1); + + if (name.size() == 0) { + Detail::print_error("Invalid extern_segment", "LibCompiler"); + throw std::runtime_error("invalid_extern_segment"); + } + + std::string result = std::to_string(name.size()); + result += kUndefinedSymbol; + + // mangle this + for (char& j : name) { + if (j == ' ' || j == ',') j = '$'; + } + + result += name; + + if (name.find(".code64") != std::string::npos) { + // data is treated as code. + kCurrentRecord.fKind = LibCompiler::kPefCode; + } else if (name.find(".data64") != std::string::npos) { + // no code will be executed from here. + kCurrentRecord.fKind = LibCompiler::kPefData; + } else if (name.find(".zero64") != std::string::npos) { + // this is a bss section. + kCurrentRecord.fKind = LibCompiler::kPefZero; + } + + // this is a special case for the start stub. + // we want this so that li can find it. + + if (name == kPefStart) { + kCurrentRecord.fKind = LibCompiler::kPefCode; + } + + // now we can tell the code size of the previous kCurrentRecord. + + if (!kRecords.empty()) kRecords[kRecords.size() - 1].fSize = kBytes.size(); + + memset(kCurrentRecord.fName, 0, kAESymbolLen); + memcpy(kCurrentRecord.fName, result.c_str(), result.size()); + + ++kCounter; + + memset(kCurrentRecord.fPad, kAENullType, kAEPad); + + kRecords.emplace_back(kCurrentRecord); + + return true; + } + // public_segment is a special keyword used by Assembler to tell the AE output stage to + // mark this section as a header. it currently supports .code64, .data64., + // .zero64 + else if (LibCompiler::find_word(line, "public_segment")) { + if (kOutputAsBinary) { + Detail::print_error("Invalid public_segment directive in flat binary mode.", "LibCompiler"); + throw std::runtime_error("invalid_public_segment_bin"); + } + + auto name = line.substr(line.find("public_segment") + strlen("public_segment")); + + std::string name_copy = name; + + for (char& j : name) { + if (j == ' ') j = '$'; + } + + if (name.find(".code64") != std::string::npos) { + // data is treated as code. + + name_copy.erase(name_copy.find(".code64"), strlen(".code64")); + kCurrentRecord.fKind = LibCompiler::kPefCode; + } else if (name.find(".data64") != std::string::npos) { + // no code will be executed from here. + + name_copy.erase(name_copy.find(".data64"), strlen(".data64")); + kCurrentRecord.fKind = LibCompiler::kPefData; + } else if (name.find(".zero64") != std::string::npos) { + // this is a bss section. + + name_copy.erase(name_copy.find(".zero64"), strlen(".zero64")); + kCurrentRecord.fKind = LibCompiler::kPefZero; + } + + // this is a special case for the start stub. + // we want this so that li can find it. + + if (name == kPefStart) { + kCurrentRecord.fKind = LibCompiler::kPefCode; + } + + while (name_copy.find(" ") != std::string::npos) name_copy.erase(name_copy.find(" "), 1); + + kOriginLabel.push_back(std::make_pair(name_copy, kOrigin)); + ++kOrigin; + + // now we can tell the code size of the previous kCurrentRecord. + + if (!kRecords.empty()) kRecords[kRecords.size() - 1].fSize = kBytes.size(); + + memset(kCurrentRecord.fName, 0, kAESymbolLen); + memcpy(kCurrentRecord.fName, name.c_str(), name.size()); + + ++kCounter; + + memset(kCurrentRecord.fPad, kAENullType, kAEPad); + + kRecords.emplace_back(kCurrentRecord); + + return true; + } + + return false; +} + +// \brief algorithms and helpers. + +namespace Detail::algorithm { +// \brief authorize a brief set of characters. +static inline bool is_not_alnum_space(char c) { + return !(isalpha(c) || isdigit(c) || (c == ' ') || (c == '\t') || (c == ',') || (c == '(') || + (c == ')') || (c == '"') || (c == '\'') || (c == '[') || (c == ']') || (c == '+') || + (c == '_') || (c == ':') || (c == '@') || (c == '.')); +} + +bool is_valid_arm64(std::string str) { + return std::find_if(str.begin(), str.end(), is_not_alnum_space) == str.end(); +} +} // namespace Detail::algorithm + +///////////////////////////////////////////////////////////////////////////////////////// + +// @brief Check for line (syntax check) + +///////////////////////////////////////////////////////////////////////////////////////// + +std::string