Implement DATALOAD instruction

Author: gumb0

lib/evmone/advanced_analysis.hpp

@@ -90,9 +90,9 @@ struct AdvancedExecutionState : ExecutionState
     AdvancedExecutionState() noexcept : stack{stack_space.bottom()} {}
 
     AdvancedExecutionState(const evmc_message& message, evmc_revision revision,
-        const evmc_host_interface& host_interface, evmc_host_context* host_ctx,
-        bytes_view _code) noexcept
-      : ExecutionState{message, revision, host_interface, host_ctx, _code},
+        const evmc_host_interface& host_interface, evmc_host_context* host_ctx, bytes_view _code,
+        bytes_view _data) noexcept
+      : ExecutionState{message, revision, host_interface, host_ctx, _code, _data},
         stack{stack_space.bottom()}
     {}
 

lib/evmone/advanced_execution.cpp

@@ -45,7 +45,8 @@ evmc_result execute(evmc_vm* /*unused*/, const evmc_host_interface* host, evmc_h
     }
     else
         analysis = analyze(rev, container);
-    auto state = std::make_unique<AdvancedExecutionState>(*msg, rev, *host, ctx, container);
+    auto state =
+        std::make_unique<AdvancedExecutionState>(*msg, rev, *host, ctx, container, bytes_view{});
     return execute(*state, analysis);
 }
 }  // namespace evmone::advanced

lib/evmone/baseline.cpp

@@ -87,9 +87,11 @@ CodeAnalysis analyze_eof1(bytes_view container)
     for (const auto offset : header.code_offsets)
         relative_offsets.push_back(offset - code_sections_offset);
 
+    const auto data_section = container.substr(code_sections_end);
+
     // FIXME: Better way of getting EOF version.
     const auto eof_version = container[2];
-    return CodeAnalysis{executable_code, {}, eof_version, relative_offsets};
+    return CodeAnalysis{executable_code, {}, eof_version, relative_offsets, data_section};
 }
 }  // namespace
 
@@ -375,8 +377,8 @@ evmc_result execute(evmc_vm* c_vm, const evmc_host_interface* host, evmc_host_co
 {
     auto vm = static_cast<VM*>(c_vm);
     const auto jumpdest_map = analyze(rev, {code, code_size});
-    auto state =
-        std::make_unique<ExecutionState>(*msg, rev, *host, ctx, bytes_view{code, code_size});
+    auto state = std::make_unique<ExecutionState>(
+        *msg, rev, *host, ctx, bytes_view{code, code_size}, jumpdest_map.data);
     return execute(*vm, *state, jumpdest_map);
 }
 }  // namespace evmone::baseline

lib/evmone/baseline.hpp

@@ -31,6 +31,8 @@ class CodeAnalysis
     /// section. We flatten the sections for cheap execution.
     CodeOffsets code_offsets;
 
+    bytes_view data;  ///< Data section.
+
 private:
     /// Padded code for faster legacy code execution.
     /// If not nullptr the executable_code must point to it.
@@ -43,11 +45,13 @@ class CodeAnalysis
         m_padded_code{std::move(padded_code)}
     {}
 
-    CodeAnalysis(bytes_view code, JumpdestMap map, uint8_t version, CodeOffsets offsets)
+    CodeAnalysis(bytes_view code, JumpdestMap map, uint8_t version, CodeOffsets offsets,
+        bytes_view data_section)
       : executable_code{code},
         jumpdest_map{std::move(map)},
         eof_version{version},
-        code_offsets{std::move(offsets)}
+        code_offsets{std::move(offsets)},
+        data{data_section}
     {}
 };
 static_assert(std::is_move_constructible_v<CodeAnalysis>);

lib/evmone/execution_state.hpp

@@ -138,6 +138,8 @@ class ExecutionState
     /// For EOF-formatted code this is a reference to entire container.
     bytes_view original_code;
 
+    bytes_view data;
+
     evmc_status_code status = EVMC_SUCCESS;
     size_t output_offset = 0;
     size_t output_size = 0;
@@ -164,13 +166,14 @@ class ExecutionState
     ExecutionState() noexcept = default;
 
     ExecutionState(const evmc_message& message, evmc_revision revision,
-        const evmc_host_interface& host_interface, evmc_host_context* host_ctx,
-        bytes_view _code) noexcept
+        const evmc_host_interface& host_interface, evmc_host_context* host_ctx, bytes_view _code,
+        bytes_view _data) noexcept
       : gas_left{message.gas},
         msg{&message},
         host{host_interface, host_ctx},
         rev{revision},
-        original_code{_code}
+        original_code{_code},
+        data{_data}
     {}
 
