diff options
-rw-r--r-- | CMakeLists.txt | 1 | ||||
-rw-r--r-- | liblll/All.h | 6 | ||||
-rw-r--r-- | liblll/CMakeLists.txt | 23 | ||||
-rw-r--r-- | liblll/CodeFragment.cpp | 586 | ||||
-rw-r--r-- | liblll/CodeFragment.h | 64 | ||||
-rw-r--r-- | liblll/Compiler.cpp | 95 | ||||
-rw-r--r-- | liblll/Compiler.h | 38 | ||||
-rw-r--r-- | liblll/CompilerState.cpp | 82 | ||||
-rw-r--r-- | liblll/CompilerState.h | 57 | ||||
-rw-r--r-- | liblll/Exceptions.h | 44 | ||||
-rw-r--r-- | liblll/Parser.cpp | 145 | ||||
-rw-r--r-- | liblll/Parser.h | 41 |
12 files changed, 1182 insertions, 0 deletions
diff --git a/CMakeLists.txt b/CMakeLists.txt index aed3c212..4b3d6071 100644 --- a/CMakeLists.txt +++ b/CMakeLists.txt @@ -30,6 +30,7 @@ add_subdirectory(libevmasm) add_subdirectory(libsolidity) add_subdirectory(solc) if (NOT EMSCRIPTEN) + add_subdirectory(liblll) add_subdirectory(test) endif() diff --git a/liblll/All.h b/liblll/All.h new file mode 100644 index 00000000..7c4192f6 --- /dev/null +++ b/liblll/All.h @@ -0,0 +1,6 @@ +#pragma once + +#include "CodeFragment.h" +#include "Compiler.h" +#include "CompilerState.h" +#include "Parser.h" diff --git a/liblll/CMakeLists.txt b/liblll/CMakeLists.txt new file mode 100644 index 00000000..0b468d7a --- /dev/null +++ b/liblll/CMakeLists.txt @@ -0,0 +1,23 @@ +cmake_policy(SET CMP0015 NEW) +set(CMAKE_AUTOMOC OFF) + +set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -DSTATICLIB") + +aux_source_directory(. SRC_LIST) + +#include_directories(BEFORE ${JSONCPP_INCLUDE_DIRS}) +include_directories(BEFORE ..) +#include_directories(${Boost_INCLUDE_DIRS}) + +set(EXECUTABLE lll) + +file(GLOB HEADERS "*.h") + +add_library(${EXECUTABLE} ${SRC_LIST} ${HEADERS}) + +eth_use(${EXECUTABLE} REQUIRED Eth::evmasm) +#target_link_libraries(${EXECUTABLE} evmasm) + +install( TARGETS ${EXECUTABLE} RUNTIME DESTINATION bin ARCHIVE DESTINATION lib LIBRARY DESTINATION lib ) +install( FILES ${HEADERS} DESTINATION include/${EXECUTABLE} ) + diff --git a/liblll/CodeFragment.cpp b/liblll/CodeFragment.cpp new file mode 100644 index 00000000..64680d5a --- /dev/null +++ b/liblll/CodeFragment.cpp @@ -0,0 +1,586 @@ +/* + This file is part of cpp-ethereum. + + cpp-ethereum is free software: you can redistribute it and/or modify + it under the terms of the GNU General Public License as published by + the Free Software Foundation, either version 3 of the License, or + (at your option) any later version. + + cpp-ethereum is distributed in the hope that it will be useful, + but WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + GNU General Public License for more details. + + You should have received a copy of the GNU General Public License + along with cpp-ethereum. If not, see <http://www.gnu.org/licenses/>. +*/ +/** @file CodeFragment.cpp + * @author Gav Wood <i@gavwood.com> + * @date 2014 + */ + +#include "CodeFragment.h" + +#include <boost/algorithm/string.hpp> +#pragma warning(push) +#pragma GCC diagnostic push +#pragma GCC diagnostic ignored "-Wunused-parameter" +#include <boost/spirit/include/support_utree.hpp> +#pragma warning(pop) +#pragma GCC diagnostic pop +#include <libdevcore/Log.h> +#include <libdevcore/CommonIO.h> +#include <libevmcore/Instruction.h> +#include "CompilerState.h" +#include "Parser.h" +using namespace std; +using namespace dev; +using namespace dev::eth; +namespace qi = boost::spirit::qi; +namespace px = boost::phoenix; +namespace sp = boost::spirit; + +void CodeFragment::finalise(CompilerState const& _cs) +{ + if (_cs.usedAlloc && _cs.vars.size() && !m_finalised) + { + m_finalised = true; + m_asm.injectStart(Instruction::MSTORE8); + m_asm.injectStart((u256)((_cs.vars.size() + 2) * 32) - 1); + m_asm.injectStart((u256)1); + } +} + +CodeFragment::CodeFragment(sp::utree const& _t, CompilerState& _s, bool _allowASM) +{ +/* cdebug << "CodeFragment. Locals:"; + for (auto const& i: _s.defs) + cdebug << i.first << ":" << toHex(i.second.m_code); + cdebug << "Args:"; + for (auto const& i: _s.args) + cdebug << i.first << ":" << toHex(i.second.m_code); + cdebug << "Outers:"; + for (auto const& i: _s.outers) + cdebug << i.first << ":" << toHex(i.second.m_code); + debugOutAST(cout, _t); + cout << endl << flush; +*/ + switch (_t.which()) + { + case sp::utree_type::list_type: + constructOperation(_t, _s); + break; + case sp::utree_type::string_type: + { + auto sr = _t.get<sp::basic_string<boost::iterator_range<char const*>, sp::utree_type::string_type>>(); + string s(sr.begin(), sr.end()); + m_asm.append(s); + break; + } + case sp::utree_type::symbol_type: + { + auto sr = _t.get<sp::basic_string<boost::iterator_range<char const*>, sp::utree_type::symbol_type>>(); + string