[compiler-rt] Use interception to access to strong functions defined in the main executable.

[mpividori]

Summary

In Windows, when sanitizers are implemented as a shared library (DLL), users can redefine and export a new definition for weak functions, in the main executable, for example:

extern "C" __declspec(dllexport) void __sanitizer_cov_trace_pc_guard(u32* guard) {
  // Different implementation provided by the client.
}

However, other client's dlls, will continue using the default implementation imported from the sanitizer dll.
This is different in linux, where all the shared libraries will consider the strong definition.

With the implementation in this diff, when the dll is initialized, it will check if the main executable exports the definition for some weak function (for example __sanitizer_cov_trace_pc_guard). If it finds that function, then it will override the function in the dll with that pointer. So, all the client's dlls with instrumentation that import __sanitizer_cov_trace_pc_guard__dll() from dll, will be using the function provided by the main executable.

In other words, when the main executable exports a strong definition for a weak function, we ensure all the dlls use that implementation instead of the default weak implementation.

So, the behavior is similar to linux. Now, every user that want to override a weak function, only has to define and export it. The same for Linux and Windows, and it will work fine. So, there is no difference on the user's side.

---

All the sanitizers will include a file sanitizer_win_weak_interception.cc that register sanitizer's weak functions to be intercepted in the binary section: WEAK

When the sanitizer dll is initialized, it will execute weak_intercept_init() implemented in: sanitizer_common/sanitizer_win_weak_interception.cc, which will consider all the CB registered in the section WEAK. So, for all the weak functions registered, we will check if a strong definition is provided in the main executable. If that is the case, we will override the dll implementation with the user provided implementation.

All the files sanitizer_win_weak_interception.cc are independent, so we don't need to include a specific list of sanitizers.
Now, we include asan_win_weak_interception.cc ubsan_win_weak_interception.cc sanitizer_coverage_win_weak_interception.cc and sanitizer_win_weak_interception.cc, in asan dll, so when it is initialized, it will consider all the weak functions from asan, ubsan and sanitizer coverage.

---

After this diff, sanitizer coverage is fixed for MD on Windows. In particular libFuzzer can provide custom implementation for all sanitizer coverage's weak functions, and they will be considered by asan dll.

[zturner]

lib/sanitizer_common/sanitizer_win_weak_interception.cc

34–35 ↗

(On Diff #85863)

I don't have a strong opinion here, but if the override fails, should it abort, or should it just behave as if the function was not found?

[mpividori]

lib/sanitizer_common/sanitizer_win_weak_interception.cc

34–35 ↗

(On Diff #85863)

I think it is appropriate to fail.
If the user has provided a strong definition (that means we can find it with InternalGetProcAddress()) then we should use it. We can not ignore it because it would be incorrect. If we can not override it, then we should fail because otherwise the program would behave in an unexpected way.
Is the same that we do on dll_thunks in https://reviews.llvm.org/D29154. If it fails to override the function, it aborts.

[alekseyshl]

LGTM

lib/sanitizer_common/sanitizer_win_weak_interception.cc

33 ↗

(On Diff #85863)

Four spaces indent.

71 ↗

(On Diff #85863)

Four spaces indent.

[mpividori]

Ok. @zturner @alekseyshl thanks for your time.

compiler-rt/trunk/lib/asan/CMakeLists.txt

if (UNIX AND NOT ${arch} MATCHES "i386|i686")
       set_source_files_properties(
         ${CMAKE_CURRENT_BINARY_DIR}/dummy.cc
         PROPERTIES
         OBJECT_DEPENDS ${CMAKE_CURRENT_BINARY_DIR}/clang_rt.asan-dynamic-${arch}.vers)
     else()
       set(VERSION_SCRIPT_FLAG)
     endif()
 
