Extract common bytecode operations into WasmBase. (#161)

Signed-off-by: Piotr Sikora <[email protected]>

Author: PiotrSikora

BUILD

@@ -25,10 +25,9 @@ cc_library(
         "src/*.h",
         "src/*.cc",
         "src/common/*.h",
-        "src/common/*.cc",
+        "src/null/*.cc",
         "src/third_party/*.h",
         "src/third_party/*.cc",
-        "src/null/*.cc",
     ]),
     deps = [
         ":include",

src/common/bytecode_util.h renamed to include/proxy-wasm/bytecode_util.h

@@ -20,7 +20,6 @@
 #include "include/proxy-wasm/wasm_vm.h"
 
 namespace proxy_wasm {
-namespace common {
 
 // Utilitiy functions which directly operate on Wasm bytecodes.
 class BytecodeUtil {
@@ -73,5 +72,4 @@ class BytecodeUtil {
   static bool parseVarint(const char *&begin, const char *end, uint32_t &ret);
 };
 
-} // namespace common
 } // namespace proxy_wasm

include/proxy-wasm/null_vm.h

@@ -35,8 +35,8 @@ struct NullVm : public WasmVm {
   std::string_view runtime() override { return "null"; }
   Cloneable cloneable() override { return Cloneable::InstantiatedModule; };
   std::unique_ptr<WasmVm> clone() override;
-  bool load(const std::string &code, bool allow_precompiled) override;
-  AbiVersion getAbiVersion() override;
+  bool load(std::string_view bytecode, std::string_view precompiled,
+            std::unordered_map<uint32_t, std::string> function_names) override;
   bool link(std::string_view debug_name) override;
   uint64_t getMemorySize() override;
   std::optional<std::string_view> getMemory(uint64_t pointer, uint64_t size) override;

include/proxy-wasm/wasm.h

@@ -58,7 +58,8 @@ class WasmBase : public std::enable_shared_from_this<WasmBase> {
   WasmBase(const std::shared_ptr<WasmHandleBase> &other, WasmVmFactory factory);
   virtual ~WasmBase();
 
-  bool initialize(const std::string &code, bool allow_precompiled = false);
+  bool load(const std::string &code, bool allow_precompiled = false);
+  bool initialize();
   void startVm(ContextBase *root_context);
   bool configure(ContextBase *root_context, std::shared_ptr<PluginBase> plugin);
   // Returns the root ContextBase or nullptr if onStart returns false.
@@ -79,9 +80,11 @@ class WasmBase : public std::enable_shared_from_this<WasmBase> {
   bool isFailed() { return failed_ != FailState::Ok; }
   FailState fail_state() { return failed_; }
 
-  const std::string &code() const { return code_; }
   const std::string &vm_configuration() const;
-  bool allow_precompiled() const { return allow_precompiled_; }
+
+  const std::string &moduleBytecode() const { return module_bytecode_; }
+  const std::string &modulePrecompiled() const { return module_precompiled_; }
+  const std::unordered_map<uint32_t, std::string> functionNames() const { return function_names_; }
 
   void timerReady(uint32_t root_context_id);
   void queueReady(uint32_t root_context_id, uint32_t token);
@@ -135,7 +138,7 @@ class WasmBase : public std::enable_shared_from_this<WasmBase> {
   virtual void error(std::string_view message) { std::cerr << message << "\n"; }
   virtual void unimplemented() { error("unimplemented proxy-wasm API"); }
 
-  AbiVersion abiVersion() { return abi_version_; }
+  AbiVersion abiVersion() const { return abi_version_; }
 
   const std::unordered_map<std::string, std::string> &envs() { return envs_; }
 
@@ -182,7 +185,7 @@ class WasmBase : public std::enable_shared_from_this<WasmBase> {
   std::string vm_id_;  // User-provided vm_id.
   std::string vm_key_; // vm_id + hash of code.
   std::unique_ptr<WasmVm> wasm_vm_;
-  Cloneable started_from_{Cloneable::NotCloneable};
+  std::optional<Cloneable> started_from_;
 
   uint32_t next_context_id_ = 1;            // 0 is reserved for the VM context.
   std::shared_ptr<ContextBase> vm_context_; // Context unrelated to any specific root or stream
@@ -260,15 +263,17 @@ class WasmBase : public std::enable_shared_from_this<WasmBase> {
   std::shared_ptr<WasmHandleBase> base_wasm_handle_;
 
