Add FuzzTest and use it to fuzz types (#7246)

FuzzTest is a state-of-the-art fuzzing framework. It supports writing
property-based fuzz targets very similar to googletest unit tests, where
the framework provides the arguments to the test function drawn from a
given domain. These fuzz tests are run continuously for one second each
when running the normal googletest unit tests, but FuzzTest also
supports building in "fuzzing mode" where the tests can be run
continuously with coverage-guided mutations until they find a bug.

Add FuzzTest as a third_party dependency that is not built by default.
To build FuzzTest and the fuzz tests that use it, set the CMake variable
`BUILD_FUZZTEST=ON`. To build in fuzzing mode, additionally set the
CMake variable `FUZZTEST_FUZZING_MODE=ON`.

One of FuzzTest's key features is its support for domain combinators,
which combine simple domains into more complex domains. For example, the
domain `VariantOf(InRange(0, 10), Arbitrary<std::string>())` produces a
std::variant that either holds an integer between 0 and 10 or an
arbitrary string. The set of available combinators is powerful enough to
build domains for arbitrarily structured types.

Use domain combinators to define a domain of WebAssembly type
definitions. The implementation of this domain follows the same general
structure as the existing heap type fuzzer: it chooses the size of rec
groups, then it chooses the supertypes and hierarchies for all the
definitions, then it generates the particular definitions. The
difference is that all random choices are made by the FuzzTest framework
rather than our own code. Whenever the domains of future choices will
depend on the outcome of the current choice, we use the `FlatMap`
combinator to make a choice from the current domain, then pass it to a
continuation that finishes constructing the final domain of types. This
leads to strange continuation-passing code, but allows us to recursively
construct the domain of type definitions.

The current implementation of the type definition domain is not ideal:
the tree of choices used to produce a particular set of type definitions
is deeper and narrower than it could be. Since a mutation of one choice
in the tree requires regenerating and changing the subtree of choices
rooted at the changed choice, having a narrower tree than necessary
means that small mutations are not as diverse as they could be and
having a deeper tree means that many mutations are larger than they
could be. The quality of the domain construction will be improved in the
future.

.flake8

@@ -6,4 +6,4 @@ ignore =
   E241,
   ; line break after binary operator
   W504
-exclude = third_party,./test/emscripten,./test/spec,./test/wasm-install,./test/lit
+exclude = third_party,./test/emscripten,./test/spec,./test/wasm-install,./test/lit,./_deps

.github/workflows/ci.yml

@@ -147,6 +147,34 @@ jobs:
     - name: test
       run: python check.py --binaryen-bin=out/bin
 
+  # Copied and modified from build-clang
+  build-fuzztest:
+    name: clang with fuzztest
+    runs-on: ubuntu-latest
+    steps:
+    - uses: actions/setup-python@v5
+      with:
+        python-version: '3.x'
+    - uses: actions/checkout@v4
+      with:
+        submodules: true
+    - name: install ninja
+      run: sudo apt-get install ninja-build
+    - name: install v8
+      run: |
+        npm install jsvu -g
+        jsvu --os=default --engines=v8
+    - name: install Python dev dependencies
+      run: pip3 install -r requirements-dev.txt
+    - name: cmake
+      run: |
+        mkdir -p out
+        cmake -S . -B out -G Ninja -DCMAKE_INSTALL_PREFIX=out/install -DCMAKE_C_COMPILER=clang -DCMAKE_CXX_COMPILER=clang++ -DBUILD_FUZZTEST=ON
+    - name: build
+      run: cmake --build out -v
+    - name: test
+      run: python check.py --binaryen-bin=out/bin
+
   # TODO(sbc): Find a way to reduce the duplicate between these sanitizer jobs
   build-asan:
     name: asan

.gitignore

@@ -21,6 +21,8 @@ CMakeFiles
 /.ninja_log
 /bin/
 /lib/
+/_deps/
+/dist/
 /config.h
 /emcc-build
 compile_commands.json

.gitmodules

@@ -4,3 +4,6 @@
 [submodule "test/spec/testsuite"]
 	path = test/spec/testsuite
 	url = https://github.com/WebAssembly/testsuite.git
+[submodule "third_party/fuzztest"]
+	path = third_party/fuzztest
+	url = https://github.com/google/fuzztest

CMakeLists.txt

@@ -36,6 +36,8 @@ option(BYN_ENABLE_LTO "Build with LTO" Off)
 # Turn this off to avoid the dependency on gtest.
 option(BUILD_TESTS "Build GTest-based tests" ON)
 
