Add support for custom loaders to the sanitizer symbolizer

Summary:
Adds a fallback mode to procmaps when the symbolizer
fails to locate a module for a given address by using
dl_iterate_phdr.

Reviewers: kubamracek, rnk, vitalybuka, eugenis

Reviewed By: eugenis

Subscribers: srhines, llvm-commits

Differential Revision: https://reviews.llvm.org/D37269

llvm-svn: 314431

compiler-rt/lib/sanitizer_common/sanitizer_common.h

@@ -727,9 +727,10 @@ class LoadedModule {
 // filling this information.
 class ListOfModules {
  public:
-  ListOfModules() : modules_(kInitialCapacity) {}
+  ListOfModules() : initialized(false) {}
   ~ListOfModules() { clear(); }
   void init();
+  void fallbackInit();  // Uses fallback init if available, otherwise clears
   const LoadedModule *begin() const { return modules_.begin(); }
   LoadedModule *begin() { return modules_.begin(); }
   const LoadedModule *end() const { return modules_.end(); }
@@ -745,10 +746,15 @@ class ListOfModules {
     for (auto &module : modules_) module.clear();
     modules_.clear();
   }
+  void clearOrInit() {
+    initialized ? clear() : modules_.Initialize(kInitialCapacity);
+    initialized = true;
+  }
 
-  InternalMmapVector<LoadedModule> modules_;
+  InternalMmapVectorNoCtor<LoadedModule> modules_;
   // We rarely have more than 16K loaded modules.
   static const uptr kInitialCapacity = 1 << 14;
+  bool initialized;
 };
 
 // Callback type for iterating over a set of memory ranges.

compiler-rt/lib/sanitizer_common/sanitizer_linux_libcdep.cc

@@ -26,6 +26,7 @@
 #include "sanitizer_placement_new.h"
 #include "sanitizer_procmaps.h"
 #include "sanitizer_stacktrace.h"
+#include "sanitizer_symbolizer.h"
 
 #include <dlfcn.h>  // for dlsym()
 #include <link.h>
@@ -424,7 +425,7 @@ typedef ElfW(Phdr) Elf_Phdr;
 # endif
 
 struct DlIteratePhdrData {
-  InternalMmapVector<LoadedModule> *modules;
+  InternalMmapVectorNoCtor<LoadedModule> *modules;
   bool first;
 };
 
@@ -462,22 +463,38 @@ extern "C" __attribute__((weak)) int dl_iterate_phdr(
     int (*)(struct dl_phdr_info *, size_t, void *), void *);
 #endif
 
-void ListOfModules::init() {
-  clear();
+static bool requiresProcmaps() {
 #if SANITIZER_ANDROID && __ANDROID_API__ <= 22
-  u32 api_level = AndroidGetApiLevel();
   // Fall back to /proc/maps if dl_iterate_phdr is unavailable or broken.
   // The runtime check allows the same library to work with
   // both K and L (and future) Android releases.
-  if (api_level <= ANDROID_LOLLIPOP_MR1) { // L or earlier
-    MemoryMappingLayout memory_mapping(false);
-    memory_mapping.DumpListOfModules(&modules_);
-    return;
-  }
+  return AndroidGetApiLevel() <= ANDROID_LOLLIPOP_MR1;
+#else
+  return false;
 #endif
+}
+
+static void procmapsInit(InternalMmapVectorNoCtor<LoadedModule> *modules) {
+  MemoryMappingLayout memory_mapping(false);
+  memory_mapping.DumpListOfModules(modules);
+}
+
+void ListOfModules::init() {
+  clearOrInit();
+  if (requiresProcmaps()) {
+    procmapsInit(&modules_);
+  } else {
     DlIteratePhdrData data = {&modules_, true};
     dl_iterate_phdr(dl_iterate_phdr_cb, &data);
   }
+}
+
+// When a custom loader is used, dl_iterate_phdr may not contain the full
+// list of modules. Allow callers to fall back to using procmaps.
+void ListOfModules::fallbackInit() {
+  clearOrInit();
+  if (!requiresProcmaps()) procmapsInit(&modules_);
+}
 
 // getrusage does not give us the current RSS, only the max RSS.
 // Still, this is better than nothing if /proc/self/statm is not available

compiler-rt/lib/sanitizer_common/sanitizer_mac.cc

@@ -411,11 +411,13 @@ void GetThreadStackAndTls(bool main, uptr *stk_addr, uptr *stk_size,
 }
 
 void ListOfModules::init() {
-  clear();
+  clearOrInit();
   MemoryMappingLayout memory_mapping(false);
   memory_mapping.DumpListOfModules(&modules_);
 }
 
+void ListOfModules::fallbackInit() { clear(); }
+
 static HandleSignalMode GetHandleSignalModeImpl(int signum) {
   switch (signum) {
     case SIGABRT:

compiler-rt/lib/sanitizer_common/sanitizer_procmaps.h

@@ -76,7 +76,7 @@ class MemoryMappingLayout {
   static void CacheMemoryMappings();
 