s(sr.begin(), sr.end()); + string us = boost::algorithm::to_upper_copy(s); + if (_allowASM && c_instructions.count(us)) + m_asm.append(c_instructions.at(us)); + else if (_s.defs.count(s)) + m_asm.append(_s.defs.at(s).m_asm); + else if (_s.args.count(s)) + m_asm.append(_s.args.at(s).m_asm); + else if (_s.outers.count(s)) + m_asm.append(_s.outers.at(s).m_asm); + else if (us.find_first_of("1234567890") != 0 && us.find_first_not_of("QWERTYUIOPASDFGHJKLZXCVBNM1234567890_") == string::npos) + { + auto it = _s.vars.find(s); + if (it == _s.vars.end()) + { + bool ok; + tie(it, ok) = _s.vars.insert(make_pair(s, make_pair(_s.stackSize, 32))); + _s.stackSize += 32; + } + m_asm.append((u256)it->second.first); + } + else + error<BareSymbol>(); + + break; + } + case sp::utree_type::any_type: + { + bigint i = *_t.get<bigint*>(); + if (i < 0 || i > bigint(u256(0) - 1)) + error<IntegerOutOfRange>(); + m_asm.append((u256)i); + break; + } + default: break; + } +} + +void CodeFragment::constructOperation(sp::utree const& _t, CompilerState& _s) +{ + if (_t.tag() == 0 && _t.empty()) + error<EmptyList>(); + else if (_t.tag() == 0 && _t.front().which() != sp::utree_type::symbol_type) + error<DataNotExecutable>(); + else + { + string s; + string us; + switch (_t.tag()) + { + case 0: + { + auto sr = _t.front().get<sp::basic_string<boost::iterator_range<char const*>, sp::utree_type::symbol_type>>(); + s = string(sr.begin(), sr.end()); + us = boost::algorithm::to_upper_copy(s); + break; + } + case 1: + us = "MLOAD"; + break; + case 2: + us = "SLOAD"; + break; + case 3: + us = "MSTORE"; + break; + case 4: + us = "SSTORE"; + break; + case 5: + us = "SEQ"; + break; + case 6: + us = "CALLDATALOAD"; + break; + default:; + } + + auto firstAsString = [&]() + { + auto i = *++_t.begin(); + if (i.tag()) + error<InvalidName>(); + if (i.which() == sp::utree_type::string_type) + { + auto sr = i.get<sp::basic_string<boost::iterator_range<char const*>, sp::utree_type::string_type>>(); + return string(sr.begin(), sr.end()); + } + else if (i.which() == sp::utree_type::symbol_type) + { + auto sr = i.get<sp::basic_string<boost::iterator_range<char const*>, sp::utree_type::symbol_type>>(); + return _s.getDef(string(sr.begin(), sr.end())).m_asm.backString(); + } + return string(); + }; + + auto varAddress = [&](string const& n) + { + auto it = _s.vars.find(n); + if (it == _s.vars.end()) + { + bool ok; + tie(it, ok) = _s.vars.insert(make_pair(n, make_pair(_s.stackSize, 32))); + _s.stackSize += 32; + } + return it->second.first; + }; + + // Operations who args are not standard stack-pushers. + bool nonStandard = true; + if (us == "ASM") + { + int c = 0; + for (auto const& i: _t) + if (c++) + m_asm.append(CodeFragment(i, _s, true).m_asm); + } + else if (us == "INCLUDE") + { + if (_t.size() != 2) + error<IncorrectParameterCount>(); + m_asm.append(CodeFragment::compile(contentsString(firstAsString()), _s).m_asm); + } + else if (us == "SET") + { + if (_t.size() != 3) + error<IncorrectParameterCount>(); + int c = 0; + for (auto const& i: _t) + if (c++ == 2) + m_asm.append(CodeFragment(i, _s, false).m_asm); + m_asm.append((u256)varAddress(firstAsString())); + m_asm.append(Instruction::MSTORE); + } + else if (us == "GET") + { + if (_t.size() != 2) + error<IncorrectParameterCount>(); + m_asm.append((u256)varAddress(firstAsString())); + m_asm.append(Instruction::MLOAD); + } + else if (us == "REF") + m_asm.append((u256)varAddress(firstAsString())); + else if (us == "DEF") + { + string n; + unsigned ii = 0; + if (_t.size() != 3 && _t.size() != 4) + error<IncorrectParameterCount>(); + vector<string> args; + for (auto const& i: _t) + { + if (ii == 1) + { + if (i.tag()) + error<InvalidName>(); + if (i.which() == sp::utree_type::string_type) + { + auto sr = i.get<sp::basic_string<boost::iterator_range<char const*>, sp::utree_type::string_type>>(); + n = string(sr.begin(), sr.end()); + } + else if (i.which() == sp::utree_type::symbol_type) + { + auto sr = i.get<sp::basic_string<boost::iterator_range<char const*>, sp::utree_type::symbol_type>>(); + n = _s.getDef(string(sr.begin(), sr.end())).m_asm.backString(); + } + } + else if (ii == 2) + if (_t.size() == 3) + _s.defs[n] = CodeFragment(i, _s); + else + for (auto const& j: i) + { + if (j.tag() || j.which() != sp::utree_type::symbol_type) + error<InvalidMacroArgs>(); + auto sr = j.get<sp::basic_string<boost::iterator_range<char const*>, sp::utree_type::symbol_type>>(); + args.push_back(string(sr.begin(), sr.end())); + } + else if (ii == 3) + { + auto k = make_pair(n, args.size()); + _s.macros[k].code = i; + _s.macros[k].env = _s.outers; + _s.macros[k].args = args; + for (auto const& i: _s.args) + _s.macros[k].env[i.first] = i.second; + for (auto const& i: _s.defs) + _s.macros[k].env[i.first] = i.second; + } + ++ii; + } + } + else if (us == "LIT") + { + if (_t.size() < 3) + error<IncorrectParameterCount>(); + unsigned ii = 0; + CodeFragment pos; + bytes data; + for (auto const& i: _t) + { + if (ii == 1) + { + pos = CodeFragment(i, _s); + if (pos.m_asm.deposit() != 1) + error<InvalidDeposit>(); + } + else if (ii == 2 && !i.tag() && i.which() == sp::utree_type::string_type) + { + auto sr = i.get<sp::basic_string<boost::iterator_range<char const*>, sp::utree_type::string_type>>(); + data = bytes((byte const*)sr.begin(), (byte const*)sr.end()); + } + else if (ii >= 2 && !i.tag() && i.which() == sp::utree_type::any_type) + { + bigint bi = *i.get<bigint*>(); + if (bi < 0) + error<IntegerOutOfRange>(); + else if (bi > bigint(u256(0) - 1)) + { + if (ii == 2 && _t.size() == 3) + { + // One big int - allow it as hex. + data.resize(bytesRequired(bi)); + toBigEndian(bi, data); + } + else + error<IntegerOutOfRange>(); + } + else + { + data.resize(data.size() + 32); + *(h256*)(&data.back() - 31) = (u256)bi; + } + } + else if (ii) + error<InvalidLiteral>(); + ++ii; + } + m_asm.append((u256)data.size()); + m_asm.append(Instruction::DUP1); + m_asm.append(data); + m_asm.append(pos.m_asm, 1); + m_asm.append(Instruction::CODECOPY); + } + else + nonStandard = false; + + if (nonStandard) + return; + + std::map<std::string, Instruction> const c_arith = { { "+", Instruction::ADD }, { "-", Instruction::SUB }, { "*", Instruction::MUL }, { "/", Instruction::DIV }, { "%", Instruction::MOD }, { "&", Instruction::AND }, { "|", Instruction::OR }, { "^", Instruction::XOR } }; + std::map<std::string, pair<Instruction, bool>> const c_binary = { { "<", { Instruction::LT, false } }, { "<=", { Instruction::GT, true } }, { ">", { Instruction::GT, false } }, { ">=", { Instruction::LT, true } }, { "S<", { Instruction::SLT, false } }, { "S<=", { Instruction::SGT, true } }, { "S>", { Instruction::SGT, false } }, { "S>=", { Instruction::SLT, true } }, { "=", { Instruction::EQ, false } }, { "!=", { Instruction::EQ, true } } }; + std::map<std::string, Instruction> const c_unary = { { "!", Instruction::ISZERO } }; + + vector<CodeFragment> code; + CompilerState ns = _s; + ns.vars.clear(); + ns.usedAlloc = false; + int c = _t.tag() ? 1 : 0; + for (auto const& i: _t) + if (c++) + { + if (us == "LLL" && c == 1) + code.push_back(CodeFragment(i, ns)); + else + code.push_back(CodeFragment(i, _s)); + } + auto requireSize = [&](unsigned s) { if (code.size() != s) error<IncorrectParameterCount>(); }; + auto requireMinSize = [&](unsigned s) { if (code.size() < s) error<IncorrectParameterCount>(); }; + auto requireMaxSize = [&](unsigned s) { if (code.size() > s) error<IncorrectParameterCount>(); }; + auto requireDeposit = [&](unsigned i, int s) { if (code[i].m_asm.deposit() != s) error<InvalidDeposit>(); }; + + if (_s.macros.count(make_pair(s, code.size()))) + { + Macro const& m = _s.macros.at(make_pair(s, code.size())); + CompilerState cs = _s; + for (auto const& i: m.env) + cs.outers[i.first] = i.second; + for (auto const& i: cs.defs) + cs.outers[i.first] = i.second; + cs.defs.clear(); + for (unsigned i = 0; i < m.args.size(); ++i) + { + //requireDeposit(i, 1); + cs.args[m.args[i]] = code[i]; + } + m_asm.append(CodeFragment(m.code, cs).m_asm); + for (auto const& i: cs.defs) + _s.defs[i.first] = i.second; + for (auto const& i: cs.macros) + _s.macros.insert(i); + } + else if (c_instructions.count(us)) + { + auto it = c_instructions.find(us); + int ea = instructionInfo(it->second).args; + if (ea >= 0) + requireSize(ea); + else + requireMinSize(-ea); + + for (unsigned i = code.size(); i; --i) + m_asm.append(code[i - 1].m_asm, 1); + m_asm.append(it->second); + } + else if (c_arith.count(us)) + { + auto it = c_arith.find(us); + requireMinSize(1); + for (unsigned i = code.size(); i; --i) + { + requireDeposit(i - 1, 1); + m_asm.append(code[i - 1].m_asm, 1); + } + for (unsigned i = 1; i < code.size(); ++i) + m_asm.append(it->second); + } + else if (c_binary.count(us)) + { + auto it = c_binary.find(us); + requireSize(2); + requireDeposit(0, 1); + requireDeposit(1, 1); + m_asm.append(code[1].m_asm, 1); + m_asm.append(code[0].m_asm, 1); + m_asm.append(it->second.first); + if (it->second.second) + m_asm.append(Instruction::ISZERO); + } + else if (c_unary.count(us)) + { + auto it = c_unary.find(us); + requireSize(1); + requireDeposit(0, 1); + m_asm.append(code[0].m_asm, 1); + m_asm.append(it->second); + } + else if (us == "IF") + { + requireSize(3); + requireDeposit(0, 1); + int minDep = min(code[1].m_asm.deposit(), code[2].m_asm.deposit()); + + m_asm.append(code[0].m_asm); + auto pos = m_asm.appendJumpI(); + m_asm.onePath(); + m_asm.append(code[2].m_asm, minDep); + auto end = m_asm.appendJump(); + m_asm.otherPath(); + m_asm << pos.tag(); + m_asm.append(code[1].m_asm, minDep); + m_asm << end.tag(); + m_asm.donePaths(); + } + else if (us == "WHEN" || us == "UNLESS") + { + requireSize(2); + requireDeposit(0, 1); + + m_asm.append(code[0].m_asm); + if (us == "WHEN") + m_asm.append(Instruction::ISZERO); + auto end = m_asm.appendJumpI(); + m_asm.onePath(); + m_asm.otherPath(); + m_asm.append(code[1].m_asm, 0); + m_asm << end.tag(); + m_asm.donePaths(); + } + else if (us == "WHILE") + { + requireSize(2); + requireDeposit(0, 1); + + auto begin = m_asm.append(); + m_asm.append(code[0].m_asm); + m_asm.append(Instruction::ISZERO); + auto end = m_asm.appendJumpI(); + m_asm.append(code[1].m_asm, 0); + m_asm.appendJump(begin); + m_asm << end.tag(); + } + else if (us == "FOR") + { + requireSize(4); + requireDeposit(1, 1); + + m_asm.append(code[0].m_asm, 0); + auto begin = m_asm.append(); + m_asm.append(code[1].m_asm); + m_asm.append(Instruction::ISZERO); + auto end = m_asm.appendJumpI(); + m_asm.append(code[3].m_asm, 0); + m_asm.append(code[2].m_asm, 0); + m_asm.appendJump(begin); + m_asm << end.tag(); + } + else if (us == "ALLOC") + { + requireSize(1); + requireDeposit(0, 1); + + m_asm.append(Instruction::MSIZE); + m_asm.append(u256(0)); + m_asm.append(u256(1)); + m_asm.append(code[0].m_asm, 1); + m_asm.append(Instruction::MSIZE); + m_asm.append(Instruction::ADD); + m_asm.append(Instruction::SUB); + m_asm.append(Instruction::MSTORE8); + + _s.usedAlloc = true; + } + else if (us == "LLL") + { + requireMinSize(2); + requireMaxSize(3); + requireDeposit(1, 1); + + auto subPush = m_asm.appendSubSize(code[0].assembly(ns)); + m_asm.append(Instruction::DUP1); + if (code.size() == 3) + { + requireDeposit(2, 1); + m_asm.append(code[2].m_asm, 1); + m_asm.append(Instruction::LT); + m_asm.append(Instruction::ISZERO); + m_asm.append(Instruction::MUL); + m_asm.append(Instruction::DUP1); + } + m_asm.append(subPush); + m_asm.append(code[1].m_asm, 1); + m_asm.append(Instruction::CODECOPY); + } + else if (us == "&&" || us == "||") + { + requireMinSize(1); + for (unsigned i = 0; i < code.size(); ++i) + requireDeposit(i, 1); + + auto end = m_asm.newTag(); + if (code.size() > 1) + { + m_asm.append((u256)(us == "||" ? 1 : 0)); + for (unsigned i = 1; i < code.size(); ++i) + { + // Check if true - predicate + m_asm.append(code[i - 1].m_asm, 1); + if (us == "&&") + m_asm.append(Instruction::ISZERO); + m_asm.appendJumpI(end); + } + m_asm.append(Instruction::POP); + } + + // Check if true - predicate + m_asm.append(code.back().m_asm, 1); + + // At end now. + m_asm.append(end); + } + else if (us == "~") + { + requireSize(1); + requireDeposit(0, 1); + + m_asm.append(code[0].m_asm, 1); + m_asm.append((u256)1); + m_asm.append((u256)0); + m_asm.append(Instruction::SUB); + m_asm.append(Instruction::SUB); + } + else if (us == "SEQ") + { + unsigned ii = 0; + for (auto const& i: code) + if (++ii < code.size()) + m_asm.append(i.m_asm, 0); + else + m_asm.append(i.m_asm); + } + else if (us == "RAW") + { + for (auto const& i: code) + m_asm.append(i.m_asm); + m_asm.popTo(1); + } + else if (us.find_first_of("1234567890") != 0 && us.find_first_not_of("QWERTYUIOPASDFGHJKLZXCVBNM1234567890_") == string::npos) + m_asm.append((u256)varAddress(s)); + else + error<InvalidOperation>(); + } +} + +CodeFragment CodeFragment::compile(string const& _src, CompilerState& _s) +{ + CodeFragment ret; + sp::utree o; + parseTreeLLL(_src, o); + if (!o.empty()) + ret = CodeFragment(o, _s); + _s.treesToKill.push_back(o); + return ret; +} diff --git a/liblll/CodeFragment.h b/liblll/CodeFragment.h new file mode 100644 index 00000000..03f812b6 --- /dev/null +++ b/liblll/CodeFragment.h @@ -0,0 +1,64 @@ +/* + This file is part of cpp-ethereum. + + cpp-ethereum is free software: you can redistribute it and/or modify + it under the terms of the GNU General Public License as published by + the Free Software Foundation, either version 3 of the License, or + (at your option) any later version. + + cpp-ethereum is distributed in the hope that it will be useful, + but WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + GNU General Public License for more details. + + You should have received a copy of the GNU General Public License + along with cpp-ethereum. If not, see <http://www.gnu.org/licenses/>. +*/ +/** @file CodeFragment.h + * @author Gav Wood <i@gavwood.com> + * @date 2014 + */ + +#pragma once + +#include <libdevcore/Common.h> +#include <libevmcore/Instruction.h> +#include <libevmasm/Assembly.h> +#include "Exceptions.h" + +namespace boost { namespace spirit { class utree; } } +namespace sp = boost::spirit; + +namespace dev +{ +namespace eth +{ + +struct CompilerState; + +class CodeFragment +{ +public: + CodeFragment() {} + CodeFragment(sp::utree const& _t, CompilerState& _s, bool _allowASM = false); + + static CodeFragment compile(std::string const& _src, CompilerState& _s); + + /// Consolidates data and compiles code. + Assembly& assembly(CompilerState const& _cs) { finalise(_cs); return m_asm; } + +private: + void finalise(CompilerState const& _cs); + + template <class T> void error() const { BOOST_THROW_EXCEPTION(T() ); } + void constructOperation(sp::utree const& _t, CompilerState& _s); + + bool m_finalised = false; + Assembly m_asm; +}; + +static const CodeFragment NullCodeFragment; + +} +} + diff --git a/liblll/Compiler.cpp b/liblll/Compiler.cpp new file mode 100644 index 00000000..f1538789 --- /dev/null +++ b/liblll/Compiler.cpp @@ -0,0 +1,95 @@ +/* + This file is part of cpp-ethereum. + + cpp-ethereum is free software: you can redistribute it and/or modify + it under the terms of the GNU General Public License as published by + the Free Software Foundation, either version 3 of the License, or + (at your option) any later version. + + cpp-ethereum is distributed in the hope that it will be useful, + but WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + GNU General Public License for more details. + + You should have received a copy of the GNU General Public License + along with cpp-ethereum. If not, see <http://www.gnu.org/licenses/>. +*/ +/** @file Compiler.cpp + * @author Gav Wood <i@gavwood.com> + * @date 2014 + */ + +#include "Compiler.h" +#include "Parser.h" +#include "CompilerState.h" +#include "CodeFragment.h" + +using namespace std; +using namespace dev; +using namespace dev::eth; + +bytes dev::eth::compileLLL(string const& _src, bool _opt, vector<string>* _errors) +{ + try + { + CompilerState cs; + cs.populateStandard(); + auto f = CodeFragment::compile(_src, cs); + bytes ret = f.assembly(cs).optimise(_opt).assemble().bytecode; + for (auto i: cs.treesToKill) + killBigints(i); + return ret; + } + catch (Exception const& _e) + { + if (_errors) + { + _errors->push_back("Parse error."); + _errors->push_back(diagnostic_information(_e)); + } + } + catch (std::exception) + { + if (_errors) + _errors->push_back("Parse error."); + } + return bytes(); +} + +std::string dev::eth::compileLLLToAsm(std::string const& _src, bool _opt, std::vector<std::string>* _errors) +{ + try + { + CompilerState cs; + cs.populateStandard(); + string ret = CodeFragment::compile(_src, cs).assembly(cs).optimise(_opt).out(); + for (auto i: cs.treesToKill) + killBigints(i); + return ret; + } + catch (Exception const& _e) + { + if (_errors) + _errors->push_back(diagnostic_information(_e)); + } + catch (std::exception) + { + if (_errors) + _errors->push_back("Parse error."); + } + return string(); +} + +string dev::eth::parseLLL(string const& _src) +{ + sp::utree o; + try + { + parseTreeLLL(_src, o); + } + catch (...) {} + ostringstream ret; + debugOutAST(ret, o); + killBigints(o); + return ret.str(); +} diff --git a/liblll/Compiler.h b/liblll/Compiler.h new file mode 100644 index 00000000..0fadd278 --- /dev/null +++ b/liblll/Compiler.h @@ -0,0 +1,38 @@ +/* + This file is part of cpp-ethereum. + + cpp-ethereum is free software: you can redistribute it and/or modify + it under the terms of the GNU General Public License as published by + the Free Software Foundation, either version 3 of the License, or + (at your option) any later version. + + cpp-ethereum is distributed in the hope that it will be useful, + but WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + GNU General Public License for more details. + + You should have received a copy of the GNU General Public License + along with cpp-ethereum. If not, see <http://www.gnu.org/licenses/>. +*/ +/** @file Compiler.h + * @author Gav Wood <i@gavwood.com> + * @date 2014 + */ + +#pragma once + +#include <string> +#include <vector> +#include <libdevcore/Common.h> + +namespace dev +{ +namespace eth +{ + +std::string parseLLL(std::string const& _src); +std::string compileLLLToAsm(std::string const& _src, bool _opt = true, std::vector<std::string>* _errors = nullptr); +bytes compileLLL(std::string const& _src, bool _opt = true, std::vector<std::string>* _errors = nullptr); + +} +} diff --git a/liblll/CompilerState.cpp b/liblll/CompilerState.cpp new file mode 100644 index 00000000..63351bc4 --- /dev/null +++ b/liblll/CompilerState.cpp @@ -0,0 +1,82 @@ +/* + This file is part of