+option(BUILD_FUZZTEST "Build fuzztest-based tests and fuzzers" OFF)
+
 # Turn this off to build only the library.
 option(BUILD_TOOLS "Build tools" ON)
 
@@ -161,8 +163,8 @@ endfunction()
 
 function(binaryen_add_executable name sources)
   add_executable(${name} ${sources})
-  target_link_libraries(${name} Threads::Threads)
-  target_link_libraries(${name} binaryen)
+  target_link_libraries(${name} PRIVATE Threads::Threads)
+  target_link_libraries(${name} PRIVATE binaryen)
   binaryen_setup_rpath(${name})
   install(TARGETS ${name} DESTINATION ${CMAKE_INSTALL_BINDIR})
 endfunction()
@@ -258,7 +260,10 @@ if(MSVC)
 else() # MSVC
 
   add_compile_flag("-fno-omit-frame-pointer")
-  add_compile_flag("-fno-rtti")
+  if(NOT BUILD_FUZZTEST)
+    # fuzztest depends on RTTIs.
+    add_compile_flag("-fno-rtti")
+  endif()
   if(WIN32)
     add_compile_flag("-D_GNU_SOURCE")
     add_compile_flag("-D__STDC_FORMAT_MACROS")
@@ -276,10 +281,6 @@ else() # MSVC
     # explicitly undefine it:
     add_nondebug_compile_flag("-UNDEBUG")
   endif()
-  if(NOT APPLE AND NOT "${CMAKE_CXX_FLAGS}" MATCHES "-fsanitize")
-    # This flag only applies to shared libraries so don't use add_link_flag
-    set(CMAKE_SHARED_LINKER_FLAGS "${CMAKE_SHARED_LINKER_FLAGS} -Wl,--no-undefined")
-  endif()
 endif()
 
 if(EMSCRIPTEN)
@@ -416,6 +417,25 @@ else() # MSVC
     add_compile_flag("-Wno-deprecated-declarations")
   endif()
 
+  if(BUILD_FUZZTEST)
+    add_compile_flag("-DFUZZTEST")
+    fuzztest_setup_fuzzing_flags()
+
+    # Enabling fuzzing mode turns on sanitizers, which turn on additional
+    # warnings. To keep the build working, do not treat these warnings as
+    # errors.
+    add_compile_flag("-Wno-error=maybe-uninitialized")
+    add_compile_flag("-Wno-error=uninitialized")
+    add_compile_flag("-Wno-error=array-bounds")
+    add_compile_flag("-Wno-error=stringop-overread")
+    add_compile_flag("-Wno-error=missing-field-initializers")
+  endif()
+
+  if(NOT APPLE AND NOT "${CMAKE_CXX_FLAGS}" MATCHES "-fsanitize")
+    # This flag only applies to shared libraries so don't use add_link_flag
+    set(CMAKE_SHARED_LINKER_FLAGS "${CMAKE_SHARED_LINKER_FLAGS} -Wl,--no-undefined")
+  endif()
+
 endif()
 