LibCompiler::EncoderARM64::CheckLine(std::string line, std::string file) { + std::string err_str; + + if (line.empty() || LibCompiler::find_word(line, "extern_segment") || + LibCompiler::find_word(line, "public_segment") || line.find('#') != std::string::npos || + LibCompiler::find_word(line, ";")) { + if (line.find('#') != std::string::npos) { + line.erase(line.find('#')); + } else if (line.find(';') != std::string::npos) { + line.erase(line.find(';')); + } else { + /// does the line contains valid input? + if (!Detail::algorithm::is_valid_arm64(line)) { + err_str = "Line contains non alphanumeric characters.\nhere -> "; + err_str += line; + } + } + + return err_str; + } + + if (!Detail::algorithm::is_valid_arm64(line)) { + err_str = "Line contains non alphanumeric characters.\nhere -> "; + err_str += line; + + return err_str; + } + + // check for a valid instruction format. + + if (line.find(',') != std::string::npos) { + if (line.find(',') + 1 == line.size()) { + err_str += "\nInstruction lacks right register, here -> "; + err_str += line.substr(line.find(',')); + + return err_str; + } else { + bool nothing_on_right = true; + + if (line.find(',') + 1 > line.size()) { + err_str += "\nInstruction not complete, here -> "; + err_str += line; + + return err_str; + } + + auto substr = line.substr(line.find(',') + 1); + + for (auto& ch : substr) { + if (ch != ' ' && ch != '\t') { + nothing_on_right = false; + } + } + + // this means we found nothing after that ',' . + if (nothing_on_right) { + err_str += "\nInstruction not complete, here -> "; + err_str += line; + + return err_str; + } + } + } + + return err_str; +} + +bool LibCompiler::EncoderARM64::WriteNumber(const std::size_t& pos, std::string& jump_label) { + if (!isdigit(jump_label[pos])) return false; + + switch (jump_label[pos + 1]) { + case 'x': { + if (auto res = strtol(jump_label.substr(pos + 2).c_str(), nullptr, 16); !res) { + if (errno != 0) { + Detail::print_error("invalid hex number: " + jump_label, "LibCompiler"); + throw std::runtime_error("invalid_hex"); + } + } + + LibCompiler::NumberCast64 num(strtol(jump_label.substr(pos + 2).c_str(), nullptr, 16)); + + for (char& i : num.number) { + kBytes.push_back(i); + } + + if (kVerbose) { + kStdOut << "AssemblerARM64: found a base 16 number here: " << jump_label.substr(pos) + << "\n"; + } + + return true; + } + case 'b': { + if (auto res = strtol(jump_label.substr(pos + 2).c_str(), nullptr, 2); !res) { + if (errno != 0) { + Detail::print_error("invalid binary number: " + jump_label, "LibCompiler"); + throw std::runtime_error("invalid_bin"); + } + } + + LibCompiler::NumberCast64 num(strtol(jump_label.substr(pos + 2).c_str(), nullptr, 2)); + + if (kVerbose) { + kStdOut << "AssemblerARM64: found a base 2 number here: " << jump_label.substr(pos) << "\n"; + } + + for (char& i : num.number) { + kBytes.push_back(i); + } + + return true; + } + case 'o': { + if (auto res = strtol(jump_label.substr(pos + 2).c_str(), nullptr, 7); !res) { + if (errno != 0) { + Detail::print_error("invalid octal number: " + jump_label, "LibCompiler"); + throw std::runtime_error("invalid_octal"); + } + } + + LibCompiler::NumberCast64 num(strtol(jump_label.substr(pos + 2).c_str(), nullptr, 7)); + + if (kVerbose) { + kStdOut << "AssemblerARM64: found a base 8 number here: " << jump_label.substr(pos) << "\n"; + } + + for (char& i : num.number) { + kBytes.push_back(i); + } + + return true; + } + default: { + break; + } + } + + /* check for errno and stuff like that */ + if (auto res = strtol(jump_label.substr(pos).c_str(), nullptr, 10); !res) { + if (errno != 0) { + return false; + } + } + + LibCompiler::NumberCast64 num(strtol(jump_label.substr(pos).c_str(), nullptr, 10)); + + for (char& i : num.number) { + kBytes.push_back(i); + } + + if (kVerbose) { + kStdOut << "AssemblerARM64: found a base 10 number here: " << jump_label.substr(pos) << "\n"; + } + + return true; +} + +///////////////////////////////////////////////////////////////////////////////////////// + +/// @brief Read and write an instruction to the output