   // Used by the base_wasm to enable non-clonable thread local Wasm(s) to be constructed.
-  std::string code_;
-  std::string vm_configuration_;
-  bool allow_precompiled_ = false;
-  bool stop_iteration_ = false;
-  FailState failed_ = FailState::Ok; // Wasm VM fatal error.
+  std::string module_bytecode_;
+  std::string module_precompiled_;
+  std::unordered_map<uint32_t, std::string> function_names_;
 
   // ABI version.
   AbiVersion abi_version_ = AbiVersion::Unknown;
 
+  std::string vm_configuration_;
+  bool stop_iteration_ = false;
+  FailState failed_ = FailState::Ok; // Wasm VM fatal error.
+
   // Plugin Stats/Metrics
   uint32_t next_counter_metric_id_ = static_cast<uint32_t>(MetricType::Counter);
   uint32_t next_gauge_metric_id_ = static_cast<uint32_t>(MetricType::Gauge);

include/proxy-wasm/wasm_vm.h

@@ -19,6 +19,7 @@
 #include <memory>
 #include <optional>
 #include <string>
+#include <unordered_map>
 
 #include "include/proxy-wasm/word.h"
 
@@ -174,14 +175,13 @@ class WasmVm {
    * Load the WASM code from a file. Return true on success. Once the module is loaded it can be
    * queried, e.g. to see which version of emscripten support is required. After loading, the
    * appropriate ABI callbacks can be registered and then the module can be link()ed (see below).
-   * @param code the WASM binary code (or registered NullVm plugin name).
-   * @param allow_precompiled if true, allows supporting VMs (e.g. WAVM) to load the binary
-   * machine code from a user-defined section of the WASM file. Because that code is not verified by
-   * the proxy process it is up to the user to ensure that the code is both safe and is built for
-   * the linked in version of WAVM.
+   * @param bytecode the Wasm bytecode or registered NullVm plugin name.
+   * @param precompiled (optional) the precompiled Wasm module.
+   * @param function_names (optional) an index-to-name mapping for the functions.
    * @return whether or not the load was successful.
    */
-  virtual bool load(const std::string &code, bool allow_precompiled) = 0;
+  virtual bool load(std::string_view bytecode, std::string_view precompiled,
+                    const std::unordered_map<uint32_t, std::string> function_names) = 0;
 
   /**
    * Link the WASM code to the host-provided functions, e.g. the ABI. Prior to linking, the module
@@ -192,12 +192,6 @@ class WasmVm {
    */
   virtual bool link(std::string_view debug_name) = 0;
 
-  /**
-   * Get ABI version of the module.
-   * @return the ABI version.
-   */
-  virtual AbiVersion getAbiVersion() = 0;
-
   /**
    * Get size of the currently allocated memory in the VM.
    * @return the size of memory in bytes.

src/common/bytecode_util.cc renamed to src/bytecode_util.cc

@@ -12,11 +12,11 @@
 // See the License for the specific language governing permissions and
 // limitations under the License.
 
-#include "src/common/bytecode_util.h"
+#include "include/proxy-wasm/bytecode_util.h"
+
 #include <cstring>
 
 namespace proxy_wasm {
-namespace common {
 
 bool BytecodeUtil::checkWasmHeader(std::string_view bytecode) {
   // Wasm file header is 8 bytes (magic number + version).
@@ -240,5 +240,4 @@ bool BytecodeUtil::parseVarint(const char *&pos, const char *end, uint32_t &ret)
   return ret != static_cast<uint32_t>(-1);
 }
 
-} // namespace common
 } // namespace proxy_wasm

src/null/null_vm.cc

@@ -40,24 +40,23 @@ std::unique_ptr<WasmVm> NullVm::clone() {
   auto cloned_null_vm = std::make_unique<NullVm>(*this);
   if (integration())
     cloned_null_vm->integration().reset(integration()->clone());
-  cloned_null_vm->load(plugin_name_, false /* unused */);
+  cloned_null_vm->load(plugin_name_, {} /* unused */, {} /* unused */);
   return cloned_null_vm;
 }
 
