CC: | "Simon Peyton-Jones" <simonpj [...] microsoft.com> |
Subject: | GHC bugs |
Date: | Thu, 14 Jul 2005 10:28:28 +0100 |
To: | <pugsbugs [...] perl.org>, <malaire [...] users.sourceforge.net>, <autrijus [...] autrijus.org> |
From: | "Simon Marlow" <simonmar [...] microsoft.com> |
Hi,
I've been trying in vain over the last day or so to reproduce the
following GHC bugs, both allegedly demonstrated by Pugs:
[ 1226252 ] internal error: EVACUATED object entered!
http://sourceforge.net/tracker/index.php?func=detail&aid=1222522&group_i
d=8032&atid=108032
[ 1222522 ] scavenge_stack: weird activation record found on stack:37548
http://sourceforge.net/tracker/index.php?func=detail&aid=1222522&group_i
d=8032&atid=108032
I checked out the exact revisions using SVN as specified in the bug
reports, and I've tried various combinations of {1386,x86_64},
{Linux,FreeBSD}, {GHC 6.4, 6.4.1-STABLE}, {optimised,unoptimised}. I do
indeed get test failures for the t/oo/desctruction.t test, but no
crashes. For example, on the FreeBSD/x86/optimised build I just did:
~/local/pugs > GHCRTS=-H1m PERL_DL_NONLAZY=1 /usr/local/bin/perl
"-I../../inc" "-I../inc" "-Iinc" "-MExtUtils::Command::MM" "-e"
"ENV->{HARNESS_PERL} = q{/home/simonmar/local/pugs/blib/script/pugs};
@ARGV = map glob, @ARGV; ENV->{PERL6LIB} =
q{/home/simonmar/local/pugs/blib6/lib}; test_harness(0, 'inc',
'/home/simonmar/local/pugs/blib/lib',
'/home/simonmar/local/pugs/blib/arch')" t/oo/destruction.t
t/oo/destruction....NOK 5# Looks like you failed 2 tests of 6
t/oo/destruction....FAILED tests 4-5
Failed 2/6 tests, 66.67% okay
Failed Test Stat Wstat Total Fail Failed List of Failed
------------------------------------------------------------------------
-------
t/oo/destruction.t 6 2 33.33% 4-5
Failed 1/1 test scripts, 0.00% okay. 2/6 subtests failed, 66.67% okay.
I can vary the success rate by altering GHC's garbage collector options
(e.g. the GHCRTS setting in the command above).
At this point I'm stuck. I don't doubt there's a bug somewhere, but
it's not showing itself for me.
If someone can provide me with a temporary account on a machine that
demonstrates the bug, that would be great. SSH key available on
request. Even better if you could set me up with an account and a Pugs
build tree all ready to go.
I've spent too much time on this to give up now!
Cheers,
Simon