array. + +///////////////////////////////////////////////////////////////////////////////////////// + +bool LibCompiler::EncoderARM64::WriteLine(std::string line, std::string file) { + if (LibCompiler::find_word(line, "public_segment")) return false; + + if (!Detail::algorithm::is_valid_arm64(line)) return false; + + return true; +} + +// Last rev 13-1-24 diff --git a/dev/LibCompiler/src/Backend/AssemblerPowerPC.cc b/dev/LibCompiler/src/Backend/AssemblerPowerPC.cc new file mode 100644 index 0000000..86a70b9 --- /dev/null +++ b/dev/LibCompiler/src/Backend/AssemblerPowerPC.cc @@ -0,0 +1,909 @@ +/* ------------------------------------------- + + Copyright (C) 2024-2025 Amlal EL Mahrouss, all rights reserved + +------------------------------------------- */ + +///////////////////////////////////////////////////////////////////////////////////////// + +/// @file AssemblerPower.cxx +/// @author EL Mahrouss Amlal +/// @brief POWER Assembler. + +/// REMINDER: when dealing with an undefined symbol use (string +/// size):LinkerFindSymbol:(string) so that li will look for it. + +///////////////////////////////////////////////////////////////////////////////////////// + +#define __ASM_NEED_PPC__ 1 + +#include <LibCompiler/AE.h> +#include <LibCompiler/Backend/PowerPC.h> +#include <LibCompiler/ErrorID.h> +#include <LibCompiler/Frontend.h> +#include <LibCompiler/PEF.h> +#include <LibCompiler/Util/LCAsmUtils.h> +#include <LibCompiler/Version.h> +#include <algorithm> +#include <filesystem> +#include <fstream> +#include <iostream> +#include <vector> + +///////////////////// + +// ANSI ESCAPE CODES + +///////////////////// + +#define kBlank "\e[0;30m" +#define kRed "\e[0;31m" +#define kWhite "\e[0;97m" +#define kYellow "\e[0;33m" + +constexpr auto cPowerIPAlignment = 0x4U; + +static CharType kOutputArch = LibCompiler::kPefArchPowerPC; + +static std::size_t kCounter = 1UL; + +static std::uintptr_t kOrigin = kPefBaseOrigin; +static std::vector<std::pair<std::string, std::uintptr_t>> kOriginLabel; + +static std::vector<uint8_t> kBytes; + +static LibCompiler::AERecordHeader kCurrentRecord{ + .fName = "", .fKind = LibCompiler::kPefCode, .fSize = 0, .fOffset = 0}; + +static std::vector<LibCompiler::AERecordHeader> kRecords; +static std::vector<std::string> kUndefinedSymbols; + +static const std::string kUndefinedSymbol = ":UndefinedSymbol:"; +static const std::string kRelocSymbol = ":RuntimeSymbol:"; + +// \brief forward decl. +static bool asm_read_attributes(std::string line); + +///////////////////////////////////////////////////////////////////////////////////////// + +/// @brief POWER assembler entrypoint, the program/module starts here. + +///////////////////////////////////////////////////////////////////////////////////////// + +LIBCOMPILER_MODULE(AssemblerMainPower64) { + ::signal(SIGSEGV, Detail::drvi_crash_handler); + + for (size_t i = 1; i < argc; ++i) { + if (argv[i][0] == '-') { + if (strcmp(argv[i], "--ver") == 0 || strcmp(argv[i], "--v") == 0) { + kStdOut << "AssemblerPower: POWER64 Assembler Driver.\nAssemblerPower: " << kDistVersion + << "\nAssemblerPower: " + "Copyright (c) " + "Amlal El Mahrouss\n"; + return 0; + } else if (strcmp(argv[i], "--h") == 0) { + kStdOut << "AssemblerPower: POWER64 Assembler Driver.\nAssemblerPower: Copyright (c) 2024 " + "Amlal El Mahrouss\n"; + kStdOut << "--version,/v: print program version.\n"; + kStdOut << "--verbose: print verbose output.\n"; + kStdOut << "--binary: output as flat binary.\n"; + + return 0; + } else if (strcmp(argv[i], "--binary") == 0) { + kOutputAsBinary = true; + continue; + } else if (strcmp(argv[i], "--verbose") == 0) { + kVerbose = true; + continue; + } + + kStdOut << "AssemblerPower: ignore " << argv[i] << "\n"; + continue; + } + + if (!std::filesystem::exists(argv[i])) { + kStdOut << "AssemblerPower: can't open: " << argv[i] << std::endl; + goto asm_fail_exit; + } + + std::string object_output(argv[i]); + + for (auto& ext : kAsmFileExts) { + if (object_output.find(ext) != std::string::npos) { + object_output.erase(object_output.find(ext), std::strlen(ext)); + } + } + + object_output += kOutputAsBinary ? kBinaryFileExt : kObjectFileExt; + + std::ifstream file_ptr(argv[i]); + std::ofstream file_ptr_out(object_output, std::ofstream::binary); + + if (file_ptr_out.bad()) { + if (kVerbose) { + kStdOut << "AssemblerPower: error: " << strerror(errno) << "\n"; + } + } + + std::string line; + + LibCompiler::AEHeader hdr{0}; + + memset(hdr.fPad, kAENullType, kAEPad); + + hdr.fMagic[0] = kAEMag0; + hdr.fMagic[1] = kAEMag1; + hdr.fSize = sizeof(LibCompiler::AEHeader); + hdr.fArch = kOutputArch; + + ///////////////////////////////////////////////////////////////////////////////////////// + + // COMPILATION LOOP + + ///////////////////////////////////////////////////////////////////////////////////////// + + LibCompiler::EncoderPowerPC asm64; + + while (std::getline(file_ptr, line)) { + if (auto ln = asm64.CheckLine(line, argv[i]); !ln.empty()) { + Detail::print_error(ln, argv[i]); + continue; + } + + try { + asm_read_attributes(line); + asm64.WriteLine(line, argv[i]); + } catch (const std::exception& e) { + if (kVerbose) { + std::string what = e.what(); + Detail::print_warning("exit because of: " + what, "LibCompiler"); + } + + std::filesystem::remove(object_output); + goto asm_fail_exit; + } + } + + if (!kOutputAsBinary) { + if (kVerbose) { + kStdOut << "AssemblerPower: Writing object file...\n"; + } + + // this is the final step, write everything to the file. + + auto pos = file_ptr_out.tellp(); + + hdr.fCount = kRecords.size() + kUndefinedSymbols.size(); + + file_ptr_out << hdr; + + if (kRecords.empty()) { + kStdErr << "AssemblerPower: At least one record is needed to write an object " + "file.\nAssemblerPower: Make one using `public_segment .code64 foo_bar`.\n"; + + std::filesystem::remove(object_output); + return 1; + } + + kRecords[kRecords.size() - 1].fSize = kBytes.size(); + + std::size_t record_count = 0UL; + + for (auto& record_hdr : kRecords) { + record_hdr.fFlags |= LibCompiler::kKindRelocationAtRuntime; + record_hdr.fOffset = record_count; + ++record_count; + + file_ptr_out << record_hdr; + + if (kVerbose) kStdOut << "AssemblerPower: Wrote record " << record_hdr.fName << "...\n"; + } + + // increment once again, so that we won't lie about the kUndefinedSymbols. + ++record_count; + + for (auto& sym : kUndefinedSymbols) { + LibCompiler::AERecordHeader undefined_sym{0}; + + if (kVerbose) kStdOut << "AssemblerPower: Wrote symbol " << sym << " to file...\n"; + + undefined_sym.fKind = kAENullType; + undefined_sym.fSize = sym.size(); + undefined_sym.fOffset = record_count; + + ++record_count; + + memset(undefined_sym.fPad, kAENullType, kAEPad); + memcpy(undefined_sym.fName, sym.c_str(), sym.size()); + + file_ptr_out << undefined_sym; + + ++kCounter; + } + + auto pos_end = file_ptr_out.tellp(); + + file_ptr_out.seekp(pos); + + hdr.fStartCode = pos_end; + hdr.fCodeSize = kBytes.size(); + + file_ptr_out << hdr; + + file_ptr_out.seekp(pos_end); + } else { + if (kVerbose) { + kStdOut << "AssemblerPower: Write raw binary...\n"; + } + } + + // byte from byte, we write this. + for (auto& byte : kBytes) { + file_ptr_out.write(reinterpret_cast<const char*>(&byte), sizeof(byte)); + } + + if (kVerbose) kStdOut << "AssemblerPower: Wrote file with program in it.\n"; + + file_ptr_out.flush(); + file_ptr_out.close(); + + if (kVerbose) kStdOut << "AssemblerPower: Exit succeeded.\n"; + + return 0; + } + +asm_fail_exit: + + if (kVerbose) kStdOut << "AssemblerPower: Exit failed.\n"; + + return LIBCOMPILER_EXEC_ERROR; +} + +///////////////////////////////////////////////////////////////////////////////////////// + +// @brief Check for attributes +// returns true if any was found. + +///////////////////////////////////////////////////////////////////////////////////////// + +static bool asm_read_attributes(std::string line) { + // extern_segment is the opposite of public_segment, it signals to the li + // that we need this symbol. + if (LibCompiler::find_word(line, "extern_segment")) { + if (kOutputAsBinary) { + Detail::print_error("Invalid extern_segment directive in flat binary mode.", "LibCompiler"); + throw std::runtime_error("invalid_extern_segment_bin"); + } + + auto