Tried running DPDK under Address Sanitizer with the thread data race
option, and ran into something that needs some discussion.
Is this just an ASAN bug?
The fatal log is:
==================
WARNING: ThreadSanitizer: data race (pid=28275)
Write of size 1 at 0xfffff5000580 by main thread:
#0 pthread_barrier_destroy <null> (libtsan.so.0+0x314b3)
#1 rte_ctrl_thread_create <null> (librte_eal.so.20.0+0x2f9db)
#2 rte_eal_intr_init <null> (librte_eal.so.20.0+0x1e573)
#3 rte_eal_init <null> (librte_eal.so.20.0+0xcb9b)
#4 <null> <null> (dpdk-backplane+0xb2ab)
#5 main <null> (dpdk-backplane+0xa1ab)
Previous read of size 1 at 0xfffff5000580 by thread T3:
#0 pthread_barrier_wait <null> (libtsan.so.0+0x315cf)
#1 rte_thread_init <null> (librte_eal.so.20.0+0x2f44f)
#2 <null> <null> (libtsan.so.0+0x2cc63)
Location is heap block of size 48 at 0xfffff5000570 allocated by main thread:
#0 malloc <null> (libtsan.so.0+0x2e4ef)
#1 rte_ctrl_thread_create <null> (librte_eal.so.20.0+0x2f873)
#2 rte_eal_intr_init <null> (librte_eal.so.20.0+0x1e573)
#3 rte_eal_init <null> (librte_eal.so.20.0+0xcb9b)
#4 <null> <null> (dpdk-backplane+0xb2ab)
#5 main <null> (dpdk-backplane+0xa1ab)
Thread T3 'eal-intr-thread' (tid=28284, running) created by main thread at:
#0 pthread_create <null> (libtsan.so.0+0x2f5a7)
#1 rte_ctrl_thread_create <null> (librte_eal.so.20.0+0x2f8bb)
#2 rte_eal_intr_init <null> (librte_eal.so.20.0+0x1e573)
#3 rte_eal_init <null> (librte_eal.so.20.0+0xcb9b)
#4 <null> <null> (dpdk-backplane+0xb2ab)
#5 main <null> (dpdk-backplane+0xa1ab)
SUMMARY: ThreadSanitizer: data race (/lib/libtsan.so.0+0x314b3) in
pthread_barrier_destroy
==================
The code is:
pthread_barrier_init(¶ms->configured, NULL, 2);
ret = pthread_create(thread, attr, ctrl_thread_init, (void *)params);
...
ret = pthread_barrier_wait(¶ms->configured);
if (ret == PTHREAD_BARRIER_SERIAL_THREAD) {
pthread_barrier_destroy(¶ms->configured);