/* This file is part of solidity. solidity 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. solidity 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 solidity. If not, see . */ /** * Translates Yul code from EVM dialect to eWasm dialect. */ #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include using namespace std; using namespace dev; using namespace yul; using namespace langutil; namespace { static string const polyfill{R"({ function or_bool(a, b, c, d) -> r { r := i64.or(i64.or(a, b), i64.or(c, d)) } // returns a + y + c plus carry value on 64 bit values. // c should be at most 1 function add_carry(x, y, c) -> r, r_c { let t := i64.add(x, y) r := i64.add(t, c) r_c := i64.or( i64.lt_u(t, x), i64.lt_u(r, t) ) } function add(x1, x2, x3, x4, y1, y2, y3, y4) -> r1, r2, r3, r4 { let carry r4, carry := add_carry(x4, y4, 0) r3, carry := add_carry(x3, y3, carry) r2, carry := add_carry(x2, y2, carry) r1, carry := add_carry(x1, y1, carry) } function bit_negate(x) -> y { y := i64.xor(x, 0xffffffffffffffff) } function sub(x1, x2, x3, x4, y1, y2, y3, y4) -> r1, r2, r3, r4 { // x - y = x + (~y + 1) let carry r4, carry := add_carry(x4, bit_negate(y4), 1) r3, carry := add_carry(x3, bit_negate(y3), carry) r2, carry := add_carry(x2, bit_negate(y2), carry) r1, carry := add_carry(x1, bit_negate(y1), carry) } function split(x) -> hi, lo { hi := i64.shr_u(x, 32) lo := i64.and(x, 0xffffffff) } // Multiplies two 64 bit values resulting in a 128 bit // value split into two 64 bit values. function mul_64x64_128(x, y) -> hi, lo { let xh, xl := split(x) let yh, yl := split(y) let t0 := i64.mul(xl, yl) let t1 := i64.mul(xh, yl) let t2 := i64.mul(xl, yh) let t3 := i64.mul(xh, yh) let t0h, t0l := split(t0) let u1 := i64.add(t1, t0h) let u1h, u1l := split(u1) let u2 := i64.add(t2, u1l) lo := i64.or(i64.shl(u2, 32), t0l) hi := i64.add(t3, i64.add(i64.shr_u(u2, 32), u1h)) } // Multiplies two 128 bit values resulting in a 256 bit // value split into four 64 bit values. function mul_128x128_256(x1, x2, y1, y2) -> r1, r2, r3, r4 { let ah, al := mul_64x64_128(x1, y1) let bh, bl := mul_64x64_128(x1, y2) let ch, cl := mul_64x64_128(x2, y1) let dh, dl := mul_64x64_128(x2, y2) r4 := dl let carry1, carry2 let t1, t2 r3, carry1 := add_carry(bl, cl, 0) r3, carry2 := add_carry(r3, dh, 0) t1, carry1 := add_carry(bh, ch, carry1) r2, carry2 := add_carry(t1, al, carry2) r1 := i64.add(i64.add(ah, carry1), carry2) } function mul(x1, x2, x3, x4, y1, y2, y3, y4) -> r1, r2, r3, r4 { // TODO it would actually suffice to have mul_128x128_128 for the first two. let b1, b2, b3, b4 := mul_128x128_256(x3, x4, y1, y2) let c1, c2, c3, c4 := mul_128x128_256(x1, x2, y3, y4) let d1, d2, d3, d4 := mul_128x128_256(x3, x4, y3, y4) r4 := d4 r3 := d3 let t1, t2 t1, t2, r1, r2 := add(0, 0, b3, b4, 0, 0, c3, c4) t1, t2, r1, r2 := add(0, 0, r1, r2, 0, 0, d1, d2) } function div(x1, x2, x3, x4, y1, y2, y3, y4) -> r1, r2, r3, r4 { // TODO implement properly r4 := i64.div_u(x4, y4) } function sdiv(x1, x2, x3, x4, y1, y2, y3, y4) -> r1, r2, r3, r4 { // TODO implement properly unreachable() } function mod(x1, x2, x3, x4, y1, y2, y3, y4) -> r1, r2, r3, r4 { // TODO implement properly r4 := i64.rem_u(x4, y4) } function smod(x1, x2, x3, x4, y1, y2, y3, y4) -> r1, r2, r3, r4 { // TODO implement properly r4 := i64.rem_u(x4, y4) } function exp(x1, x2, x3, x4, y1, y2, y3, y4) -> r1, r2, r3, r4 { // TODO implement properly unreachable() } function byte(x1, x2, x3, x4, y1, y2, y3, y4) -> r1, r2, r3, r4 { if i64.eqz(i64.or(i64.or(x1, x2), x3)) { let component switch i64.div_u(x4, 8) case 0 { component := y1 } case 1 { component := y2 } case 2 { component := y3 } case 3 { component := y4 } x4 := i64.mul(i64.rem_u(x4, 8), 8) r4 := i64.shr_u(component, i64.sub(56, x4)) r4 := i64.and(0xff, r4) } } function xor(x1, x2, x3, x4, y1, y2, y3, y4) -> r1, r2, r3, r4 { r1 := i64.xor(x1, y1) r2 := i64.xor(x2, y2) r3 := i64.xor(x3, y3) r4 := i64.xor(x4, y4) } function or(x1, x2, x3, x4, y1, y2, y3, y4) -> r1, r2, r3, r4 { r1 := i64.or(x1, y1) r2 := i64.or(x2, y2) r3 := i64.or(x3, y3) r4 := i64.or(x4, y4) } function and(x1, x2, x3, x4, y1, y2, y3, y4) -> r1, r2, r3, r4 { r1 := i64.and(x1, y1) r2 := i64.and(x2, y2) r3 := i64.and(x3, y3) r4 := i64.and(x4, y4) } function not(x1, x2, x3, x4) -> r1, r2, r3, r4 { let mask := 0xffffffffffffffff r1, r2, r3, r4 := xor(x1, x2, x3, x4, mask, mask, mask, mask) } function iszero(x1, x2, x3, x4) -> r1, r2, r3, r4 { r4 := i64.eqz(i64.or(i64.or(x1, x2), i64.or(x3, x4))) } function eq(x1, x2, x3, x4, y1, y2, y3, y4) -> r1, r2, r3, r4 { if i64.eq(x1, y1) { if i64.eq(x2, y2) { if i64.eq(x3, y3) { if i64.eq(x4, y4) { r4 := 1 } } } } } // returns 0 if a == b, -1 if a < b and 1 if a > b function cmp(a, b) -> r { switch i64.lt_u(a, b) case 1 { r := 0xffffffffffffffff } default { r := i64.ne(a, b) } } function lt(x1, x2, x3, x4, y1, y2, y3, y4) -> z1, z2, z3, z4 { switch cmp(x1, y1) case 0 { switch cmp(x2, y2) case 0 { switch cmp(x3, y3) case 0 { z4 := i64.lt_u(x4, y4) } case 1 { z4 := 0 } default { z4 := 1 } } case 1 { z4 := 0 } default { z4 := 1 } } case 1 { z4 := 0 } default { z4 := 1 } } function gt(x1, x2, x3, x4, y1, y2, y3, y4) -> z1, z2, z3, z4 { z1, z2, z3, z4 := lt(y1, y2, y3, y4, x1, x2, x3, x4) } function slt(x1, x2, x3, x4, y1, y2, y3, y4) -> z1, z2, z3, z4 { // TODO correct? x1 := i64.add(x1, 0x8000000000000000) y1 := i64.add(y1, 0x8000000000000000) z1, z2, z3, z4 := lt(x1, x2, x3, x4, y1, y2, y3, y4) } function sgt(x1, x2, x3, x4, y1, y2, y3, y4) -> z1, z2, z3, z4 { z1, z2, z3, z4 := slt(y1, y2, y3, y4, x1, x2, x3, x4) } function shl_single(a, amount) -> x, y { // amount < 64 x := i64.shr_u(a, i64.sub(64, amount)) y := i64.shl(a, amount) } function shl(x1, x2, x3, x4, y1, y2, y3, y4) -> z1, z2, z3, z4 { if i64.and(i64.eqz(x1), i64.eqz(x2)) { if