name = line.substr(line.find("extern_segment") + strlen("extern_segment") + 1); + + if (name.size() == 0) { + Detail::print_error("Invalid extern_segment", "LibCompiler"); + throw std::runtime_error("invalid_extern_segment"); + } + + std::string result = std::to_string(name.size()); + result += kUndefinedSymbol; + + // mangle this + for (char& j : name) { + if (j == ' ' || j == ',') j = '$'; + } + + result += name; + + if (name.find(".code64") != std::string::npos) { + // data is treated as code. + kCurrentRecord.fKind = LibCompiler::kPefCode; + } else if (name.find(".data64") != std::string::npos) { + // no code will be executed from here. + kCurrentRecord.fKind = LibCompiler::kPefData; + } else if (name.find(".zero64") != std::string::npos) { + // this is a bss section. + kCurrentRecord.fKind = LibCompiler::kPefZero; + } + + // this is a special case for the start stub. + // we want this so that li can find it. + + if (name == kPefStart) { + kCurrentRecord.fKind = LibCompiler::kPefCode; + } + + // now we can tell the code size of the previous kCurrentRecord. + + if (!kRecords.empty()) kRecords[kRecords.size() - 1].fSize = kBytes.size(); + + memset(kCurrentRecord.fName, 0, kAESymbolLen); + memcpy(kCurrentRecord.fName, result.c_str(), result.size()); + + ++kCounter; + + memset(kCurrentRecord.fPad, kAENullType, kAEPad); + + kRecords.emplace_back(kCurrentRecord); + + return true; + } + // public_segment is a special keyword used by AssemblerPower to tell the AE output stage to + // mark this section as a header. it currently supports .code64, .data64., + // .zero64 + else if (LibCompiler::find_word(line, "public_segment")) { + if (kOutputAsBinary) { + Detail::print_error("Invalid public_segment directive in flat binary mode.", "LibCompiler"); + throw std::runtime_error("invalid_public_segment_bin"); + } + + auto name = line.substr(line.find("public_segment") + strlen("public_segment")); + + std::string name_copy = name; + + for (char& j : name) { + if (j == ' ') j = '$'; + } + + if (name.find(".code64") != std::string::npos) { + // data is treated as code. + + name_copy.erase(name_copy.find(".code64"), strlen(".code64")); + kCurrentRecord.fKind = LibCompiler::kPefCode; + } else if (name.find(".data64") != std::string::npos) { + // no code will be executed from here. + + name_copy.erase(name_copy.find(".data64"), strlen(".data64")); + kCurrentRecord.fKind = LibCompiler::kPefData; + } else if (name.find(".zero64") != std::string::npos) { + // this is a bss section. + + name_copy.erase(name_copy.find(".zero64"), strlen(".zero64")); + kCurrentRecord.fKind = LibCompiler::kPefZero; + } + + // this is a special case for the start stub. + // we want this so that li can find it. + + if (name == kPefStart) { + kCurrentRecord.fKind = LibCompiler::kPefCode; + } + + while (name_copy.find(" ") != std::string::npos) name_copy.erase(name_copy.find(" "), 1); + + kOriginLabel.push_back(std::make_pair(name_copy, kOrigin)); + ++kOrigin; + + // now we can tell the code size of the previous kCurrentRecord. + + if (!kRecords.empty()) kRecords[kRecords.size() - 1].fSize = kBytes.size(); + + memset(kCurrentRecord.fName, 0, kAESymbolLen); + memcpy(kCurrentRecord.fName, name.c_str(), name.size()); + + ++kCounter; + + memset(kCurrentRecord.fPad, kAENullType, kAEPad); + + kRecords.emplace_back(kCurrentRecord); + + return true; + } + + return false; +} + +// \brief algorithms and helpers. + +namespace Detail::algorithm { +// \brief authorize a brief set of characters. +static inline bool is_not_alnum_space(char c) { + return !