+    set(ASAN_DYNAMIC_WEAK_INTERCEPTION)
+    if (MSVC)
+      add_compiler_rt_object_libraries(AsanWeakInterception
+        ${SANITIZER_COMMON_SUPPORTED_OS}
+        ARCHS ${ASAN_SUPPORTED_ARCH}
+        SOURCES asan_win_weak_interception.cc
+        CFLAGS ${ASAN_CFLAGS} -DSANITIZER_DYNAMIC
+        DEFS ${ASAN_COMMON_DEFINITIONS})
+      set(ASAN_DYNAMIC_WEAK_INTERCEPTION
+          AsanWeakInterception
+          UbsanWeakInterception
+          SancovWeakInterception
+          SanitizerCommonWeakInterception)
+    endif()
+
     add_compiler_rt_runtime(clang_rt.asan
       SHARED
       ARCHS ${arch}
       OBJECT_LIBS ${ASAN_COMMON_RUNTIME_OBJECT_LIBS}
               RTAsan_dynamic
               # The only purpose of RTAsan_dynamic_version_script_dummy is to carry
               # a dependency of the shared runtime on the version script. With CMake
               # 3.1 or later it can be replaced with a straightforward
               # add_dependencies(clang_rt.asan-dynamic-${arch} clang_rt.asan-dynamic-${arch}-version-list)
               RTAsan_dynamic_version_script_dummy
               RTUbsan_cxx
+              ${ASAN_DYNAMIC_WEAK_INTERCEPTION}
       CFLAGS ${ASAN_DYNAMIC_CFLAGS}
       LINK_FLAGS ${ASAN_DYNAMIC_LINK_FLAGS}
                 ${VERSION_SCRIPT_FLAG}
       LINK_LIBS ${ASAN_DYNAMIC_LIBS}
       DEFS ${ASAN_DYNAMIC_DEFINITIONS}
       PARENT_TARGET asan)
 
     if (UNIX AND NOT ${arch} MATCHES "i386|i686")

compiler-rt/trunk/lib/asan/asan_win_weak_interception.cc

+//===-- asan_win_weak_interception.cc -------------------------------------===//
+//
+//                     The LLVM Compiler Infrastructure
+//
+// This file is distributed under the University of Illinois Open Source
+// License. See LICENSE.TXT for details.
+//
+//===----------------------------------------------------------------------===//
+// This module should be included in Address Sanitizer when it is implemented as
+// a shared library on Windows (dll), in order to delegate the calls of weak
+// functions to the implementation in the main executable when a strong
+// definition is provided.
+//===----------------------------------------------------------------------===//
+#ifdef SANITIZER_DYNAMIC
+#include "sanitizer_common/sanitizer_win_weak_interception.h"
+#include "asan_interface_internal.h"
+// Check if strong definitions for weak functions are present in the main
+// executable. If that is the case, override dll functions to point to strong
+// implementations.
+#define INTERFACE_FUNCTION(Name)
+#define INTERFACE_WEAK_FUNCTION(Name) INTERCEPT_SANITIZER_WEAK_FUNCTION(Name)
+#include "asan_interface.inc"
+#endif // SANITIZER_DYNAMIC

compiler-rt/trunk/lib/sanitizer_common/CMakeLists.txt

 add_compiler_rt_object_libraries(RTSanitizerCommonLibc
   ${OS_OPTION}
   ARCHS ${SANITIZER_COMMON_SUPPORTED_ARCH}
   SOURCES ${SANITIZER_LIBCDEP_SOURCES}
   CFLAGS ${SANITIZER_CFLAGS}
   DEFS ${SANITIZER_COMMON_DEFINITIONS})
 
 if(MSVC)
+  add_compiler_rt_object_libraries(SanitizerCommonWeakInterception
+    ${SANITIZER_COMMON_SUPPORTED_OS}
+    ARCHS ${SANITIZER_COMMON_SUPPORTED_ARCH}
+    SOURCES sanitizer_win_weak_interception.cc
+    CFLAGS ${SANITIZER_CFLAGS} -DSANITIZER_DYNAMIC
+    DEFS ${SANITIZER_COMMON_DEFINITIONS})
+  add_compiler_rt_object_libraries(SancovWeakInterception
+    ${SANITIZER_COMMON_SUPPORTED_OS}
+    ARCHS ${SANITIZER_COMMON_SUPPORTED_ARCH}
+    SOURCES sanitizer_coverage_win_weak_interception.cc
+    CFLAGS ${SANITIZER_CFLAGS} -DSANITIZER_DYNAMIC
+    DEFS ${SANITIZER_COMMON_DEFINITIONS})
+
   add_compiler_rt_object_libraries(SanitizerCommonDllThunk
     ${SANITIZER_COMMON_SUPPORTED_OS}
     ARCHS ${SANITIZER_COMMON_SUPPORTED_ARCH}
     SOURCES sanitizer_win_dll_thunk.cc
     CFLAGS ${SANITIZER_CFLAGS} -DSANITIZER_DLL_THUNK
     DEFS ${SANITIZER_COMMON_DEFINITIONS})
   add_compiler_rt_object_libraries(SancovDllThunk
     ${SANITIZER_COMMON_SUPPORTED_OS}

compiler-rt/trunk/lib/sanitizer_common/sanitizer_coverage_win_weak_interception.cc