i64.eqz(x3) { if i64.lt_u(x4, 256) { if i64.ge_u(x4, 128) { y1 := y3 y2 := y4 y3 := 0 y4 := 0 x4 := i64.sub(x4, 128) } if i64.ge_u(x4, 64) { y1 := y2 y2 := y3 y3 := y4 y4 := 0 x4 := i64.sub(x4, 64) } let t, r t, z4 := shl_single(y4, x4) r, z3 := shl_single(y3, x4) z3 := i64.or(z3, t) t, z2 := shl_single(y2, x4) z2 := i64.or(z2, r) r, z1 := shl_single(y1, x4) z1 := i64.or(z1, t) } } } } function shr_single(a, amount) -> x, y { // amount < 64 y := i64.shl(a, i64.sub(64, amount)) x := i64.shr_u(a, amount) } function shr(x1, x2, x3, x4, y1, y2, y3, y4) -> z1, z2, z3, z4 { if i64.and(i64.eqz(x1), i64.eqz(x2)) { if i64.eqz(x3) { if i64.lt_u(x4, 256) { if i64.ge_u(x4, 128) { y4 := y2 y3 := y1 y2 := 0 y1 := 0 x4 := i64.sub(x4, 128) } if i64.ge_u(x4, 64) { y4 := y3 y3 := y2 y2 := y1 y1 := 0 x4 := i64.sub(x4, 64) } let t z4, t := shr_single(y4, x4) z3, t := shr_single(y3, x4) z4 := i64.or(z4, t) z2, t := shr_single(y2, x4) z3 := i64.or(z3, t) z1, t := shr_single(y1, x4) z2 := i64.or(z2, t) } } } } function sar(x1, x2, x3, x4, y1, y2, y3, y4) -> z1, z2, z3, z4 { // TODO implement unreachable() } function addmod(x1, x2, x3, x4, y1, y2, y3, y4, m1, m2, m3, m4) -> z1, z2, z3, z4 { // TODO implement unreachable() } function mulmod(x1, x2, x3, x4, y1, y2, y3, y4, m1, m2, m3, m4) -> z1, z2, z3, z4 { // TODO implement unreachable() } function signextend(x1, x2, x3, x4, y1, y2, y3, y4) -> z1, z2, z3, z4 { // TODO implement unreachable() } function u256_to_i64(x1, x2, x3, x4) -> v { if i64.ne(0, i64.or(i64.or(x1, x2), x3)) { invalid() } v := x4 } function u256_to_i32(x1, x2, x3, x4) -> v { if i64.ne(0, i64.or(i64.or(x1, x2), x3)) { invalid() } if i64.ne(0, i64.shr_u(x4, 32)) { invalid() } v := x4 } function u256_to_i32ptr(x1, x2, x3, x4) -> v { v := u256_to_i32(x1, x2, x3, x4) } function keccak256(x1, x2, x3, x4, y1, y2, y3, y4) -> z1, z2, z3, z4 { // TODO implement unreachable() } function address() -> z1, z2, z3, z4 { eth.getAddress(0) z1, z2, z3, z4 := mload_internal(0) } function balance(x1, x2, x3, x4) -> z1, z2, z3, z4 { // TODO implement unreachable() } function selfbalance() -> z1, z2, z3, z4 { // TODO: not part of current Ewasm spec unreachable() } function chainid() -> z1, z2, z3, z4 { // TODO: not part of current Ewasm spec unreachable() } function origin() -> z1, z2, z3, z4 { eth.getTxOrigin(0) z1, z2, z3, z4 := mload_internal(0) } function caller() -> z1, z2, z3, z4 { eth.getCaller(0) z1, z2, z3, z4 := mload_internal(0) } function callvalue() -> z1, z2, z3, z4 { eth.getCallValue(0) z1, z2, z3, z4 := mload_internal(0) } function calldataload(x1, x2, x3, x4) -> z1, z2, z3, z4 { eth.callDataCopy(0, u256_to_i32(x1, x2, x3, x4), 32) z1, z2, z3, z4 := mload_internal(0) } function calldatasize() -> z1, z2, z3, z4 { z4 := eth.getCallDataSize() } function calldatacopy(x1, x2, x3, x4, y1, y2, y3, y4, z1, z2, z3, z4) { eth.callDataCopy( // scratch - TODO: overflow check