 # Declare libbinaryen
@@ -483,83 +503,84 @@ endif()
 if(EMSCRIPTEN)
   # binaryen.js WebAssembly variant
   add_executable(binaryen_wasm ${binaryen_SOURCES})
-  target_link_libraries(binaryen_wasm binaryen)
-  target_link_libraries(binaryen_wasm "-sFILESYSTEM")
-  target_link_libraries(binaryen_wasm "-sEXPORT_NAME=Binaryen")
-  target_link_libraries(binaryen_wasm "-sNODERAWFS=0")
+  target_link_libraries(binaryen_wasm PRIVATE binaryen)
+  target_link_libraries(binaryen_wasm PRIVATE "-sFILESYSTEM")
+  target_link_libraries(binaryen_wasm PRIVATE "-sEXPORT_NAME=Binaryen")
+  target_link_libraries(binaryen_wasm PRIVATE "-sNODERAWFS=0")
   # Do not error on the repeated NODERAWFS argument
-  target_link_libraries(binaryen_wasm "-Wno-unused-command-line-argument")
+  target_link_libraries(binaryen_wasm PRIVATE "-Wno-unused-command-line-argument")
   # Emit a single file for convenience of people using binaryen.js as a library,
   # so they only need to distribute a single file.
   if(EMSCRIPTEN_ENABLE_SINGLE_FILE)
-    target_link_libraries(binaryen_wasm "-sSINGLE_FILE")
-  endif()
-  target_link_libraries(binaryen_wasm "-sEXPORT_ES6")
-  target_link_libraries(binaryen_wasm "-sEXPORTED_RUNTIME_METHODS=stringToUTF8OnStack,stringToAscii")
-  target_link_libraries(binaryen_wasm "-sEXPORTED_FUNCTIONS=_malloc,_free")
-  target_link_libraries(binaryen_wasm "--post-js=${CMAKE_CURRENT_SOURCE_DIR}/src/js/binaryen.js-post.js")
-  target_link_libraries(binaryen_wasm "-msign-ext")
-  target_link_libraries(binaryen_wasm "-mbulk-memory")
-  target_link_libraries(binaryen_wasm optimized "--closure=1")
+    target_link_libraries(binaryen_wasm PRIVATE "-sSINGLE_FILE")
+  endif()
+  target_link_libraries(binaryen_wasm PRIVATE "-sEXPORT_ES6")
+  target_link_libraries(binaryen_wasm PRIVATE "-sEXPORTED_RUNTIME_METHODS=stringToUTF8OnStack,stringToAscii")
+  target_link_libraries(binaryen_wasm PRIVATE "-sEXPORTED_FUNCTIONS=_malloc,_free")
+  target_link_libraries(binaryen_wasm PRIVATE "--post-js=${CMAKE_CURRENT_SOURCE_DIR}/src/js/binaryen.js-post.js")
+  target_link_libraries(binaryen_wasm PRIVATE "-msign-ext")
+  target_link_libraries(binaryen_wasm PRIVATE "-mbulk-memory")
+  target_link_libraries(binaryen_wasm PRIVATE optimized "--closure=1")
   # TODO: Fix closure warnings! (#5062)
-  target_link_libraries(binaryen_wasm optimized "-Wno-error=closure")
-  target_link_libraries(binaryen_wasm optimized "-flto")
-  target_link_libraries(binaryen_wasm debug "--profiling")
+  target_link_libraries(binaryen_wasm PRIVATE optimized "-Wno-error=closure")
+  target_link_libraries(binaryen_wasm PRIVATE optimized "-flto")
+  target_link_libraries(binaryen_wasm PRIVATE debug "--profiling")
   # Avoid catching exit as that can confuse error reporting in Node,
   # see https://github.com/emscripten-core/emscripten/issues/17228
-  target_link_libraries(binaryen_wasm "-sNODEJS_CATCH_EXIT=0")
+  target_link_libraries(binaryen_wasm PRIVATE "-sNODEJS_CATCH_EXIT=0")
   install(TARGETS binaryen_wasm DESTINATION ${CMAKE_INSTALL_BINDIR})
 
