Re: S48 - Crash at GC write barrier

Robert Ransom <[email protected]> writes:

> I ran the Scheme 48 test suite earlier today, and it crashed with the
> following output:
>
>
> [posix-tests
> [posix-core-tests
>  (file-mode-predicates)
>  (file-modes)
>  (make-directory)
>  (time)
>  (set-working-directory!)
>  (i/o-flags)
>  (append-mode)
>  (file-timesscheme48vm: ./c/bibop/area_roots.h:72: s48_write_barrier_inline: Assertion `address < area->frontier' failed.
>
>
> I ran “make check” again, and could not reproduce the crash; I also
> couldn't find a core file to debug this with anywhere on my system.

The BIBOP GC is sufficiently complex to make it likely there are still
problems left, even though it's been quite extensively debugged.  (We
did GC hackathons in 2008 and 2009.)  But without something
semi-reproducible, I'm afraid there's nil chance of finding the problem.
With the POSIX support, I'd think that the foreign code is actually a
more likely suspect.  (You might loop "make check" overnight or
something to see if you can make it pop up again.)

-- 
Cheers =8-} Mike
Friede, Völkerverständigung und überhaupt blabla