i64.add(u256_to_i32ptr(x1, x2, x3, x4), 64), u256_to_i32(y1, y2, y3, y4), u256_to_i32(z1, z2, z3, z4) ) } // Needed? function codesize() -> z1, z2, z3, z4 { eth.getCodeSize(0) z1, z2, z3, z4 := mload_internal(0) } function codecopy(x1, x2, x3, x4, y1, y2, y3, y4, z1, z2, z3, z4) { eth.codeCopy( // scratch - TODO: overflow check i64.add(u256_to_i32ptr(x1, x2, x3, x4), 64), u256_to_i32(y1, y2, y3, y4), u256_to_i32(z1, z2, z3, z4) ) } function datacopy(x1, x2, x3, x4, y1, y2, y3, y4, z1, z2, z3, z4) { // TODO correct? codecopy(x1, x2, x3, x4, y1, y2, y3, y4, z1, z2, z3, z4) } function gasprice() -> z1, z2, z3, z4 { eth.getTxGasPrice(0) z1, z2, z3, z4 := mload_internal(0) } function extcodesize(x1, x2, x3, x4) -> z1, z2, z3, z4 { // TODO implement unreachable() } function extcodehash(x1, x2, x3, x4) -> z1, z2, z3, z4 { // TODO: not part of current Ewasm spec unreachable() } function extcodecopy(v1, v2, v3, v4, x1, x2, x3, x4, y1, y2, y3, y4, z1, z2, z3, z4) { // TODO implement unreachable() } function returndatasize() -> z1, z2, z3, z4 { z4 := eth.getReturnDataSize() } function returndatacopy(x1, x2, x3, x4, y1, y2, y3, y4, z1, z2, z3, z4) { eth.returnDataCopy( // scratch - TODO: overflow check i64.add(u256_to_i32ptr(x1, x2, x3, x4), 64), u256_to_i32(y1, y2, y3, y4), u256_to_i32(z1, z2, z3, z4) ) } function blockhash(x1, x2, x3, x4) -> z1, z2, z3, z4 { // TODO implement unreachable() } function coinbase() -> z1, z2, z3, z4 { // TODO implement unreachable() } function timestamp() -> z1, z2, z3, z4 { z4 := eth.getBlockTimestamp() } function number() -> z1, z2, z3, z4 { z4 := eth.getBlockNumber() } function difficulty() -> z1, z2, z3, z4 { eth.getBlockDifficulty(0) z1, z2, z3, z4 := mload_internal(0) } function gaslimit() -> z1, z2, z3, z4 { z4 := eth.getBlockGasLimit() } function pop(x1, x2, x3, x4) { } function endian_swap_16(x) -> y { let hi := i64.and(i64.shl(x, 8), 0xff00) let lo := i64.and(i64.shr_u(x, 8), 0xff) y := i64.or(hi, lo) } function endian_swap_32(x) -> y { let hi := i64.shl(endian_swap_16(x), 16) let lo := endian_swap_16(i64.shr_u(x, 16)) y := i64.or(hi, lo) } function endian_swap(x) -> y { let hi := i64.shl(endian_swap_32(x), 32) let lo := endian_swap_32(i64.shr_u(x, 32)) y := i64.or(hi, lo) } function save_temp_mem_32() -> t1, t2, t3, t4 { t1 := i64.load(0) t2 := i64.load(8) t3 := i64.load(16) t4 := i64.load(24) } function restore_temp_mem_32(t1, t2, t3, t4) { i64.store(0, t1) i64.store(8, t2) i64.store(16, t3) i64.store(24, t4) } function save_temp_mem_64() -> t1, t2, t3, t4, t5, t6, t7, t8 { t1 := i64.load(0) t2 := i64.load(8) t3 := i64.load(16) t4 := i64.load(24) t5 := i64.load(32) t6 := i64.load(40) t7 := i64.load(48) t8 := i64.load(54) } function restore_temp_mem_64(t1, t2, t3, t4, t5, t6, t7, t8) { i64.store(0, t1) i64.store(8, t2) i64.store(16, t3) i64.store(24, t4) i64.store(32, t5) i64.store(40, t6) i64.store(48, t7) i64.store(54, t8) } function