I think I've found a bug in SBCL 0.8.13's dead thread reaping. If a
thread which is not the initial thread makes a thread and the parent
dies before the child, sbcl never notices the child's death when it
comes (I guess linux throws away the SIGCHLD when it sees its parent
is gone, like for normal processes), and complains about being unable
to kill it at quit time. I imagine it never frees the memory mmap'd
at creation-time either.
In case my description isn't clear:
* (sb-thread:make-thread (lambda ()
(sb-thread:make-thread (lambda ()
[wait a couple of seconds to give both time to die]
debugger invoked on a SIMPLE-ERROR in thread 30764:
Syscall interrupt_thread failed: No such process
You can type HELP for debugger help, or (SB-EXT:QUIT) to exit from SBCL.
(no restarts: If you didn't do this on purpose, please report it as a bug.)
(SB-THREAD:INTERRUPT-THREAD 2 30766 QUIT)[:EXTERNAL]