+//===-- sanitizer_coverage_win_weak_interception.cc -----------------------===//
+//
+//                     The LLVM Compiler Infrastructure
+//
+// This file is distributed under the University of Illinois Open Source
+// License. See LICENSE.TXT for details.
+//
+//===----------------------------------------------------------------------===//
+// This module should be included in Sanitizer Coverage when it implemented as a
+// shared library on Windows (dll), in order to delegate the calls of weak
+// functions to the implementation in the main executable when a strong
+// definition is provided.
+//===----------------------------------------------------------------------===//
+#ifdef SANITIZER_DYNAMIC
+#include "sanitizer_win_weak_interception.h"
+#include "sanitizer_interface_internal.h"
+#include "sancov_flags.h"
+// Check if strong definitions for weak functions are present in the main
+// executable. If that is the case, override dll functions to point to strong
+// implementations.
+#define INTERFACE_FUNCTION(Name)
+#define INTERFACE_WEAK_FUNCTION(Name) INTERCEPT_SANITIZER_WEAK_FUNCTION(Name)
+#include "sanitizer_coverage_interface.inc"
+#endif // SANITIZER_DYNAMIC

compiler-rt/trunk/lib/sanitizer_common/sanitizer_win_weak_interception.h

+//===-- sanitizer_win_weak_interception.h ---------------------------------===//
+//
+//                     The LLVM Compiler Infrastructure
+//
+// This file is distributed under the University of Illinois Open Source
+// License. See LICENSE.TXT for details.
+//
+//===----------------------------------------------------------------------===//
+// This header provide helper macros to delegate calls of weak functions to the
+// implementation in the main executable when a strong definition is present.
+//===----------------------------------------------------------------------===//
+#ifndef SANITIZER_WIN_WEAK_INTERCEPTION_H
+#define SANITIZER_WIN_WEAK_INTERCEPTION_H
+#include "sanitizer_internal_defs.h"
+
+namespace __sanitizer {
+int interceptWhenPossible(uptr dll_function, const char *real_function);
+}
+
+// ----------------- Function interception helper macros -------------------- //
+// Weak functions, could be redefined in the main executable, but that is not
+// necessary, so we shouldn't die if we can not find a reference.
+#define INTERCEPT_WEAK(Name) interceptWhenPossible((uptr) Name, #Name);
+
+#define INTERCEPT_SANITIZER_WEAK_FUNCTION(Name)                                \
+  static int intercept_##Name() {                                              \
+    return __sanitizer::interceptWhenPossible((__sanitizer::uptr) Name, #Name);\
+  }                                                                            \
+  __pragma(section(".WEAK$M", long, read))                                     \
+  __declspec(allocate(".WEAK$M")) int (*__weak_intercept_##Name)() =           \
+      intercept_##Name;
+
+#endif // SANITIZER_WIN_WEAK_INTERCEPTION_H

