I came across this when reviewing a problem where a submission inside run_time_error just never stopped running. That is, the rlimit applied never causes the process to die.
I developed a minimal example, which is attached.
When I attach from GDB the process that is running, I see the following stack trace, which appears to point to a deadlock in the memory manager:
(gdb) bt
#0 0x0000000000452d2c in __lll_lock_wait_private ()
#1 0x0000000000410d30 in __vfscanf_internal ()
#2 0x0000000000040000 in ?? ()
#3 0x00000000004d50a0 in ?? ()
#4 0x0000000000401783 in init_cacheinfo ()
#5 0x0000000000402ba4 in __libc_csu_init ()
#6 0x00000000004023f0 in __libc_start_main ()
#7 0x0000000000401b2a in _start ()
I came across this when reviewing a problem where a submission inside
run_time_errorjust never stopped running. That is, the rlimit applied never causes the process to die.I developed a minimal example, which is attached.
When I attach from GDB the process that is running, I see the following stack trace, which appears to point to a deadlock in the memory manager:
rteproblem.tar.gz