     /// Resets the contents of the ExecutionState so that it could be reused.

lib/evmone/instructions.hpp

@@ -890,6 +890,31 @@ inline code_iterator swapn(StackTop stack, ExecutionState& state, code_iterator
     return pos + 2;
 }
 
+inline evmc_status_code dataload(StackTop stack, ExecutionState& state) noexcept
+{
+    auto& index = stack.top();
+
+    if (state.msg->input_size < index)
+        index = 0;
+    else
+    {
+        if (state.data.size() < index)
+            return EVMC_INVALID_MEMORY_ACCESS; // TODO better error
+        const auto begin = static_cast<size_t>(index);
+
+        const auto end = begin + 32;
+        if (end > state.data.size())
+            return EVMC_INVALID_MEMORY_ACCESS; // TODO better error
+
+        uint8_t data[32] = {};
+        for (size_t i = 0; i < 32; ++i)
+            data[i] = state.data[begin + i];
+
+        index = intx::be::load<uint256>(data);
+    }
+    return EVMC_SUCCESS;
+}
+
 template <size_t NumTopics>
 inline evmc_status_code log(StackTop stack, ExecutionState& state) noexcept
 {

lib/evmone/instructions_opcodes.hpp

@@ -163,6 +163,8 @@ enum Opcode : uint8_t
     OP_DUPN = 0xb5,
     OP_SWAPN = 0xb6,
 
+    OP_DATALOAD = 0xb7,
+
     OP_CREATE = 0xf0,
     OP_CALL = 0xf1,
     OP_CALLCODE = 0xf2,

lib/evmone/instructions_traits.hpp

@@ -171,6 +171,7 @@ constexpr inline GasCostTable gas_costs = []() noexcept {
     table[EVMC_CANCUN][OP_RJUMPV] = 4;
     table[EVMC_CANCUN][OP_CALLF] = 5;
     table[EVMC_CANCUN][OP_RETF] = 3;
+    table[EVMC_CANCUN][OP_DATALOAD] = 3;
 
     table[EVMC_PRAGUE] = table[EVMC_CANCUN];
 
@@ -375,6 +376,7 @@ constexpr inline std::array<Traits, 256> traits = []() noexcept {
 
     table[OP_DUPN] = {"DUPN", 1, false, 0, 1, EVMC_CANCUN};
     table[OP_SWAPN] = {"SWAPN", 1, false, 0, 0, EVMC_CANCUN};
+    table[OP_DATALOAD] = {"DATALOAD", 0, false, 1, 0, EVMC_CANCUN};
 
     table[OP_CREATE] = {"CREATE", 0, false, 3, -2, EVMC_FRONTIER};
     table[OP_CALL] = {"CALL", 0, false, 7, -6, EVMC_FRONTIER};

lib/evmone/instructions_xmacro.hpp

@@ -226,7 +226,7 @@
     ON_OPCODE_UNDEFINED(0xb4)                               \
     ON_OPCODE_IDENTIFIER(OP_DUPN, dupn)                     \
     ON_OPCODE_IDENTIFIER(OP_SWAPN, swapn)                   \
-    ON_OPCODE_UNDEFINED(0xb7)                               \
+    ON_OPCODE_IDENTIFIER(OP_DATALOAD, dataload)             \
     ON_OPCODE_UNDEFINED(0xb8)                               \
     ON_OPCODE_UNDEFINED(0xb9)                               \
     ON_OPCODE_UNDEFINED(0xba)                               \

test/unittests/execution_state_test.cpp

@@ -26,7 +26,7 @@ TEST(execution_state, construct)
     const evmc_host_interface host_interface{};
     const uint8_t code[]{0x0f};
     const evmone::ExecutionState st{
-        msg, EVMC_MAX_REVISION, host_interface, nullptr, {code, std::size(code)}};
+        msg, EVMC_MAX_REVISION, host_interface, nullptr, {code, std::size(code)}, {}};
 
     EXPECT_EQ(st.gas_left, -1);
     EXPECT_EQ(st.memory.size(), 0);

test/unittests/instructions_test.cpp

@@ -110,6 +110,7 @@ constexpr bool instruction_only_in_evmone(evmc_revision rev, Opcode op) noexcept
     case OP_RETF:
     case OP_DUPN:
     case OP_SWAPN:
+    case OP_DATALOAD:
         return true;
     default:
         return false;