Merge pull request #74 from ethereum/beginblock_instr

BEGINBLOCK intrinsic instruction

lib/evmone/analysis.cpp

@@ -17,7 +17,7 @@ bool is_terminator(uint8_t c) noexcept
 }
 }  // namespace
 
-int code_analysis::find_jumpdest(int offset) noexcept
+int code_analysis::find_jumpdest(int offset) const noexcept
 {
     // TODO: Replace with lower_bound().
     for (const auto& d : jumpdest_map)
@@ -51,6 +51,7 @@ code_analysis analyze(
     const exec_fn_table& fns, evmc_revision rev, const uint8_t* code, size_t code_size) noexcept
 {
     code_analysis analysis;
+    // TODO: Check if final result exceeds the reservation.
     analysis.instrs.reserve(code_size + 1);
 
     auto* instr_table = evmc_get_instruction_metrics_table(rev);
@@ -61,19 +62,27 @@ code_analysis analyze(
     {
         // TODO: Loop in reverse order for easier GAS analysis.
         const auto c = code[i];
-        auto& instr = analysis.instrs.emplace_back(fns[c]);
 
         const bool jumpdest = c == OP_JUMPDEST;
+
         if (!block || jumpdest)
         {
             // Create new block.
             block = &analysis.blocks.emplace_back();
-            instr.block_index = static_cast<int>(analysis.blocks.size() - 1);
 
-            if (jumpdest)
+            // Create BEGINBLOCK instruction which either replaces JUMPDEST or is injected
+            // in case there is no JUMPDEST.
+            auto& beginblock_instr = analysis.instrs.emplace_back(fns[OPX_BEGINBLOCK]);
+            beginblock_instr.arg.p.number = static_cast<int>(analysis.blocks.size() - 1);
+
+            if (jumpdest)  // Add the jumpdest to the map.
                 analysis.jumpdest_map.emplace_back(static_cast<int>(i), instr_index);
+            else  // Increase instruction count because additional BEGINBLOCK was injected.
+                ++instr_index;
         }
 
+        auto& instr = jumpdest ? analysis.instrs.back() : analysis.instrs.emplace_back(fns[c]);
+
         auto metrics = instr_table[c];
         block->gas_cost += metrics.gas_cost;
         auto stack_req = metrics.num_stack_arguments - block->stack_diff;

lib/evmone/analysis.hpp

@@ -4,6 +4,7 @@
 #pragma once
 
 #include <evmc/evmc.hpp>
+#include <evmc/instructions.h>
 #include <evmc/utils.h>
 #include <intx/intx.hpp>
 #include <array>
@@ -66,13 +67,26 @@ static_assert(sizeof(instr_argument) == sizeof(void*), "Incorrect size of instr_
 
 using exec_fn = void (*)(execution_state&, instr_argument arg);
 
+/// The evmone intrinsic opcodes.
+///
+/// These intrinsic instructions may be injected to the code in the analysis phase.
+/// They contain additional and required logic to be executed by the interpreter.
+enum intrinsic_opcodes
+{
+    /// The BEGINBLOCK instruction.
+    ///
+    /// This instruction is defined as alias for JUMPDEST and replaces all JUMPDEST instructions.
+    /// It is also injected at beginning of basic blocks not being the valid jump destination.
+    /// It checks basic block execution requirements and terminates execution if they are not met.
+    OPX_BEGINBLOCK = OP_JUMPDEST
+};
+
 using exec_fn_table = std::array<exec_fn, 256>;
 
 struct instr_info
 {
     exec_fn fn = nullptr;
     instr_argument arg;
-    int block_index = -1;
 
     explicit constexpr instr_info(exec_fn f) noexcept : fn{f}, arg{} {};
 };
@@ -99,7 +113,7 @@ struct code_analysis
     std::vector<std::pair<int, int>> jumpdest_map;
 
     // TODO: Exported for unit tests. Rework unit tests?
-    EVMC_EXPORT int find_jumpdest(int offset) noexcept;
+    EVMC_EXPORT int find_jumpdest(int offset) const noexcept;
 };
 
