Got scudo error if exit std::thread.

[satics]

I got a crash error when enabling the ASan in the Android native code like below.

---

signal 6 (SIGABRT), code -1 (SI_QUEUE), fault addr --------
Abort message: 'Scudo ERROR: corrupted chunk header at address 0x7da31e30
'
r0 00000000 r1 00000f2a r2 00000006 r3 76eb2038
r4 76eb204c r5 76eb2030 r6 00000721 r7 0000016b
r8 76eb2038 r9 76eb2048 r10 76eb2068 r11 76eb2058
ip 00000f2a sp 76eb2008 lr aaf94981 pc aaf94994
backtrace:
#00 pc 00038994 /apex/com.android.runtime/lib/bionic/libc.so (abort+172) (BuildId: bb138a50a6b8f8361f25f87cc01c5671)
#1 pc 0002e70f /apex/com.android.runtime/lib/bionic/libc.so (scudo::die()+2) (BuildId: bb138a50a6b8f8361f25f87cc01c5671)
#2 pc 0002eb1d /apex/com.android.runtime/lib/bionic/libc.so (scudo::ScopedErrorReport::~ScopedErrorReport()+16) (BuildId: bb138a50a6b8f8361f25f87cc01c5671)
#3 pc 0002eb9f /apex/com.android.runtime/lib/bionic/libc.so (scudo::reportHeaderCorruption(void*)+34) (BuildId: bb138a50a6b8f8361f25f87cc01c5671)
#4 pc 0002f973 /apex/com.android.runtime/lib/bionic/libc.so (scudo::Allocator<scudo::AndroidConfig, &(scudo_malloc_postinit)>::deallocate(void*, scudo::Chunk::Origin, unsigned long, unsigned long)+234) (BuildId: bb138a50a6b8f8361f25f87cc01c5671)
#5 pc 0002f87b /apex/com.android.runtime/lib/bionic/libc.so (scudo_free+18) (BuildId: bb138a50a6b8f8361f25f87cc01c5671)
#6 pc 00081535 /apex/com.android.runtime/lib/bionic/libc.so (pthread_key_clean_all()+104) (BuildId: bb138a50a6b8f8361f25f87cc01c5671)
#7 pc 0008115d /apex/com.android.runtime/lib/bionic/libc.so (pthread_exit+38) (BuildId: bb138a50a6b8f8361f25f87cc01c5671)
#8 pc 000810a1 /apex/com.android.runtime/lib/bionic/libc.so (__pthread_start(void*)+42) (BuildId: bb138a50a6b8f8361f25f87cc01c5671)
#9 pc 00039e9d /apex/com.android.runtime/lib/bionic/libc.so (__start_thread+30) (BuildId: bb138a50a6b8f8361f25f87cc01c5671)

NOTE: use pthread is ok.

Is there any bug in using std::thread if enable asan?