mload(x1, x2, x3, x4) -> z1, z2, z3, z4 { let pos := u256_to_i32ptr(x1, x2, x3, x4) // Make room for the scratch space // TODO do we need to check for overflow? pos := i64.add(pos, 64) z1, z2, z3, z4 := mload_internal(pos) } function mload_internal(pos) -> z1, z2, z3, z4 { z1 := endian_swap(i64.load(pos)) z2 := endian_swap(i64.load(i64.add(pos, 8))) z3 := endian_swap(i64.load(i64.add(pos, 16))) z4 := endian_swap(i64.load(i64.add(pos, 24))) } function mstore(x1, x2, x3, x4, y1, y2, y3, y4) { let pos := u256_to_i32ptr(x1, x2, x3, x4) // Make room for the scratch space // TODO do we need to check for overflow? pos := i64.add(pos, 64) mstore_internal(pos, y1, y2, y3, y4) } function mstore_internal(pos, y1, y2, y3, y4) { i64.store(pos, endian_swap(y1)) i64.store(i64.add(pos, 8), endian_swap(y2)) i64.store(i64.add(pos, 16), endian_swap(y3)) i64.store(i64.add(pos, 24), endian_swap(y4)) } function mstore8(x1, x2, x3, x4, y1, y2, y3, y4) { // TODO implement unreachable() } // Needed? function msize() -> z1, z2, z3, z4 { // TODO implement unreachable() } function sload(x1, x2, x3, x4) -> z1, z2, z3, z4 { mstore_internal(0, x1, x2, x3, x4) eth.storageLoad(0, 32) z1, z2, z3, z4 := mload_internal(32) } function sstore(x1, x2, x3, x4, y1, y2, y3, y4) { mstore_internal(0, x1, x2, x3, x4) mstore_internal(32, y1, y2, y3, y4) eth.storageStore(0, 32) } // Needed? function pc() -> z1, z2, z3, z4 { // TODO implement unreachable() } function gas() -> z1, z2, z3, z4 { z4 := eth.getGasLeft() } function log0(p1, p2, p3, p4, s1, s2, s3, s4) { // TODO implement unreachable() } function log1( p1, p2, p3, p4, s1, s2, s3, s4, t11, t12, t13, t14 ) { // TODO implement unreachable() } function log2( p1, p2, p3, p4, s1, s2, s3, s4, t11, t12, t13, t14, t21, t22, t23, t24 ) { // TODO implement unreachable() } function log3( p1, p2, p3, p4, s1, s2, s3, s4, t11, t12, t13, t14, t21, t22, t23, t24, t31, t32, t33, t34 ) { // TODO implement unreachable() } function log4( p1, p2, p3, p4, s1, s2, s3, s4, t11, t12, t13, t14, t21, t22, t23, t24, t31, t32, t33, t34, t41, t42, t43, t44, ) { // TODO implement unreachable() } function create(x1, x2, x3, x4, y1, y2, y3, y4, z1, z2, z3, z4) -> a1, a2, a3, a4 { // TODO implement unreachable() } function call( a1, a2, a3, a4, b1, b2, b3, b4, c1, c2, c3, c4, d1, d2, d3, d4, e1, e2, e3, e4, f1, f2, f3, f4, g1, g2, g3, g4 ) -> x1, x2, x3, x4 { // TODO implement unreachable() } function callcode( a1, a2, a3, a4, b1, b2, b3, b4, c1, c2, c3, c4, d1, d2, d3, d4, e1, e2, e3, e4, f1, f2, f3, f4, g1, g2, g3, g4 ) -> x1, x2, x3, x4 { // TODO implement unreachable() } function delegatecall( a1, a2, a3, a4, b1, b2, b3, b4, c1, c2, c3, c4, d1, d2, d3, d4, e1, e2, e3, e4, f1, f2, f3, f4 ) -> x1, x2, x3, x4 { // TODO implement unreachable() } function staticcall( a1, a2, a3, a4, b1, b2, b3, b4, c1, c2, c3, c4, d1, d2, d3, d4, e1, e2, e3, e4, f1, f2, f3, f4 ) -> x1, x2, x3, x4 { // TODO implement unreachable() } function create2( a1, a2, a3, a4, b1, b2, b3, b4, c1, c2, c3, c4, d1, d2, d3, d4 ) -> x1, x2, x3, x4 { // TODO: not part of current Ewasm spec unreachable() } function selfdestruct(a1, a2, a3, a4) { mstore(0, 0, 0, 0, a1, a2, a3, a4) // In EVM, addresses are padded to 32 bytes, so discard the first 12. eth.selfDestruct(12) } function return(x1, x2, x3, x4, y1, y2, y3, y4) { eth.finish( // scratch - TODO: overflow check i64.add(u256_to_i32ptr(x1, x2, x3, x4), 64), u256_to_i32(y1, y2, y3, y4) ) } function revert(x1, x2, x3, x4, y1, y2, y3, y4) { eth.revert( // scratch - TODO: overflow check i64.add(u256_to_i32ptr(x1, x2, x3, x4), 64), u256_to_i32(y1, y2, y3, y4) ) } function invalid() { unreachable() } })"}; } Object EVMToEWasmTranslator::run(Object const& _object) { if (!m_polyfill) parsePolyfill(); Block ast = std::get(Disambiguator(m_dialect, *_object.analysisInfo)(*_object.code)); set reservedIdentifiers; NameDispenser nameDispenser{m_dialect, ast, reservedIdentifiers}; OptimiserStepContext context{m_dialect, nameDispenser, reservedIdentifiers}; FunctionHoister::run(context, ast); FunctionGrouper::run(context, ast); MainFunction{}(ast); ForLoopConditionIntoBody::run(context, ast); ExpressionSplitter::run(context, ast); WordSizeTransform::run(m_dialect, ast, nameDispenser); NameDisplacer{nameDispenser, m_polyfillFunctions}(ast); for (auto const& st: m_polyfill->statements) ast.statements.emplace_back(ASTCopier{}.translate(st)); Object ret; ret.name = _object.name; ret.code = make_shared(move(ast)); ret.analysisInfo = make_shared(); ErrorList errors; ErrorReporter errorReporter(errors); AsmAnalyzer analyzer(*ret.analysisInfo, errorReporter, WasmDialect::instance(), {}, _object.dataNames()); if (!analyzer.analyze(*ret.code)) { // TODO the errors here are "wrong" because they have invalid source references! string message; for (auto const& err: errors) message += langutil::SourceReferenceFormatter::formatErrorInformation(*err); yulAssert(false, message); } for (auto const& subObjectNode: _object.subObjects) if (Object const* subObject = dynamic_cast(subObjectNode.get())) ret.subObjects.push_back(make_shared(run(*subObject))); else ret.subObjects.push_back(make_shared(dynamic_cast(*subObjectNode))); ret.subIndexByName = _object.subIndexByName; return ret; } void EVMToEWasmTranslator::parsePolyfill() { ErrorList errors; ErrorReporter errorReporter(errors); shared_ptr scanner{make_shared(CharStream(polyfill, ""))}; m_polyfill = Parser(errorReporter, WasmDialect::instance()).parse(scanner, false); if (!errors.empty()) { string message; for (auto const& err: errors) message += langutil::SourceReferenceFormatter::formatErrorInformation(*err); yulAssert(false, message); } m_polyfillFunctions.clear(); for (auto const& statement: m_polyfill->statements) m_polyfillFunctions.insert(std::get(statement).name); }