cpp-ethereum. + + cpp-ethereum is free software: you can redistribute it and/or modify + it under the terms of the GNU General Public License as published by + the Free Software Foundation, either version 3 of the License, or + (at your option) any later version. + + cpp-ethereum is distributed in the hope that it will be useful, + but WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + GNU General Public License for more details. + + You should have received a copy of the GNU General Public License + along with cpp-ethereum. If not, see <http://www.gnu.org/licenses/>. +*/ +/** @file CompilerState.cpp + * @author Gav Wood <i@gavwood.com> + * @date 2014 + */ + +#include "CompilerState.h" +#include "CodeFragment.h" + +using namespace std; +using namespace dev; +using namespace dev::eth; + +CompilerState::CompilerState() +{ +} + +CodeFragment const& CompilerState::getDef(std::string const& _s) +{ + if (defs.count(_s)) + return defs.at(_s); + else if (args.count(_s)) + return args.at(_s); + else if (outers.count(_s)) + return outers.at(_s); + else + return NullCodeFragment; +} + +void CompilerState::populateStandard() +{ + static const string s = "{" + "(def 'gav 0x51ba59315b3a95761d0863b05ccc7a7f54703d99)" + "(def 'config 0x661005d2720d855f1d9976f88bb10c1a3398c77f)" + "(def 'allgas (- (gas) 21))" + "(def 'send (to value) (call allgas to value 0 0 0 0))" + "(def 'send (gaslimit to value) (call gaslimit to value 0 0 0 0))" + "(def 'msg (gaslimit to value data datasize outsize) { (set x outsize) (set y (alloc @32)) (call gaslimit to value data datasize @0 @32) @0 })" + "(def 'msg (gaslimit to value data datasize) { (call gaslimit to value data datasize 0 32) @0 })" + "(def 'msg (gaslimit to value data) { [0]:data (msg gaslimit to value 0 32) })" + "(def 'msg (to value data) { [0]:data (msg allgas to value 0 32) })" + "(def 'msg (to data) { [0]:data (msg allgas to 0 0 32) })" + "(def 'create (value code) { [0]:(msize) (create value @0 (lll code @0)) })" + "(def 'create (code) { [0]:(msize) (create 0 @0 (lll code @0)) })" + "(def 'sha3 (val) { [0]:val (sha3 0 32) })" + "(def 'sha3pair (a b) { [0]:a [32]:b (sha3 0 64) })" + "(def 'sha3trip (a b c) { [0]:a [32]:b [64]:c (sha3 0 96) })" + "(def 'return (val) { [0]:val (return 0 32) })" + "(def 'returnlll (code) (return 0 (lll code 0)) )" + "(def 'makeperm (name pos) { (def name (sload pos)) (def name (v) (sstore pos v)) } )" + "(def 'permcount 0)" + "(def 'perm (name) { (makeperm name permcount) (def 'permcount (+ permcount 1)) } )" + "(def 'namereg (msg config 0))" + "(def 'coinreg (msg config 1))" + "(def 'gavcoin (msg config 2))" + "(def 'sendgavcoin (to value) { [32]'send [64]:to [96]:value (call allgas gavcoin 0 32 96 0 0) })" + "(def 'regname (name) { [32]'register [64]name (call allgas namereg 0 32 64 0 0) })" + "(def 'regcoin (name) { [32]name (call allgas coinreg 0 32 32 0 0) })" + "(def 'regcoin (name denom) { [32]name [64]denom (call allgas coinreg 0 32 64 0 0) })" + "(def 'ecrecover (r s v hash) { [0] r [32] s [64] v [96] hash (msg allgas 1 0 0 128) })" + "(def 'sha256 (data datasize) (msg allgas 2 0 data datasize))" + "(def 'ripemd160 (data datasize) (msg allgas 3 0 data datasize))" + "(def 'sha256 (val) { [0]:val (sha256 0 32) })" + "(def 'ripemd160 (val) { [0]:val (ripemd160 0 32) })" + "}"; + CodeFragment::compile(s, *this); +} diff --git a/liblll/CompilerState.h b/liblll/CompilerState.h new file mode 100644 index 00000000..bfe56f92 --- /dev/null +++ b/liblll/CompilerState.h @@ -0,0 +1,57 @@ +/* + This file is part of cpp-ethereum. + + cpp-ethereum is free software: you can redistribute it and/or modify + it under the terms of the GNU General Public License as published by + the Free Software Foundation, either version 3 of the License, or + (at your option) any later version. + + cpp-ethereum is distributed in the hope that it will be useful, + but WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + GNU General Public License for more details. + + You should have received a copy of the GNU General Public License + along with cpp-ethereum. If not, see <http://www.gnu.org/licenses/>. +*/ +/** @file CompilerState.h + * @author Gav Wood <i@gavwood.com> + * @date 2014 + */ + +#pragma once + +#include <boost/spirit/include/support_utree.hpp> +#include "CodeFragment.h" + +namespace dev +{ +namespace eth +{ + +struct Macro +{ + std::vector<std::string> args; + boost::spirit::utree code; + std::map<std::string, CodeFragment> env; +}; + +struct CompilerState +{ + CompilerState(); + + CodeFragment const& getDef(std::string const& _s); + void populateStandard(); + + unsigned stackSize = 128; + std::map<std::string, std::pair<unsigned, unsigned>> vars; ///< maps