(isalpha(c) || isdigit(c) || (c == ' ') || (c == '\t') || (c == ',') || (c == '(') || + (c == ')') || (c == '"') || (c == '\'') || (c == '[') || (c == ']') || (c == '+') || + (c == '_') || (c == ':') || (c == '@') || (c == '.')); +} + +bool is_valid_power64(std::string str) { + return std::find_if(str.begin(), str.end(), is_not_alnum_space) == str.end(); +} +} // namespace Detail::algorithm + +///////////////////////////////////////////////////////////////////////////////////////// + +// @brief Check for line (syntax check) + +///////////////////////////////////////////////////////////////////////////////////////// + +std::string LibCompiler::EncoderPowerPC::CheckLine(std::string line, std::string file) { + std::string err_str; + + if (line.empty() || LibCompiler::find_word(line, "extern_segment") || + LibCompiler::find_word(line, "public_segment") || line.find('#') != std::string::npos || + LibCompiler::find_word(line, ";")) { + if (line.find('#') != std::string::npos) { + line.erase(line.find('#')); + } else if (line.find(';') != std::string::npos) { + line.erase(line.find(';')); + } else { + /// does the line contains valid input? + if (!Detail::algorithm::is_valid_power64(line)) { + err_str = "Line contains non alphanumeric characters.\nhere -> "; + err_str += line; + } + } + + return err_str; + } + + if (!Detail::algorithm::is_valid_power64(line)) { + err_str = "Line contains non alphanumeric characters.\nhere -> "; + err_str += line; + + return err_str; + } + + // check for a valid instruction format. + + if (line.find(',') != std::string::npos) { + if (line.find(',') + 1 == line.size()) { + err_str += "\nInstruction lacks right register, here -> "; + err_str += line.substr(line.find(',')); + + return err_str; + } else { + bool nothing_on_right = true; + + if (line.find(',') + 1 > line.size()) { + err_str += "\nInstruction not complete, here -> "; + err_str += line; + + return err_str; + } + + auto substr = line.substr(line.find(',') + 1); + + for (auto& ch : substr) { + if (ch != ' ' && ch != '\t') { + nothing_on_right = false; + } + } + + // this means we found nothing after that ',' . + if (nothing_on_right) { + err_str += "\nInstruction not complete, here -> "; + err_str += line; + + return err_str; + } + } + } + + // these do take an argument. + std::vector<std::string> operands_inst = {"stw", "li"}; + + // these don't. + std::vector<std::string> filter_inst = {"blr", "bl", "sc"}; + + for (auto& opcode_risc : kOpcodesPowerPC) { + if (LibCompiler::find_word(line, opcode_risc.name)) { + for (auto& op : operands_inst) { + // if only the instruction was found. + if (line == op) { + err_str += "\nMalformed "; + err_str += op; + err_str += " instruction, here -> "; + err_str += line; + } + } + + // if it is like that -> addr1, 0x0 + if (auto it = std::find(filter_inst.begin(), filter_inst.end(), opcode_risc.name); + it == filter_inst.cend()) { + if (LibCompiler::find_word(line, opcode_risc.name)) { + if (!isspace(line[line.find(opcode_risc.name) + strlen(opcode_risc.name)])) { + err_str += "\nMissing space between "; + err_str += opcode_risc.name; + err_str += " and operands.\nhere -> "; + err_str += line; + } + } + } + + return err_str; + } + } + + err_str += "Unrecognized instruction: " + line; + + return err_str; +} + +bool LibCompiler::EncoderPowerPC::WriteNumber(const std::size_t& pos, std::string& jump_label) { + if (!isdigit(jump_label[pos])) return false; + + switch (jump_label[pos + 1]) { + case 'x': { + if (auto res = strtol(jump_label.substr(pos + 2).c_str(), nullptr, 16); !res) { + if (errno != 0) { + Detail::print_error("invalid hex number: " + jump_label, "LibCompiler"); + throw std::runtime_error("invalid_hex"); + } + } + + LibCompiler::NumberCast64 num(strtol(jump_label.substr(pos + 2).c_str(), nullptr, 16)); + + for (char& i : num.number) { + kBytes.push_back(i); + } + + if (kVerbose) { + kStdOut << "AssemblerPower: found a base 16 number here: " << jump_label.substr(pos) + << "\n"; + } + + return true; + } + case 'b': { + if (auto res = strtol(jump_label.substr(pos + 2).c_str(), nullptr, 2); !res) { + if (errno != 0) { + Detail::print_error("invalid binary number: " + jump_label, "LibCompiler"); + throw std::runtime_error("invalid_bin"); + } + } + + LibCompiler::NumberCast64 num(strtol(jump_label.substr(pos + 2).c_str(), nullptr, 2)); + + if (kVerbose) { + kStdOut << "AssemblerPower: found a base 2 number here: " << jump_label.substr(pos) << "\n"; + } + + for (char& i : num.number) { + kBytes.push_back(i); + } + + return true; + } + case 'o': { + if (auto res = strtol(jump_label.substr(pos + 2).c_str(), nullptr, 