   # binaryen.js JavaScript variant
   add_executable(binaryen_js ${binaryen_SOURCES})
-  target_link_libraries(binaryen_js binaryen)
-  target_link_libraries(binaryen_js "-sWASM=0")
-  target_link_libraries(binaryen_js "-sWASM_ASYNC_COMPILATION=0")
+  target_link_libraries(binaryen_js PRIVATE binaryen)
+  target_link_libraries(binaryen_js PRIVATE "-sWASM=0")
+  target_link_libraries(binaryen_js PRIVATE "-sWASM_ASYNC_COMPILATION=0")
+
   if(${CMAKE_CXX_COMPILER_VERSION} STREQUAL "6.0.1")
     # only valid with fastcomp and WASM=0
-    target_link_libraries(binaryen_js "-sELIMINATE_DUPLICATE_FUNCTIONS")
+    target_link_libraries(binaryen_js PRIVATE "-sELIMINATE_DUPLICATE_FUNCTIONS")
   endif()
   # Disabling filesystem and setting web environment for js_of_ocaml
   # so it doesn't try to detect the "node" environment
   if(JS_OF_OCAML)
-    target_link_libraries(binaryen_js "-sFILESYSTEM=0")
-    target_link_libraries(binaryen_js "-sENVIRONMENT=web,worker")
+    target_link_libraries(binaryen_js PRIVATE "-sFILESYSTEM=0")
+    target_link_libraries(binaryen_js PRIVATE "-sENVIRONMENT=web,worker")
   else()
-    target_link_libraries(binaryen_js "-sFILESYSTEM=1")
+    target_link_libraries(binaryen_js PRIVATE "-sFILESYSTEM=1")
   endif()
-  target_link_libraries(binaryen_js "-sNODERAWFS=0")
+  target_link_libraries(binaryen_js PRIVATE "-sNODERAWFS=0")
   # Do not error on the repeated NODERAWFS argument
-  target_link_libraries(binaryen_js "-Wno-unused-command-line-argument")
+  target_link_libraries(binaryen_js PRIVATE "-Wno-unused-command-line-argument")
   if(EMSCRIPTEN_ENABLE_SINGLE_FILE)
-    target_link_libraries(binaryen_js "-sSINGLE_FILE")
+    target_link_libraries(binaryen_js PRIVATE "-sSINGLE_FILE")
   endif()
-  target_link_libraries(binaryen_js "-sEXPORT_NAME=Binaryen")
+  target_link_libraries(binaryen_js PRIVATE "-sEXPORT_NAME=Binaryen")
   # Currently, js_of_ocaml can only process ES5 code
   if(JS_OF_OCAML)
-    target_link_libraries(binaryen_js "-sEXPORT_ES6=0")
+    target_link_libraries(binaryen_js PRIVATE "-sEXPORT_ES6=0")
   else()
-    target_link_libraries(binaryen_js "-sEXPORT_ES6=1")
+    target_link_libraries(binaryen_js PRIVATE "-sEXPORT_ES6=1")
   endif()
-  target_link_libraries(binaryen_js "-sEXPORTED_RUNTIME_METHODS=stringToUTF8OnStack,stringToAscii")
-  target_link_libraries(binaryen_js "-sEXPORTED_FUNCTIONS=_malloc,_free")
-  target_link_libraries(binaryen_js "--post-js=${CMAKE_CURRENT_SOURCE_DIR}/src/js/binaryen.js-post.js")
+  target_link_libraries(binaryen_js PRIVATE "-sEXPORTED_RUNTIME_METHODS=stringToUTF8OnStack,stringToAscii")
+  target_link_libraries(binaryen_js PRIVATE "-sEXPORTED_FUNCTIONS=_malloc,_free")
+  target_link_libraries(binaryen_js PRIVATE "--post-js=${CMAKE_CURRENT_SOURCE_DIR}/src/js/binaryen.js-post.js")
   # js_of_ocaml needs a specified variable with special comment to provide the library to consumers
   if(JS_OF_OCAML)
-    target_link_libraries(binaryen_js "--extern-pre-js=${CMAKE_CURRENT_SOURCE_DIR}/src/js/binaryen.jsoo-extern-pre.js")
+    target_link_libraries(binaryen_js PRIVATE "--extern-pre-js=${CMAKE_CURRENT_SOURCE_DIR}/src/js/binaryen.jsoo-extern-pre.js")
   endif()
-  target_link_libraries(binaryen_js optimized "--closure=1")
+  target_link_libraries(binaryen_js PRIVATE optimized "--closure=1")
   # Currently, js_of_ocaml can only process ES5 code
   if(JS_OF_OCAML)
-    target_link_libraries(binaryen_js optimized "--closure-args=\"--language_out=ECMASCRIPT5\"")
+    target_link_libraries(binaryen_js PRIVATE optimized "--closure-args=\"--language_out=ECMASCRIPT5\"")
   endif()
   # TODO: Fix closure warnings! (#5062)
-  target_link_libraries(binaryen_js optimized "-Wno-error=closure")
-  target_link_libraries(binaryen_js optimized "-flto")
-  target_link_libraries(binaryen_js debug "--profiling")
-  target_link_libraries(binaryen_js debug "-sASSERTIONS")
+  target_link_libraries(binaryen_js PRIVATE optimized "-Wno-error=closure")
+  target_link_libraries(binaryen_js PRIVATE optimized "-flto")
+  target_link_libraries(binaryen_js PRIVATE debug "--profiling")
+  target_link_libraries(binaryen_js PRIVATE debug "-sASSERTIONS")
   # Avoid catching exit as that can confuse error reporting in Node,
   # see https://github.com/emscripten-core/emscripten/issues/17228
-  target_link_libraries(binaryen_js "-sNODEJS_CATCH_EXIT=0")
+  target_link_libraries(binaryen_js PRIVATE "-sNODEJS_CATCH_EXIT=0")
   install(TARGETS binaryen_js DESTINATION ${CMAKE_INSTALL_BINDIR})
 endif()
 

check.py

@@ -43,11 +43,11 @@ def run_version_tests():
     print('[ checking --version ... ]\n')
 