 EVMC_EXPORT code_analysis analyze(

lib/evmone/execution.cpp

@@ -29,31 +29,6 @@ evmc_result execute(evmc_instance*, evmc_context* ctx, evmc_revision rev, const
     while (state.run)
     {
         auto& instr = analysis.instrs[state.pc];
-        if (instr.block_index >= 0)
-        {
-            auto& block = analysis.blocks[static_cast<size_t>(instr.block_index)];
-
-            state.gas_left -= block.gas_cost;
-            if (state.gas_left < 0)
-            {
-                state.status = EVMC_OUT_OF_GAS;
-                break;
-            }
-
-            if (static_cast<int>(state.stack.size()) < block.stack_req)
-            {
-                state.status = EVMC_STACK_UNDERFLOW;
-                break;
-            }
-
-            if (static_cast<int>(state.stack.size()) + block.stack_max > 1024)
-            {
-                state.status = EVMC_STACK_OVERFLOW;
-                break;
-            }
-
-            state.current_block_cost = block.gas_cost;
-        }
 
         // Advance the PC not to allow jump opcodes to overwrite it.
         ++state.pc;

lib/evmone/instructions.cpp

@@ -8,6 +8,8 @@
 #include <evmc/helpers.hpp>
 #include <evmc/instructions.h>
 
+#include <cassert>
+
 namespace evmone
 {
 namespace
@@ -576,6 +578,9 @@ void op_jumpi(execution_state& state, instr_argument) noexcept
         state.pc = static_cast<size_t>(pc);
     }
 
+    // OPT: The pc must be the BEGINBLOCK (even in fallback case),
+    //      so we can execute it straight away.
+
     state.stack.pop_back();
     state.stack.pop_back();
 }
@@ -597,11 +602,6 @@ void op_gas(execution_state& state, instr_argument arg) noexcept
     state.stack.push_back(gas);
 }
 
-void op_jumpdest(execution_state&, instr_argument) noexcept
-{
-    // OPT: We can skip JUMPDEST instruction in analysis.
-}
-
 void op_gasprice(execution_state& state, instr_argument) noexcept
 {
     auto x = intx::be::uint256(state.host.get_tx_context().tx_gas_price.bytes);
@@ -1359,6 +1359,36 @@ void op_selfdestruct(execution_state& state, instr_argument) noexcept
     state.run = false;
 }
 
+void opx_beginblock(execution_state& state, instr_argument arg) noexcept
+{
+    assert(arg.p.number >= 0);
+    auto& block = state.analysis->blocks[static_cast<size_t>(arg.p.number)];
+
+    state.gas_left -= block.gas_cost;
+    if (state.gas_left < 0)
+    {
+        state.status = EVMC_OUT_OF_GAS;
+        state.run = false;
+        return;
+    }
+
+    if (static_cast<int>(state.stack.size()) < block.stack_req)
+    {
+        state.status = EVMC_STACK_UNDERFLOW;
+        state.run = false;
+        return;
+    }
+
+    if (static_cast<int>(state.stack.size()) + block.stack_max > 1024)
+    {
+        state.status = EVMC_STACK_OVERFLOW;
+        state.run = false;
+        return;
+    }
+
+    state.current_block_cost = block.gas_cost;
+}
+
 constexpr exec_fn_table create_op_table_frontier() noexcept
 {
     auto table = exec_fn_table{};
@@ -1421,7 +1451,7 @@ constexpr exec_fn_table create_op_table_frontier() noexcept
     table[OP_PC] = op_pc;
     table[OP_MSIZE] = op_msize;
     table[OP_GAS] = op_gas;
-    table[OP_JUMPDEST] = op_jumpdest;
+    table[OPX_BEGINBLOCK] = opx_beginblock;  // Replaces JUMPDEST.
     for (auto op = size_t{OP_PUSH1}; op <= OP_PUSH32; ++op)
         table[op] = op_push_full;
     for (auto op = size_t{OP_DUP1}; op <= OP_DUP16; ++op)

test/unittests/analysis_test.cpp

@@ -5,8 +5,11 @@
 #include <evmc/instructions.h>
 #include <evmone/analysis.hpp>
 #include <gtest/gtest.h>
+#include <test/utils/bytecode.hpp>
 #include <test/utils/utils.hpp>
 
+using namespace evmone;
+
 constexpr auto rev = EVMC_BYZANTIUM;
 
 const auto fake_fn_table = []() noexcept
@@ -21,17 +24,20 @@ const auto fake_fn_table = []() noexcept
 
