rlimit fails on some submissions

[ghamerly]

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 ()

rteproblem.tar.gz