     not_executable_suffix = ['.DS_Store', '.txt', '.js', '.ilk', '.pdb', '.dll', '.wasm', '.manifest']
-    not_executable_prefix = ['binaryen-lit', 'binaryen-unittests']
+    executable_prefix = ['wasm']
     bin_files = [os.path.join(shared.options.binaryen_bin, f) for f in os.listdir(shared.options.binaryen_bin)]
     executables = [f for f in bin_files if os.path.isfile(f) and
                    not any(f.endswith(s) for s in not_executable_suffix) and
-                   not any(os.path.basename(f).startswith(s) for s in not_executable_prefix)]
+                   any(os.path.basename(f).startswith(s) for s in executable_prefix)]
     executables = sorted(executables)
     assert len(executables)
 

test/gtest/CMakeLists.txt

@@ -1,4 +1,8 @@
-include_directories(SYSTEM ${PROJECT_SOURCE_DIR}/third_party/googletest/googletest/include)
+if(BUILD_FUZZTEST)
+  include_directories(SYSTEM ${PROJECT_SOURCE_DIR}/third_party/fuzztest)
+else()
+  include_directories(SYSTEM ${PROJECT_SOURCE_DIR}/third_party/googletest/googletest/include)
+endif()
 
 set(unittest_SOURCES
   arena.cpp
@@ -21,10 +25,21 @@ set(unittest_SOURCES
   validator.cpp
 )
 
+if(BUILD_FUZZTEST)
+  set(unittest_SOURCES ${unittest_SOURCES} type-domains.cpp)
+endif()
+
 # suffix_tree.cpp includes LLVM header using std::iterator (deprecated in C++17)
 if (NOT MSVC)
   set_source_files_properties(suffix_tree.cpp PROPERTIES COMPILE_FLAGS -Wno-deprecated-declarations)
 endif()
 
+enable_testing()
+include(GoogleTest)
 binaryen_add_executable(binaryen-unittests "${unittest_SOURCES}")
-target_link_libraries(binaryen-unittests gtest gtest_main)
+if(BUILD_FUZZTEST)
+  link_fuzztest(binaryen-unittests)
+  gtest_discover_tests(binaryen-unittests)
+else()
+  target_link_libraries(binaryen-unittests PRIVATE gtest gtest_main)
+endif()

test/gtest/type-builder.cpp

@@ -5,6 +5,10 @@
 #include "wasm-type.h"
 #include "gtest/gtest.h"
 
+#ifdef FUZZTEST
+#include "type-domains.h"
+#endif
+
 using namespace wasm;
 
 TEST_F(TypeTest, TypeBuilderGrowth) {
@@ -1056,6 +1060,41 @@ TEST_F(TypeTest, TestHeapTypeRelations) {
   }
 }
 
+#ifdef FUZZTEST
+
+void TestHeapTypeRelationsFuzz(std::pair<HeapType, HeapType> pair) {
+  auto [a, b] = pair;
+  auto lub = HeapType::getLeastUpperBound(a, b);
+  auto otherLub = HeapType::getLeastUpperBound(b, a);
+  EXPECT_EQ(lub, otherLub);
+  if (lub) {
+    EXPECT_EQ(a.getTop(), b.getTop());
+    EXPECT_EQ(a.getBottom(), b.getBottom());
+    EXPECT_TRUE(HeapType::isSubType(a, *lub));
+    EXPECT_TRUE(HeapType::isSubType(b, *lub));
+  } else {
+    EXPECT_NE(a.getTop(), b.getTop());
+    EXPECT_NE(a.getBottom(), b.getBottom());
+  }
+  if (a == b) {
+    EXPECT_EQ(lub, a);
+    EXPECT_EQ(lub, b);
+  } else if (lub && *lub == b) {
+    EXPECT_TRUE(HeapType::isSubType(a, b));
+    EXPECT_FALSE(HeapType::isSubType(b, a));
+  } else if (lub && *lub == a) {
+    EXPECT_FALSE(HeapType::isSubType(a, b));
+    EXPECT_TRUE(HeapType::isSubType(b, a));
+  } else if (lub) {
+    EXPECT_FALSE(HeapType::isSubType(a, b));
+    EXPECT_FALSE(HeapType::isSubType(b, a));
+  }
+}
+FUZZ_TEST(TypeFuzzTest, TestHeapTypeRelationsFuzz)
+  .WithDomains(ArbitraryHeapTypePair());
+
+#endif // FUZZTEST
+
 TEST_F(TypeTest, TestSubtypeErrors) {
   Type anyref = Type(HeapType::any, Nullable);
   Type eqref = Type(HeapType::eq, Nullable);