010908 build of HEAD for alpha-dec-osf3 (harvard-puffin)

[Ken Shan]

--pWyiEgJYm5f9v55/
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
Content-Transfer-Encoding: quoted-printable

On 2001-09-10T10:53:03+0100, Simon Marlow wrote:
> A likely cause of a nondeterministic crash is using data from below the
> C stack pointer at some stage (I think you had a bug like this before,
> right?).  To verify, you can increase or decrease the frequency of the
> timer signal, or turn it off altogether, by modifying the call to
> initialize_virtual_timer() in RtsStartup.c.

And now we have a delicate but deterministic crash building STABLE on
the Alpha (see yesterday's nightly build message).  Hrm.  (It's
delicate in that if you run the compiler slightly differently -- with
command-line arguments ordered differently or with .hi present, etc.
-- the crash moves or goes away.)

I would be very happy if either (preferably both) of these bugs are
due to using data from below the C stack pointer at some stage.  But!
I don't see where the C stack pointer is touched other than in
StgRun()...?

(off to turn off the virtual timer in the HEAD build to see)

--=20
Edit this signature at http://www.digitas.harvard.edu/cgi-bin/ken/sig
Flags flying at half staff in the hearts of men, like so many of
the most important things, are uncountable by digital means.

--pWyiEgJYm5f9v55/
Content-Type: application/pgp-signature
Content-Disposition: inline

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.0.6 (GNU/Linux)
Comment: For info see http://www.gnupg.org

iD8DBQE7oEd9zjAc4f+uuBURAj/iAKD7BfN7qyU5MkWOdzCnjiyAFq+lagCgqmtO
smc9SJhUHL4Lc8AvDCiNzsI=
=Yh0z
-----END PGP SIGNATURE-----

--pWyiEgJYm5f9v55/--