name to stack offset & size. + std::map<std::string, CodeFragment> defs; + std::map<std::string, CodeFragment> args; + std::map<std::string, CodeFragment> outers; + std::map<std::pair<std::string, unsigned>, Macro> macros; + std::vector<boost::spirit::utree> treesToKill; + bool usedAlloc = false; +}; + +} +} diff --git a/liblll/Exceptions.h b/liblll/Exceptions.h new file mode 100644 index 00000000..1e9671b3 --- /dev/null +++ b/liblll/Exceptions.h @@ -0,0 +1,44 @@ +/* + This file is part of cpp-ethereum. + + cpp-ethereum is free software: you can redistribute it and/or modify + it under the terms of the GNU General Public License as published by + the Free Software Foundation, either version 3 of the License, or + (at your option) any later version. + + cpp-ethereum is distributed in the hope that it will be useful, + but WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + GNU General Public License for more details. + + You should have received a copy of the GNU General Public License + along with cpp-ethereum. If not, see <http://www.gnu.org/licenses/>. +*/ +/** @file Exceptions.h + * @author Gav Wood <i@gavwood.com> + * @date 2014 + */ + +#pragma once + +#include <libdevcore/Exceptions.h> + +namespace dev +{ +namespace eth +{ + +/// Compile a Low-level Lisp-like Language program into EVM-code. +class CompilerException: public dev::Exception {}; +class InvalidOperation: public CompilerException {}; +class IntegerOutOfRange: public CompilerException {}; +class EmptyList: public CompilerException {}; +class DataNotExecutable: public CompilerException {}; +class IncorrectParameterCount: public CompilerException {}; +class InvalidName: public CompilerException {}; +class InvalidMacroArgs: public CompilerException {}; +class InvalidLiteral: public CompilerException {}; +class BareSymbol: public CompilerException {}; + +} +} diff --git a/liblll/Parser.cpp b/liblll/Parser.cpp new file mode 100644 index 00000000..df30f389 --- /dev/null +++ b/liblll/Parser.cpp @@ -0,0 +1,145 @@ +/* + This file is part of cpp-ethereum. + + cpp-ethereum is free software: you can redistribute it and/or modify + it under the terms of the GNU General Public License as published by + the Free Software Foundation, either version 3 of the License, or + (at your option) any later version. + + cpp-ethereum is distributed in the hope that it will be useful, + but WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + GNU General Public License for more details. + + You should have received a copy of the GNU General Public License + along with cpp-ethereum. If not, see <http://www.gnu.org/licenses/>. +*/ +/** @file Parser.cpp + * @author Gav Wood <i@gavwood.com> + * @date 2014 + */ + +#include "Parser.h" + +#define BOOST_RESULT_OF_USE_DECLTYPE +#define BOOST_SPIRIT_USE_PHOENIX_V3 +#include <boost/spirit/include/qi.hpp> +#include <boost/spirit/include/phoenix.hpp> +#include <boost/spirit/include/support_utree.hpp> + +using namespace std; +using namespace dev; +using namespace dev::eth; +namespace qi = boost::spirit::qi; +namespace px = boost::phoenix; +namespace sp = boost::spirit; + +void dev::eth::killBigints(sp::utree const& _this) +{ + switch (_this.which()) + { + case sp::utree_type::list_type: for (auto const& i: _this) killBigints(i); break; + case sp::utree_type::any_type: delete _this.get<bigint*>(); break; + default:; + } +} + +void dev::eth::debugOutAST(ostream& _out, sp::utree const& _this) +{ + switch (_this.which()) + { + case sp::utree_type::list_type: + switch (_this.tag()) + { + case 0: _out << "( "; for (auto const& i: _this) { debugOutAST(_out, i); _out << " "; } _out << ")"; break; + case 1: _out << "@ "; debugOutAST(_out, _this.front()); break; + case 2: _out << "@@ "; debugOutAST(_out, _this.front()); break; + case 3: _out << "[ "; debugOutAST(_out, _this.front()); _out << " ] "; debugOutAST(_out, _this.back()); break; + case 4: _out << "[[ "; debugOutAST(_out, _this.front()); _out << " ]] "; debugOutAST(_out, _this.back()); break; + case 5: _out << "{ "; for (auto const& i: _this) { debugOutAST(_out, i); _out << " "; } _out << "}"; break; + case 6: _out << "$ "; debugOutAST(_out, _this.front()); break; + default:; + } + + break; + case sp::utree_type::int_type: _out << _this.get<int>(); break; + case sp::utree_type::string_type: _out << "\"" << _this.get<sp::basic_string<boost::iterator_range<char const*>, sp::utree_type::string_type>>() << "\""; break; + case sp::utree_type::symbol_type: _out << _this.get<sp::basic_string<boost::iterator_range<char const*>, sp::utree_type::symbol_type>>(); break; + case sp::utree_type::any_type: _out << *_this.get<bigint*>(); break; + default: _out << "nil"; + } +} + +namespace dev { namespace eth { +namespace parseTreeLLL_ { + +template<unsigned N> +struct tagNode +{ + void operator()(sp::utree& n, qi::rule<string::const_iterator, qi::ascii::space_type, sp::utree()>::context_type& c) const + { + (boost::fusion::at_c<0>(c.attributes) = n).tag(N); + } +}; + +}}} + +void dev::eth::parseTreeLLL(string const& _s, sp::utree& o_out) +{ + using qi::standard::space; + using qi::standard::space_type; + using dev::eth::parseTreeLLL_::tagNode; + using symbol_type = sp::basic_string<std::string, sp::utree_type::symbol_type>; + using it = string::const_iterator; + + static const u256 ether = u256(1000000000) * 1000000000; + static const u256 finney = u256(1000000000) * 1000000; + static const u256 szabo = u256(1000000000) * 1000; + + qi::rule<it, space_type, sp::utree()> element; + qi::rule<it, string()> str = '"' > qi::lexeme[+(~qi::char_(std::string("\"") + '\0'))] > '"'; + qi::rule<it, string()> strsh = '\'' > qi::lexeme[+(~qi::char_(std::string(" ;$@()[]{}:\n\t") + '\0'))]; + qi::rule<it, symbol_type()> symbol = qi::lexeme[+(~qi::char_(std::string(" $@[]{}:();\"\x01-\x1f\x7f") + '\0'))]; + qi::rule<it, string()> intstr = qi::lexeme[ qi::no_case["0x"][qi::_val = "0x"] >> *qi::char_("0-9a-fA-F")[qi::_val += qi::_1]] | qi::lexeme[+qi::char_("0-9")[qi::_val += qi::_1]]; + qi::rule<it, bigint()> integer = intstr; + qi::rule<it, bigint()> multiplier = qi::lit("wei")[qi::_val = 1] | qi::lit("szabo")[qi::_val = szabo] | qi::lit("finney")[qi::_val = finney] | qi::lit("ether")[qi::_val = ether]; + qi::rule<it, space_type, bigint()> quantity = integer[qi::_val = qi::_1] >> -multiplier[qi::_val *= qi::_1]; + qi::rule<it, space_type, sp::utree()> atom = quantity[qi::_val = px::construct<sp::any_ptr>(px::new_<bigint>(qi::_1))] | (str | strsh)[qi::_val = qi::_1] | symbol[qi::_val = qi::_1]; + qi::rule<it, space_type, sp::utree::list_type()> seq = '{' > *element > '}'; + qi::rule<it, space_type, sp::utree::list_type()> mload = '@' > element; + qi::rule<it, space_type, sp::utree::list_type()> sload = qi::lit("@@") > element; + qi::rule<it, space_type, sp::utree::list_type()> mstore = '[' > element > ']' > -qi::lit(":") > element; + qi::rule<it, space_type, sp::utree::list_type()> sstore = qi::lit("[[") > element > qi::lit("]]") > -qi::lit(":") > element; + qi::rule<it, space_type, sp::utree::list_type()> calldataload = qi::lit("$") > element; + qi::rule<it, space_type, sp::utree::list_type()> list = '(' > *element > ')'; + + qi::rule<it, space_type, sp::utree()> extra = sload[tagNode<2>()] | mload[tagNode<1>()] | sstore[tagNode<4>()] | mstore[tagNode<3>()] | seq[tagNode<5>()] | calldataload[tagNode<6>()]; + element = atom | list | extra; + + string s; + s.reserve(_s.size()); + bool incomment = false; + bool instring = false; + bool insstring = false; + for (auto i: _s) + { + if (i == ';' && !instring && !insstring) + incomment = true; + else if (i == '\n') + incomment = instring = insstring = false; + else if (i == '"' && !insstring) + instring = !instring; + else if (i == '\'') + insstring = true; + else if (i == ' ') + insstring = false; + if (!incomment) + s.push_back(i); + } + auto ret = s.cbegin(); + qi::phrase_parse(ret, s.cend(), element, space, qi::skip_flag::dont_postskip, o_out); + for (auto i = ret; i != s.cend(); ++i) + if (!isspace(*i)) + BOOST_THROW_EXCEPTION(std::exception()); +} + diff --git a/liblll/Parser.h b/liblll/Parser.h new file mode 100644 index 00000000..b21989f0 --- /dev/null +++ b/liblll/Parser.h @@ -0,0 +1,41 @@ +/* + This file is part of cpp-ethereum. + + cpp-ethereum is free software: you can redistribute it and/or modify + it under the terms of the GNU General Public License as published by + the Free Software Foundation, either version 3 of the License, or + (at your option) any later version. + + cpp-ethereum is distributed in the hope that it will be useful, + but WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + GNU General Public License for more details. + + You should have received a copy of the GNU General Public License + along with cpp-ethereum. If not, see <http://www.gnu.org/licenses/>. +*/ +/** @file Parser.h + * @author Gav Wood <i@gavwood.com> + * @date 2014 + */ + +#pragma once + +#include <string> +#include <vector> +#include <libdevcore/Common.h> + +namespace boost { namespace spirit { class utree; } } +namespace sp = boost::spirit; + +namespace dev +{ +namespace eth +{ + +void killBigints(sp::utree const& _this); +void parseTreeLLL(std::string const& _s, sp::utree& o_out); +void debugOutAST(std::ostream& _out, sp::utree const& _this); + +} +} |