 TEST(analysis, example1)
 {
-    auto code = from_hex("602a601e5359600055");
-    auto analysis = evmone::analyze(fake_fn_table, rev, &code[0], code.size());
+    const auto code = push(0x2a) + push(0x1e) + OP_MSTORE8 + OP_MSIZE + push(0) + OP_SSTORE;
+    const auto analysis = analyze(fake_fn_table, rev, &code[0], code.size());
 
-    ASSERT_EQ(analysis.instrs.size(), 7);
+    ASSERT_EQ(analysis.instrs.size(), 8);
 
-    EXPECT_EQ(analysis.instrs[0].fn, fake_fn_table[OP_PUSH1]);
+    EXPECT_EQ(analysis.instrs[0].fn, fake_fn_table[OPX_BEGINBLOCK]);
+    EXPECT_EQ(analysis.instrs[0].arg.p.number, 0);
     EXPECT_EQ(analysis.instrs[1].fn, fake_fn_table[OP_PUSH1]);
-    EXPECT_EQ(analysis.instrs[2].fn, fake_fn_table[OP_MSTORE8]);
-    EXPECT_EQ(analysis.instrs[3].fn, fake_fn_table[OP_MSIZE]);
-    EXPECT_EQ(analysis.instrs[4].fn, fake_fn_table[OP_PUSH1]);
-    EXPECT_EQ(analysis.instrs[5].fn, fake_fn_table[OP_SSTORE]);
+    EXPECT_EQ(analysis.instrs[2].fn, fake_fn_table[OP_PUSH1]);
+    EXPECT_EQ(analysis.instrs[3].fn, fake_fn_table[OP_MSTORE8]);
+    EXPECT_EQ(analysis.instrs[4].fn, fake_fn_table[OP_MSIZE]);
+    EXPECT_EQ(analysis.instrs[5].fn, fake_fn_table[OP_PUSH1]);
+    EXPECT_EQ(analysis.instrs[6].fn, fake_fn_table[OP_SSTORE]);
+    EXPECT_EQ(analysis.instrs[7].fn, fake_fn_table[OP_STOP]);
 
     ASSERT_EQ(analysis.blocks.size(), 1);
     EXPECT_EQ(analysis.blocks[0].gas_cost, 14);
@@ -42,14 +48,16 @@ TEST(analysis, example1)
 
 TEST(analysis, stack_up_and_down)
 {
-    auto code = from_hex("81808080808080505050505050505050506000");
-    auto analysis = evmone::analyze(fake_fn_table, rev, &code[0], code.size());
+    const auto code = OP_DUP2 + 6 * OP_DUP1 + 10 * OP_POP + push(0);
+    const auto analysis = analyze(fake_fn_table, rev, &code[0], code.size());
 
-    ASSERT_EQ(analysis.instrs.size(), 19);
-    EXPECT_EQ(analysis.instrs[0].fn, fake_fn_table[OP_DUP2]);
-    EXPECT_EQ(analysis.instrs[1].fn, fake_fn_table[OP_DUP1]);
-    EXPECT_EQ(analysis.instrs[7].fn, fake_fn_table[OP_POP]);
-    EXPECT_EQ(analysis.instrs[17].fn, fake_fn_table[OP_PUSH1]);
+    ASSERT_EQ(analysis.instrs.size(), 20);
+    EXPECT_EQ(analysis.instrs[0].fn, fake_fn_table[OPX_BEGINBLOCK]);
+    EXPECT_EQ(analysis.instrs[0].arg.p.number, 0);
+    EXPECT_EQ(analysis.instrs[1].fn, fake_fn_table[OP_DUP2]);
+    EXPECT_EQ(analysis.instrs[2].fn, fake_fn_table[OP_DUP1]);
+    EXPECT_EQ(analysis.instrs[8].fn, fake_fn_table[OP_POP]);
+    EXPECT_EQ(analysis.instrs[18].fn, fake_fn_table[OP_PUSH1]);
 
     ASSERT_EQ(analysis.blocks.size(), 1);
     EXPECT_EQ(analysis.blocks[0].gas_cost, 7 * 3 + 10 * 2 + 3);
@@ -60,26 +68,27 @@ TEST(analysis, stack_up_and_down)
 
