wasm: Simplify the parsed instruction format

This reduces the memory usage of the instructions by a lot and simplifies the code that works with them.
robinlinden · Jan 9, 2025 · b27749a · b27749a
1 parent 1696c9f
commit b27749a
Show file tree

Hide file tree

Showing 10 changed files with 168 additions and 186 deletions.
diff --git a/wasm/byte_code_parser.cpp b/wasm/byte_code_parser.cpp
@@ -661,11 +661,13 @@ tl::expected<Module, ModuleParseError> ByteCodeParser::parse_module(std::istream
     return module;
 }
 
-// NOLINTNEXTLINE(misc-no-recursion)
 std::optional<std::vector<instructions::Instruction>> ByteCodeParser::parse_instructions(std::istream &is) {
     using namespace instructions;
     std::vector<Instruction> instructions{};
 
+    // If an End-opcode is encountered when nesting == 0, we're done.
+    int nesting = 0;
+
     while (true) {
         std::uint8_t opcode{};
         if (!is.read(reinterpret_cast<char *>(&opcode), sizeof(opcode))) {
@@ -679,12 +681,8 @@ std::optional<std::vector<instructions::Instruction>> ByteCodeParser::parse_inst
                     return std::nullopt;
                 }
 
-                auto block_instructions = parse_instructions(is);
-                if (!block_instructions) {
-                    return std::nullopt;
-                }
-
-                instructions.emplace_back(Block{*std::move(type), *std::move(block_instructions)});
+                instructions.emplace_back(Block{*std::move(type)});
+                ++nesting;
                 break;
             }
             case Loop::kOpcode: {
@@ -693,12 +691,8 @@ std::optional<std::vector<instructions::Instruction>> ByteCodeParser::parse_inst
                     return std::nullopt;
                 }
 
-                auto block_instructions = parse_instructions(is);
-                if (!block_instructions) {
-                    return std::nullopt;
-                }
-
-                instructions.emplace_back(Loop{*std::move(type), *std::move(block_instructions)});
+                instructions.emplace_back(Loop{*std::move(type)});
+                ++nesting;
                 break;
             }
             case Branch::kOpcode: {
@@ -721,7 +715,13 @@ std::optional<std::vector<instructions::Instruction>> ByteCodeParser::parse_inst
                 instructions.emplace_back(Return{});
                 break;
             case End::kOpcode:
-                return instructions;
+                instructions.emplace_back(End{});
+                if (nesting == 0) {
+                    return instructions;
+                }
+
+                --nesting;
+                break;
             case I32Const::kOpcode: {
                 auto value = wasm::Leb128<std::int32_t>::decode_from(is);
                 if (!value) {

diff --git a/wasm/byte_code_parser_test.cpp b/wasm/byte_code_parser_test.cpp
@@ -411,7 +411,7 @@ void global_section_tests(etest::Suite &s) {
         a.expect_eq(module.global_section,
                 wasm::GlobalSection{.globals{{
                         .type{wasm::ValueType::Int32, wasm::GlobalType::Mutability::Const},
-                        .init{wasm::instructions::I32Const{42}},
+                        .init{wasm::instructions::I32Const{42}, wasm::instructions::End{}},
                 }}});
     });
 
@@ -421,7 +421,7 @@ void global_section_tests(etest::Suite &s) {
         a.expect_eq(module.global_section,
                 wasm::GlobalSection{.globals{{
                         .type{wasm::ValueType::Int32, wasm::GlobalType::Mutability::Var},
-                        .init{wasm::instructions::I32Const{42}},
+                        .init{wasm::instructions::I32Const{42}, wasm::instructions::End{}},
                 }}});
     });
 
@@ -433,11 +433,11 @@ void global_section_tests(etest::Suite &s) {
                 wasm::GlobalSection{.globals{
                         {
                                 .type{wasm::ValueType::Int32, wasm::GlobalType::Mutability::Var},
-                                .init{wasm::instructions::I32Const{42}},
+                                .init{wasm::instructions::I32Const{42}, wasm::instructions::End{}},
                         },
                         {
                                 .type{wasm::ValueType::Int32, wasm::GlobalType::Mutability::Const},
-                                .init{wasm::instructions::I32Const{42}},
+                                .init{wasm::instructions::I32Const{42}, wasm::instructions::End{}},
                         },
                 }});
     });
@@ -633,7 +633,9 @@ void code_section_tests(etest::Suite &s) {
 
         wasm::CodeSection expected{.entries{
                 wasm::CodeEntry{
-                        .code{wasm::instructions::I32Const{0b11}, wasm::instructions::I32PopulationCount{}},
+                        .code{wasm::instructions::I32Const{0b11},
+                                wasm::instructions::I32PopulationCount{},
+                                wasm::instructions::End{}},
                         .locals{{1, wasm::ValueType::Int32}},
                 },
         }};
@@ -647,11 +649,11 @@ void code_section_tests(etest::Suite &s) {
 
         wasm::CodeSection expected{.entries{
                 wasm::CodeEntry{
-                        .code{wasm::instructions::I32Const{42}},
+                        .code{wasm::instructions::I32Const{42}, wasm::instructions::End{}},
                         .locals{{1, wasm::ValueType::Int32}},
                 },
                 wasm::CodeEntry{
-                        .code{},
+                        .code{wasm::instructions::End{}},
                         .locals{{5, wasm::ValueType::Int64}, {6, wasm::ValueType::Float32}},
                 },
         }};

diff --git a/wasm/instructions.cpp b/wasm/instructions.cpp
diff --git a/wasm/instructions.h b/wasm/instructions.h
@@ -11,7 +11,6 @@
 #include <cstdint>
 #include <string_view>
 #include <variant>
-#include <vector>
 
 namespace wasm::instructions {
 
@@ -140,16 +139,14 @@ struct Block {
     static constexpr std::uint8_t kOpcode = 0x02;
     static constexpr std::string_view kMnemonic = "block";
     BlockType type{};
-    std::vector<Instruction> instructions;
-    [[nodiscard]] bool operator==(Block const &) const;
+    [[nodiscard]] bool operator==(Block const &) const = default;
 };
 
 struct Loop {
     static constexpr std::uint8_t kOpcode = 0x03;
     static constexpr std::string_view kMnemonic = "loop";
     BlockType type{};
-    std::vector<Instruction> instructions;
-    [[nodiscard]] bool operator==(Loop const &) const;
+    [[nodiscard]] bool operator==(Loop const &) const = default;
 };
 
 struct Branch {