7); !res) { + if (errno != 0) { + Detail::print_error("invalid octal number: " + jump_label, "LibCompiler"); + throw std::runtime_error("invalid_octal"); + } + } + + LibCompiler::NumberCast64 num(strtol(jump_label.substr(pos + 2).c_str(), nullptr, 7)); + + if (kVerbose) { + kStdOut << "AssemblerPower: found a base 8 number here: " << jump_label.substr(pos) << "\n"; + } + + for (char& i : num.number) { + kBytes.push_back(i); + } + + return true; + } + default: { + break; + } + } + + /* check for errno and stuff like that */ + if (auto res = strtol(jump_label.substr(pos).c_str(), nullptr, 10); !res) { + if (errno != 0) { + return false; + } + } + + LibCompiler::NumberCast64 num(strtol(jump_label.substr(pos).c_str(), nullptr, 10)); + + for (char& i : num.number) { + kBytes.push_back(i); + } + + if (kVerbose) { + kStdOut << "AssemblerPower: found a base 10 number here: " << jump_label.substr(pos) << "\n"; + } + + return true; +} + +///////////////////////////////////////////////////////////////////////////////////////// + +/// @brief Read and write an instruction to the output array. + +///////////////////////////////////////////////////////////////////////////////////////// + +bool LibCompiler::EncoderPowerPC::WriteLine(std::string line, std::string file) { + if (LibCompiler::find_word(line, "public_segment")) return false; + if (!Detail::algorithm::is_valid_power64(line)) return false; + + for (auto& opcode_risc : kOpcodesPowerPC) { + // strict check here + if (LibCompiler::find_word(line, opcode_risc.name)) { + std::string name(opcode_risc.name); + std::string jump_label, cpy_jump_label; + std::vector<size_t> found_registers_index; + + // check funct7 type. + switch (opcode_risc.ops->type) { + default: { + NumberCast32 num(opcode_risc.opcode); + + for (auto ch : num.number) { + kBytes.emplace_back(ch); + } + break; + } + case BADDR: + case PCREL: { + auto num = GetNumber32(line, name); + + kBytes.emplace_back(num.number[0]); + kBytes.emplace_back(num.number[1]); + kBytes.emplace_back(num.number[2]); + kBytes.emplace_back(0x48); + + break; + } + /// General purpose, float, vector operations. Everything that involve + /// registers. + case G0REG: + case FREG: + case VREG: + case GREG: { + // \brief how many registers we found. + std::size_t found_some_count = 0UL; + std::size_t register_count = 0UL; + std::string opcodeName = opcode_risc.name; + std::size_t register_sum = 0; + + NumberCast64 num(opcode_risc.opcode); + + for (size_t line_index = 0UL; line_index < line.size(); line_index++) { + if (line[line_index] == kAsmRegisterPrefix[0] && isdigit(line[line_index + 1])) { + std::string register_syntax = kAsmRegisterPrefix; + register_syntax += line[line_index + 1]; + + if (isdigit(line[line_index + 2])) register_syntax += line[line_index + 2]; + + std::string reg_str; + reg_str += line[line_index + 1]; + + if (isdigit(line[line_index + 2])) reg_str += line[line_index + 2]; + + // it ranges from r0 to r19 + // something like r190 doesn't exist in the instruction set. + if (isdigit(line[line_index + 3]) && isdigit(line[line_index + 2])) { + reg_str += line[line_index + 3]; + Detail::print_error("invalid register index, r" + reg_str + + "\nnote: The POWER accepts registers from r0 to r32.", + file); + throw std::runtime_error("invalid_register_index"); + } + + // finally cast to a size_t + std::size_t reg_index = strtol(reg_str.c_str(), nullptr, 10); + + if (reg_index > kAsmRegisterLimit) { + Detail::print_error("invalid register index, r" + reg_str, file); + throw std::runtime_error("invalid_register_index"); + } + + if (opcodeName == "li") { + char numIndex = 0; + + for (size_t i = 0; i != reg_index; i++) { + numIndex += 0x20; + } + + auto num = GetNumber32(line, reg_str); + + kBytes.push_back(num.number[0]); + kBytes.push_back(num.number[1]); + kBytes.push_back(numIndex); + kBytes.push_back(0x38); + + // check if bigger than two. + for (size_t i = 2; i < 4; i++) { + if (num.number[i] > 0) { + Detail::print_warning("number overflow on li operation.", file); + break; + } + } + + break; + } + + if ((opcodeName[0] == 