compiler-rt/trunk/lib/sanitizer_common/sanitizer_win_weak_interception.cc

+//===-- sanitizer_win_weak_interception.cc --------------------------------===//
+//
+//                     The LLVM Compiler Infrastructure
+//
+// This file is distributed under the University of Illinois Open Source
+// License. See LICENSE.TXT for details.
+//
+//===----------------------------------------------------------------------===//
+// This module should be included in the sanitizer when it is implemented as a
+// shared library on Windows (dll), in order to delegate the calls of weak
+// functions to the implementation in the main executable when a strong
+// definition is provided.
+//===----------------------------------------------------------------------===//
+
+#include "sanitizer_common/sanitizer_platform.h"
+#if SANITIZER_WINDOWS && SANITIZER_DYNAMIC
+#include "sanitizer_win_weak_interception.h"
+#include "sanitizer_allocator_interface.h"
+#include "sanitizer_interface_internal.h"
+#include "sanitizer_win_defs.h"
+#include "interception/interception.h"
+
+extern "C" {
+void *WINAPI GetModuleHandleA(const char *module_name);
+void abort();
+}
+
+namespace __sanitizer {
+// Try to get a pointer to real_function in the main module and override
+// dll_function with that pointer. If the function isn't found, nothing changes.
+int interceptWhenPossible(uptr dll_function, const char *real_function) {
+  uptr real = __interception::InternalGetProcAddress(
+      (void *)GetModuleHandleA(0), real_function);
+  if (real && !__interception::OverrideFunction((uptr)dll_function, real, 0))
+    abort();
+  return 0;
+}
+} // namespace __sanitizer
+
+// Include Sanitizer Common interface.
+#define INTERFACE_FUNCTION(Name)
+#define INTERFACE_WEAK_FUNCTION(Name) INTERCEPT_SANITIZER_WEAK_FUNCTION(Name)
+#include "sanitizer_common_interface.inc"
+
+#pragma section(".WEAK$A", read)  // NOLINT
+#pragma section(".WEAK$Z", read)  // NOLINT
+
+typedef void (*InterceptCB)();
+extern "C" {
+__declspec(allocate(".WEAK$A")) InterceptCB __start_weak_list;
+__declspec(allocate(".WEAK$Z")) InterceptCB __stop_weak_list;
+}
+
+static int weak_intercept_init() {
+  static bool flag = false;
+  // weak_interception_init is expected to be called by only one thread.
+  if (flag) return 0;
+  flag = true;
+
+  for (InterceptCB *it = &__start_weak_list; it < &__stop_weak_list; ++it)
+    if (*it)
+      (*it)();
+
+  // In DLLs, the callbacks are expected to return 0,
+  // otherwise CRT initialization fails.
+  return 0;
+}
+
+#pragma section(".CRT$XIB", long, read)  // NOLINT
+__declspec(allocate(".CRT$XIB")) int (*__weak_intercept_preinit)() =
+    weak_intercept_init;
+
+static void WINAPI weak_intercept_thread_init(void *mod, unsigned long reason,
+                                              void *reserved) {
+  if (reason == /*DLL_PROCESS_ATTACH=*/1) weak_intercept_init();
+}
+
+#pragma section(".CRT$XLAB", long, read)  // NOLINT
+__declspec(allocate(".CRT$XLAB")) void(WINAPI *__weak_intercept_tls_init)(
+    void *, unsigned long, void *) = weak_intercept_thread_init;
+
+#endif // SANITIZER_WINDOWS && SANITIZER_DYNAMIC

compiler-rt/trunk/lib/ubsan/CMakeLists.txt

else()
     set(UBSAN_CXX_SOURCES ${CMAKE_CURRENT_BINARY_DIR}/cxx_dummy.cc)
   endif()
 
   add_compiler_rt_object_libraries(RTUbsan_cxx
     ARCHS ${UBSAN_COMMON_SUPPORTED_ARCH}
     SOURCES ${UBSAN_CXX_SOURCES} CFLAGS ${UBSAN_CXXFLAGS})
 
   if (WIN32)
+    add_compiler_rt_object_libraries(UbsanWeakInterception
+      ${SANITIZER_COMMON_SUPPORTED_OS}
+      ARCHS ${UBSAN_SUPPORTED_ARCH}
+      SOURCES ubsan_win_weak_interception.cc
+      CFLAGS ${UBSAN_CFLAGS} -DSANITIZER_DYNAMIC
+      DEFS ${UBSAN_COMMON_DEFINITIONS})
+
     add_compiler_rt_object_libraries(UbsanDllThunk
       ${SANITIZER_COMMON_SUPPORTED_OS}
       ARCHS ${UBSAN_SUPPORTED_ARCH}
       SOURCES ubsan_win_dll_thunk.cc
       CFLAGS ${UBSAN_CFLAGS} -DSANITIZER_DLL_THUNK
       DEFS ${UBSAN_COMMON_DEFINITIONS})
 
     set(DYNAMIC_RUNTIME_THUNK_CFLAGS "-DSANITIZER_DYNAMIC_RUNTIME_THUNK")

compiler-rt/trunk/lib/ubsan/ubsan_win_weak_interception.cc

+//===-- ubsan_win_weak_interception.cc ------------------------------------===//
+//
+//                     The LLVM Compiler Infrastructure
+//
+// This file is distributed under the University of Illinois Open Source
+// License. See LICENSE.TXT for details.
+//
+//===----------------------------------------------------------------------===//
+// This module should be included in Ubsan when it is implemented as a shared
+// library on Windows (dll), in order to delegate the calls of weak functions to
+// the implementation in the main executable when a strong definition is
+// provided.
+//===----------------------------------------------------------------------===//
+#ifdef SANITIZER_DYNAMIC
+#include "sanitizer_common/sanitizer_win_weak_interception.h"
+#include "ubsan_flags.h"
+// Check if strong definitions for weak functions are present in the main
+// executable. If that is the case, override dll functions to point to strong
+// implementations.
+#define INTERFACE_FUNCTION(Name)
+#define INTERFACE_WEAK_FUNCTION(Name) INTERCEPT_SANITIZER_WEAK_FUNCTION(Name)
+#include "ubsan_interface.inc"
+#endif // SANITIZER_DYNAMIC