 // "Load" the plugin by obtaining a pointer to it from the factory.
-bool NullVm::load(const std::string &name, bool /* allow_precompiled */) {
+bool NullVm::load(std::string_view name, std::string_view,
+                  const std::unordered_map<uint32_t, std::string>) {
   if (!null_vm_plugin_factories_)
     return false;
-  auto factory = (*null_vm_plugin_factories_)[name];
+  auto factory = (*null_vm_plugin_factories_)[std::string(name)];
   if (!factory)
     return false;
   plugin_name_ = name;
   plugin_ = factory();
   plugin_->wasm_vm_ = this;
   return true;
-}
-
-AbiVersion NullVm::getAbiVersion() { return AbiVersion::ProxyWasm_0_2_1; }
+} // namespace proxy_wasm
 
 bool NullVm::link(std::string_view /* name */) { return true; }
 

src/v8/v8.cc

@@ -25,8 +25,6 @@
 #include <utility>
 #include <vector>
 
-#include "src/common/bytecode_util.h"
-
 #include "v8.h"
 #include "v8-version.h"
 #include "wasm-api/wasm.hh"
@@ -64,8 +62,8 @@ class V8 : public WasmVm {
   // WasmVm
   std::string_view runtime() override { return "v8"; }
 
-  bool load(const std::string &code, bool allow_precompiled) override;
-  AbiVersion getAbiVersion() override;
+  bool load(std::string_view bytecode, std::string_view precompiled,
+            const std::unordered_map<uint32_t, std::string> function_names) override;
   std::string_view getPrecompiledSectionName() override;
   bool link(std::string_view debug_name) override;
 
@@ -122,8 +120,6 @@ class V8 : public WasmVm {
 
   std::unordered_map<std::string, FuncDataPtr> host_functions_;
   std::unordered_map<std::string, wasm::own<wasm::Func>> module_functions_;
-
-  AbiVersion abi_version_;
   std::unordered_map<uint32_t, std::string> function_names_index_;
 };
 
@@ -251,58 +247,31 @@ template <typename T, typename U> constexpr T convertValTypesToArgsTuple(const U
 
 // V8 implementation.
 
-bool V8::load(const std::string &code, bool allow_precompiled) {
+bool V8::load(std::string_view bytecode, std::string_view precompiled,
+              const std::unordered_map<uint32_t, std::string> function_names) {
   store_ = wasm::Store::make(engine());
 
-  // Get ABI version from bytecode.
-  if (!common::BytecodeUtil::getAbiVersion(code, abi_version_)) {
-    fail(FailState::UnableToInitializeCode, "Failed to parse corrupted Wasm module");
-    return false;
-  }
+  if (!precompiled.empty()) {
+    auto vec = wasm::vec<byte_t>::make_uninitialized(precompiled.size());
+    ::memcpy(vec.get(), precompiled.data(), precompiled.size());
+    module_ = wasm::Module::deserialize(store_.get(), vec);
 
-  if (allow_precompiled) {
-    const auto section_name = getPrecompiledSectionName();
-    if (!section_name.empty()) {
-      std::string_view precompiled = {};
-      if (!common::BytecodeUtil::getCustomSection(code, section_name, precompiled)) {
-        fail(FailState::UnableToInitializeCode, "Failed to parse corrupted Wasm module");
-        return false;
-      }
-      if (!precompiled.empty()) {
-        auto vec = wasm::vec<byte_t>::make_uninitialized(precompiled.size());
-        ::memcpy(vec.get(), precompiled.data(), precompiled.size());
-
-        module_ = wasm::Module::deserialize(store_.get(), vec);
-        if (!module_) {
-          // Precompiled module that cannot be loaded is considered a hard error,
-          // so don't fallback to compiling the bytecode.
-          return false;
-        }
-      }
-    }
+  } else {
+    auto vec = wasm::vec<byte_t>::make_uninitialized(bytecode.size());
+    ::memcpy(vec.get(), bytecode.data(), bytecode.size());
+    module_ = wasm::Module::make(store_.get(), vec);
   }
 
   if (!module_) {
-    std::string stripped;
-    if (!common::BytecodeUtil::getStrippedSource(code, stripped)) {
-      fail(FailState::UnableToInitializeCode, "Failed to parse corrupted Wasm module");
-      return false;
-    };
-    wasm::vec<byte_t> stripped_vec = wasm::vec<byte_t>::make(stripped.size(), stripped.data());
-    module_ = wasm::Module::make(store_.get(), stripped_vec);
+    return false;
   }
 
