[HWASAN] Prevent crashes on thread exit

[vitalybuka]

Summary

I can't figure out how to reproduce this for test, but I see the case on
random binaries.

The known issue is with GLIBC, others may have a workaround, e.g. Bionic,
https://cs.android.com/android/platform/superproject/+/master:bionic/libc/bionic/pthread_exit.cpp;l=149
see signals blocked above.

[eugenis]

I do not understand the bionic issue. __hwasan_thread_exit is called very late (as the last thing in that function), while the signals are blocked. Why do thread cleanup functions matter?

[vitalybuka]

GLIBC only

[vitalybuka]

In D150401#4336141, @eugenis wrote:

I do not understand the bionic issue. __hwasan_thread_exit is called very late (as the last thing in that function), while the signals are blocked. Why do thread cleanup functions matter?

fixed

[vitalybuka]

typo

[eugenis]

LGTM

compiler-rt/lib/hwasan/hwasan_linux.cpp

 extern "C" void __hwasan_thread_enter() {
   hwasanThreadList().CreateCurrentThread()->EnsureRandomStateInited();
 }
 
 extern "C" void __hwasan_thread_exit() {
   Thread *t = GetCurrentThread();
   // Make sure that signal handler can not see a stale current thread pointer.
   atomic_signal_fence(memory_order_seq_cst);
-  if (t)
+  if (t) {
+    // Block async signals on the thread as the handler can be instrumented.
+    // After this point instrumented code can't access essential data from TLS
+    // and will crash.
+    BlockSignals();
     hwasanThreadList().ReleaseThread(t);
   }
+}
 
 #  if HWASAN_WITH_INTERCEPTORS
 static pthread_key_t tsd_key;
 static bool tsd_key_inited = false;
 
 void HwasanTSDThreadInit() {
   if (tsd_key_inited)
     CHECK_EQ(0, pthread_setspecific(tsd_key,