   // Adds all mapped objects into a vector.
-  void DumpListOfModules(InternalMmapVector<LoadedModule> *modules);
+  void DumpListOfModules(InternalMmapVectorNoCtor<LoadedModule> *modules);
 
  private:
   void LoadFromCache();

compiler-rt/lib/sanitizer_common/sanitizer_procmaps_common.cc

@@ -120,7 +120,7 @@ void MemoryMappingLayout::LoadFromCache() {
 }
 
 void MemoryMappingLayout::DumpListOfModules(
-    InternalMmapVector<LoadedModule> *modules) {
+    InternalMmapVectorNoCtor<LoadedModule> *modules) {
   Reset();
   InternalScopedString module_name(kMaxPathLength);
   MemoryMappedSegment segment(module_name.data(), module_name.size());

compiler-rt/lib/sanitizer_common/sanitizer_procmaps_mac.cc

@@ -353,7 +353,7 @@ bool MemoryMappingLayout::Next(MemoryMappedSegment *segment) {
 }
 
 void MemoryMappingLayout::DumpListOfModules(
-    InternalMmapVector<LoadedModule> *modules) {
+    InternalMmapVectorNoCtor<LoadedModule> *modules) {
   Reset();
   InternalScopedString module_name(kMaxPathLength);
   MemoryMappedSegment segment(module_name.data(), kMaxPathLength);

compiler-rt/lib/sanitizer_common/sanitizer_symbolizer.h

@@ -119,6 +119,7 @@ class Symbolizer final {
   void AddHooks(StartSymbolizationHook start_hook,
                 EndSymbolizationHook end_hook);
 
+  void RefreshModules();
   const LoadedModule *FindModuleForAddress(uptr address);
 
   void InvalidateModuleList();
@@ -151,6 +152,7 @@ class Symbolizer final {
                                          uptr *module_offset,
                                          ModuleArch *module_arch);
   ListOfModules modules_;
+  ListOfModules fallback_modules_;
   // If stale, need to reload the modules before looking up addresses.
   bool modules_fresh_;
 

compiler-rt/lib/sanitizer_common/sanitizer_symbolizer_libcdep.cc

@@ -163,29 +163,47 @@ bool Symbolizer::FindModuleNameAndOffsetForAddress(uptr address,
   return true;
 }
 
+void Symbolizer::RefreshModules() {
+  modules_.init();
+  fallback_modules_.fallbackInit();
+  RAW_CHECK(modules_.size() > 0);
+  modules_fresh_ = true;
+}
+
+static const LoadedModule *SearchForModule(const ListOfModules &modules,
+                                           uptr address) {
+  for (uptr i = 0; i < modules.size(); i++) {
+    if (modules[i].containsAddress(address)) {
+      return &modules[i];
+    }
+  }
+  return nullptr;
+}
+
 const LoadedModule *Symbolizer::FindModuleForAddress(uptr address) {
   bool modules_were_reloaded = false;
   if (!modules_fresh_) {
-    modules_.init();
-    RAW_CHECK(modules_.size() > 0);
-    modules_fresh_ = true;
+    RefreshModules();
     modules_were_reloaded = true;
   }
-  for (uptr i = 0; i < modules_.size(); i++) {
-    if (modules_[i].containsAddress(address)) {
-      return &modules_[i];
-    }
-  }
+  const LoadedModule *module = SearchForModule(modules_, address);
+  if (module) return module;
+
     // dlopen/dlclose interceptors invalidate the module list, but when
     // interception is disabled, we need to retry if the lookup fails in
     // case the module list changed.
 #if !SANITIZER_INTERCEPT_DLOPEN_DLCLOSE
   if (!modules_were_reloaded) {
-    modules_fresh_ = false;
-    return FindModuleForAddress(address);
+    RefreshModules();
+    module = SearchForModule(modules_, address);
+    if (module) return module;
   }
 #endif
-  return 0;
+
+  if (fallback_modules_.size()) {
+    module = SearchForModule(fallback_modules_, address);
+  }
+  return module;
 }
 
 // For now we assume the following protocol:

compiler-rt/lib/sanitizer_common/sanitizer_win.cc

@@ -524,7 +524,7 @@ static uptr GetPreferredBase(const char *modname) {
 }
 
 void ListOfModules::init() {
-  clear();
+  clearOrInit();
   HANDLE cur_process = GetCurrentProcess();
 
   // Query the list of modules.  Start by assuming there are no more than 256
@@ -583,7 +583,9 @@ void ListOfModules::init() {
     modules_.push_back(cur_module);
   }
   UnmapOrDie(hmodules, modules_buffer_size);
-};
+}
+
+void ListOfModules::fallbackInit() { clear(); }
 
 // We can't use atexit() directly at __asan_init time as the CRT is not fully
 // initialized at this point.  Place the functions into a vector and use