's' && opcodeName[1] == 't')) { + if (register_sum == 0) { + for (size_t indexReg = 0UL; indexReg < reg_index; ++indexReg) { + register_sum += 0x20; + } + } else { + register_sum += reg_index; + } + } + + if (opcodeName == "mr") { + switch (register_count) { + case 0: { + kBytes.push_back(0x78); + + char numIndex = 0x3; + + for (size_t i = 0; i != reg_index; i++) { + numIndex += 0x8; + } + + kBytes.push_back(numIndex); + + break; + } + case 1: { + char numIndex = 0x1; + + for (size_t i = 0; i != reg_index; i++) { + numIndex += 0x20; + } + + for (size_t i = 0; i != reg_index; i++) { + kBytes[kBytes.size() - 1] += 0x8; + } + + kBytes[kBytes.size() - 1] -= 0x8; + + kBytes.push_back(numIndex); + + if (reg_index >= 10 && reg_index < 20) + kBytes.push_back(0x7d); + else if (reg_index >= 20 && reg_index < 30) + kBytes.push_back(0x7e); + else if (reg_index >= 30) + kBytes.push_back(0x7f); + else + kBytes.push_back(0x7c); + + break; + } + default: + break; + } + + ++register_count; + ++found_some_count; + } + + if (opcodeName == "addi") { + if (found_some_count == 2 || found_some_count == 0) + kBytes.emplace_back(reg_index); + else if (found_some_count == 1) + kBytes.emplace_back(0x00); + + ++found_some_count; + + if (found_some_count > 3) { + Detail::print_error("Too much registers. -> " + line, file); + throw std::runtime_error("too_much_regs"); + } + } + + if (opcodeName.find("cmp") != std::string::npos) { + ++found_some_count; + + if (found_some_count > 3) { + Detail::print_error("Too much registers. -> " + line, file); + throw std::runtime_error("too_much_regs"); + } + } + + if (opcodeName.find("mf") != std::string::npos || + opcodeName.find("mt") != std::string::npos) { + char numIndex = 0; + + for (size_t i = 0; i != reg_index; i++) { + numIndex += 0x20; + } + + num.number[2] += numIndex; + + ++found_some_count; + + if (found_some_count > 1) { + Detail::print_error("Too much registers. -> " + line, file); + throw std::runtime_error("too_much_regs"); + } + + if (kVerbose) { + kStdOut << "AssemblerPower: Found register: " << register_syntax << "\n"; + kStdOut << "AssemblerPower: Amount of registers in instruction: " + << found_some_count << "\n"; + } + + if (reg_index >= 10 && reg_index < 20) + num.number[3] = 0x7d; + else if (reg_index >= 20 && reg_index < 30) + num.number[3] = 0x7e; + else if (reg_index >= 30) + num.number[3] = 0x7f; + else + num.number[3] = 0x7c; + + for (auto ch : num.number) { + kBytes.emplace_back(ch); + } + } + + found_registers_index.push_back(reg_index); + } + } + + if (opcodeName == "addi") { + kBytes.emplace_back(0x38); + } + + if (opcodeName.find("cmp") != std::string::npos) { + char rightReg = 0x0; + + for (size_t i = 0; i != found_registers_index[1]; i++) { + rightReg += 0x08; + } + + kBytes.emplace_back(0x00); + kBytes.emplace_back(rightReg); + kBytes.emplace_back(found_registers_index[0]); + kBytes.emplace_back(0x7c); + } + + if ((opcodeName[0] == 's' && opcodeName[1] == 't')) { + size_t offset = 0UL; + + if (line.find('+') != std::string::npos) { + auto number = GetNumber32(line.substr(line.find("+")), "+"); + offset = number.raw; + } + + kBytes.push_back(offset); + kBytes.push_back(0x00); + kBytes.push_back(register_sum); + + kBytes.emplace_back(0x90); + } + + if (opcodeName == "mr") { + if (register_count == 1) { + Detail::print_error("Too few registers. -> " + line, file); + throw std::runtime_error("too_few_registers"); + } + } + + // we're not in immediate addressing, reg to reg. + if (opcode_risc.ops->type != GREG) { + // remember! register to register! + if (found_some_count == 1) { + Detail::print_error( + "Unrecognized register found.\ntip: each AssemblerPower register " + "starts with 'r'.\nline: " + + line, + file); + + throw std::runtime_error("not_a_register"); + } + } + + if (found_some_count < 1 && name[0] != 'l' && name[0] != 's') { + Detail::print_error("invalid combination of opcode and registers.\nline: " + line, + file); + throw std::runtime_error("invalid_comb_op_reg"); + } + + break; + } + } + + kOrigin += cPowerIPAlignment; + break; + } + } + + return true; +} + +// Last rev 13-1-24 |