Closed ghost closed 5 years ago
Maybe also of note, we have configured this job for a fixed number of iterations and see these failures when that number is presumably reached and the fuzzer should shut down gracefully.
Maybe updating upstream honggfuzz could be promising as it contains a few commits related to signal handling?
Hi @adam-rhebo, I have no idea where it came from and how to reproduce. But it seems that upstream went through significant changes. I just released a new version with updated upstream, 0.5.42, could you check if it is now resolved?
I just released a new version with updated upstream, 0.5.42, could you check if it is now resolved?
Will do.
Seems to have done the trick:
Size:3942 (i,b,hw,edge,ip,cmp): 0/0/0/0/0/1, Tot:0/0/0/6703/14/141865
Size:979 (i,b,hw,edge,ip,cmp): 0/0/0/0/0/1, Tot:0/0/0/6703/14/141866
Size:463 (i,b,hw,edge,ip,cmp): 0/0/0/0/0/1, Tot:0/0/0/6703/14/141867
Terminating thread no. #1, left: 4
Terminating thread no. #2, left: 3
Terminating thread no. #0, left: 3
[2019-02-19T19:30:44+0000][W][91] pingThreads():201 pthread_kill(thread=0, SIGUSR1): Success
[2019-02-19T19:30:44+0000][W][91] pingThreads():201 pthread_kill(thread=1, SIGUSR1): Success
[2019-02-19T19:30:44+0000][W][91] pingThreads():201 pthread_kill(thread=2, SIGUSR1): Success
[2019-02-19T19:30:44+0000][W][91] pingThreads():201 pthread_kill(thread=0, SIGUSR1): Success
[2019-02-19T19:30:44+0000][W][91] pingThreads():201 pthread_kill(thread=1, SIGUSR1): Success
[2019-02-19T19:30:44+0000][W][91] pingThreads():201 pthread_kill(thread=2, SIGUSR1): Success
Terminating thread no. #3, left: 1
[2019-02-19T19:30:44+0000][W][91] pingThreads():201 pthread_kill(thread=0, SIGUSR1): Success
[2019-02-19T19:30:44+0000][W][91] pingThreads():201 pthread_kill(thread=1, SIGUSR1): Success
[2019-02-19T19:30:44+0000][W][91] pingThreads():201 pthread_kill(thread=2, SIGUSR1): Success
[2019-02-19T19:30:44+0000][W][91] pingThreads():201 pthread_kill(thread=3, SIGUSR1): Success
Summary iterations:10000004 time:1395 speed:7168
Since updating to version 0.5.41, we see termination failures like
repeatedly in our CI. We don't think that anything changed on the host that runs the fuzzing job but more importantly we are not sure where to start investigating. Any ideas what could have caused this?