-  if (module_) {
-    shared_module_ = module_->share();
-    assert((shared_module_ != nullptr));
-  }
+  shared_module_ = module_->share();
+  assert(shared_module_ != nullptr);
 
-  if (!common::BytecodeUtil::getFunctionNameIndex(code, function_names_index_)) {
-    fail(FailState::UnableToInitializeCode, "Failed to parse corrupted Wasm module");
-    return false;
-  };
+  function_names_index_ = function_names;
 
-  return module_ != nullptr;
+  return true;
 }
 
 std::unique_ptr<WasmVm> V8::clone() {
@@ -314,7 +283,6 @@ std::unique_ptr<WasmVm> V8::clone() {
 
   clone->module_ = wasm::Module::obtain(clone->store_.get(), shared_module_.get());
   clone->function_names_index_ = function_names_index_;
-  clone->abi_version_ = abi_version_;
 
   return clone;
 }
@@ -335,8 +303,6 @@ std::string_view V8::getPrecompiledSectionName() {
   return name;
 }
 
-AbiVersion V8::getAbiVersion() { return abi_version_; }
-
 bool V8::link(std::string_view debug_name) {
   assert(module_ != nullptr);
 

src/wamr/wamr.cc

@@ -29,9 +29,10 @@
 #include <utility>
 #include <vector>
 
+#include "include/proxy-wasm/bytecode_util.h"
+
 #include "src/wamr/types.h"
 #include "wasm_c_api.h"
-#include "src/common/bytecode_util.h"
 
 namespace proxy_wasm {
 namespace wamr {
@@ -62,9 +63,8 @@ class Wamr : public WasmVm {
   Cloneable cloneable() override { return Cloneable::NotCloneable; }
   std::unique_ptr<WasmVm> clone() override { return nullptr; }
 
-  AbiVersion getAbiVersion() override;
-
-  bool load(const std::string &code, bool allow_precompiled = false) override;
+  bool load(std::string_view bytecode, std::string_view precompiled,
+            const std::unordered_map<uint32_t, std::string> function_names) override;
   bool link(std::string_view debug_name) override;
   uint64_t getMemorySize() override;
   std::optional<std::string_view> getMemory(uint64_t pointer, uint64_t size) override;
@@ -113,27 +113,15 @@ class Wamr : public WasmVm {
 
   std::unordered_map<std::string, HostFuncDataPtr> host_functions_;
   std::unordered_map<std::string, WasmFuncPtr> module_functions_;
-  AbiVersion abi_version_;
 };
 
-bool Wamr::load(const std::string &code, bool allow_precompiled) {
+bool Wamr::load(std::string_view bytecode, std::string_view,
+                const std::unordered_map<uint32_t, std::string>) {
   store_ = wasm_store_new(engine());
 
-  // Get ABI version from bytecode.
-  if (!common::BytecodeUtil::getAbiVersion(code, abi_version_)) {
-    fail(FailState::UnableToInitializeCode, "Failed to parse corrupted Wasm module");
-    return false;
-  }
-
-  std::string stripped;
-  if (!common::BytecodeUtil::getStrippedSource(code, stripped)) {
-    fail(FailState::UnableToInitializeCode, "Failed to parse corrupted Wasm module");
-    return false;
-  };
-
-  WasmByteVec stripped_vec;
-  wasm_byte_vec_new(stripped_vec.get(), stripped.size(), stripped.data());
-  module_ = wasm_module_new(store_.get(), stripped_vec.get());
+  WasmByteVec vec;
+  wasm_byte_vec_new(vec.get(), bytecode.size(), bytecode.data());
+  module_ = wasm_module_new(store_.get(), vec.get());
 
   return module_ != nullptr;
 }
@@ -622,8 +610,6 @@ void Wamr::getModuleFunctionImpl(std::string_view function_name,
   };
 };
 
-AbiVersion Wamr::getAbiVersion() { return abi_version_; }
-
 } // namespace wamr
 
 std::unique_ptr<WasmVm> createWamrVm() { return std::make_unique<wamr::Wamr>(); }