 TEST(analysis, push)
 {
-    auto code = from_hex("6708070605040302017f00ee");
-    auto analysis = evmone::analyze(fake_fn_table, rev, &code[0], code.size());
+    const auto code = push(0x0807060504030201) + "7f00ee";
+    const auto analysis = analyze(fake_fn_table, rev, &code[0], code.size());
 
-    ASSERT_EQ(analysis.instrs.size(), 3);
+    ASSERT_EQ(analysis.instrs.size(), 4);
     ASSERT_EQ(analysis.args_storage.size(), 2);
-    EXPECT_EQ(analysis.instrs[0].arg.data, &analysis.args_storage[0][0]);
-    EXPECT_EQ(analysis.instrs[1].arg.data, &analysis.args_storage[1][0]);
+    EXPECT_EQ(analysis.instrs[0].fn, fake_fn_table[OPX_BEGINBLOCK]);
+    EXPECT_EQ(analysis.instrs[1].arg.data, &analysis.args_storage[0][0]);
+    EXPECT_EQ(analysis.instrs[2].arg.data, &analysis.args_storage[1][0]);
     EXPECT_EQ(analysis.args_storage[0][31 - 7], 0x08);
     EXPECT_EQ(analysis.args_storage[1][1], 0xee);
 }
 
 TEST(analysis, jump1)
 {
-    auto code = from_hex("6002600401565b600360005260206000f3600656");
-    auto analysis = evmone::analyze(fake_fn_table, rev, &code[0], code.size());
+    const auto code = jump(add(4, 2)) + OP_JUMPDEST + mstore(0, 3) + ret(0, 0x20) + jump(6);
+    const auto analysis = analyze(fake_fn_table, rev, &code[0], code.size());
 
     ASSERT_EQ(analysis.blocks.size(), 3);
     ASSERT_EQ(analysis.jumpdest_map.size(), 1);
-    EXPECT_EQ(analysis.jumpdest_map[0], std::pair(6, 4));
-    EXPECT_EQ(analysis.find_jumpdest(6), 4);
+    EXPECT_EQ(analysis.jumpdest_map[0], std::pair(6, 5));
+    EXPECT_EQ(analysis.find_jumpdest(6), 5);
     EXPECT_EQ(analysis.find_jumpdest(0), -1);
 }
 

test/unittests/execution_test.cpp

@@ -230,6 +230,13 @@ TEST_F(execution, bad_jumpdest)
     }
 }
 
+TEST_F(execution, jump_to_block_beginning)
+{
+    const auto code = jumpi(0, OP_MSIZE) + jump(4);
+    execute(code);
+    EXPECT_STATUS(EVMC_BAD_JUMP_DESTINATION);
+}
+
 TEST_F(execution, pc)
 {
     const auto code = OP_CALLDATASIZE + push(9) + OP_JUMPI + push(12) + OP_PC + OP_SWAP1 + OP_JUMP +

test/utils/bytecode.hpp

@@ -100,11 +100,26 @@ inline bytecode mstore(bytecode index)
     return index + OP_MSTORE;
 }
 
+inline bytecode mstore(bytecode index, bytecode value)
+{
+    return value + index + OP_MSTORE;
+}
+
 inline bytecode mstore8(bytecode index)
 {
     return index + OP_MSTORE8;
 }
 
+inline bytecode jump(bytecode target)
+{
+    return target + OP_JUMP;
+}
+
+inline bytecode jumpi(bytecode target, bytecode condition)
+{
+    return condition + target + OP_JUMPI;
+}
+
 inline bytecode ret(bytecode index, bytecode size)
 {
     return size + index + OP_RETURN;

test/utils/dump.cpp

@@ -26,10 +26,9 @@ void dump_analysis(const evmone::code_analysis& analysis)
         if (!name)
             name = "XX";
 
-
-        if (instr.block_index >= 0)
+        if (c == OPX_BEGINBLOCK)
         {
-            block = &analysis.blocks[size_t(instr.block_index)];
+            block = &analysis.blocks[size_t(instr.arg.p.number)];
 
             auto get_jumpdest_offset = [&analysis](size_t index) noexcept
             {